Data.Colour.Matrix:determinant from colour-2.3.3, A

Percentage Accurate: 73.2% → 81.7%

Time: 7.8s

Alternatives: 21

Speedup: 0.5×

• 58.3% of points produce a very large (infinite) output. You may want to add a precondition.

Specification

Math

\[\begin{array}{l} \\ \left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (+
  (- (* x (- (* y z) (* t a))) (* b (- (* c z) (* t i))))
  (* j (- (* c a) (* y i)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	return ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    code = ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)))
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	return ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
}

Python

def code(x, y, z, t, a, b, c, i, j):
	return ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)))

Julia

function code(x, y, z, t, a, b, c, i, j)
	return Float64(Float64(Float64(x * Float64(Float64(y * z) - Float64(t * a))) - Float64(b * Float64(Float64(c * z) - Float64(t * i)))) + Float64(j * Float64(Float64(c * a) - Float64(y * i))))
end

MATLAB

function tmp = code(x, y, z, t, a, b, c, i, j)
	tmp = ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := N[(N[(N[(x * N[(N[(y * z), $MachinePrecision] - N[(t * a), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - N[(b * N[(N[(c * z), $MachinePrecision] - N[(t * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] + N[(j * N[(N[(c * a), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Initial Program: 73.2% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (+
  (- (* x (- (* y z) (* t a))) (* b (- (* c z) (* t i))))
  (* j (- (* c a) (* y i)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	return ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    code = ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)))
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	return ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
}

Python

def code(x, y, z, t, a, b, c, i, j):
	return ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)))

Julia

function code(x, y, z, t, a, b, c, i, j)
	return Float64(Float64(Float64(x * Float64(Float64(y * z) - Float64(t * a))) - Float64(b * Float64(Float64(c * z) - Float64(t * i)))) + Float64(j * Float64(Float64(c * a) - Float64(y * i))))
end

MATLAB

function tmp = code(x, y, z, t, a, b, c, i, j)
	tmp = ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := N[(N[(N[(x * N[(N[(y * z), $MachinePrecision] - N[(t * a), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - N[(b * N[(N[(c * z), $MachinePrecision] - N[(t * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] + N[(j * N[(N[(c * a), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Alternative 1: 81.7% accurate, 0.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right)\\ \mathbf{if}\;t\_1 \leq \infty:\\ \;\;\;\;t\_1\\ \mathbf{else}:\\ \;\;\;\;\left(-t\right) \cdot \left(a \cdot x - i \cdot b\right)\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1
         (+
          (- (* x (- (* y z) (* t a))) (* b (- (* c z) (* t i))))
          (* j (- (* c a) (* y i))))))
   (if (<= t_1 INFINITY) t_1 (* (- t) (- (* a x) (* i b))))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
	double tmp;
	if (t_1 <= ((double) INFINITY)) {
		tmp = t_1;
	} else {
		tmp = -t * ((a * x) - (i * b));
	}
	return tmp;
}

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
	double tmp;
	if (t_1 <= Double.POSITIVE_INFINITY) {
		tmp = t_1;
	} else {
		tmp = -t * ((a * x) - (i * b));
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)))
	tmp = 0
	if t_1 <= math.inf:
		tmp = t_1
	else:
		tmp = -t * ((a * x) - (i * b))
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(Float64(x * Float64(Float64(y * z) - Float64(t * a))) - Float64(b * Float64(Float64(c * z) - Float64(t * i)))) + Float64(j * Float64(Float64(c * a) - Float64(y * i))))
	tmp = 0.0
	if (t_1 <= Inf)
		tmp = t_1;
	else
		tmp = Float64(Float64(-t) * Float64(Float64(a * x) - Float64(i * b)));
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = ((x * ((y * z) - (t * a))) - (b * ((c * z) - (t * i)))) + (j * ((c * a) - (y * i)));
	tmp = 0.0;
	if (t_1 <= Inf)
		tmp = t_1;
	else
		tmp = -t * ((a * x) - (i * b));
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(N[(x * N[(N[(y * z), $MachinePrecision] - N[(t * a), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - N[(b * N[(N[(c * z), $MachinePrecision] - N[(t * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] + N[(j * N[(N[(c * a), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[t$95$1, Infinity], t$95$1, N[((-t) * N[(N[(a * x), $MachinePrecision] - N[(i * b), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if (+.f64 (-.f64 (*.f64 x (-.f64 (*.f64 y z) (*.f64 t a))) (*.f64 b (-.f64 (*.f64 c z) (*.f64 t i)))) (*.f64 j (-.f64 (*.f64 c a) (*.f64 y i)))) < +inf.0

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   if +inf.0 < (+.f64 (-.f64 (*.f64 x (-.f64 (*.f64 y z) (*.f64 t a))) (*.f64 b (-.f64 (*.f64 c z) (*.f64 t i)))) (*.f64 j (-.f64 (*.f64 c a) (*.f64 y i))))

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in t around -inf

      \[\leadsto \color{blue}{-1 \cdot \left(t \cdot \left(a \cdot x - b \cdot i\right)\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot t\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      2. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      3. lower-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      4. lower-neg.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      5. lower--.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b \cdot i}\right) \]

      6. lower-*.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b} \cdot i\right) \]

      7. *-commutative N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

      8. lower-*.f64 38.7

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

   4. Applied rewrites 38.7%

      \[\leadsto \color{blue}{\left(-t\right) \cdot \left(a \cdot x - i \cdot b\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 2: 67.6% accurate, 0.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right)\\ \mathbf{if}\;y \leq -4.5 \cdot 10^{+231}:\\ \;\;\;\;\mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y\\ \mathbf{elif}\;y \leq -2.7 \cdot 10^{-24}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;y \leq 6.4 \cdot 10^{-295}:\\ \;\;\;\;\mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right)\\ \mathbf{elif}\;y \leq 4.4 \cdot 10^{-17}:\\ \;\;\;\;\left(\mathsf{fma}\left(z, y, -b \cdot \frac{c \cdot z - i \cdot t}{x}\right) - a \cdot t\right) \cdot x + j \cdot \left(a \cdot c\right)\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (fma (- (* c a) (* i y)) j (* (- (* z y) (* a t)) x))))
   (if (<= y -4.5e+231)
     (* (fma (- i) j (* z x)) y)
     (if (<= y -2.7e-24)
       t_1
       (if (<= y 6.4e-295)
         (fma (* j c) a (- (* i (- (* b t) (* j y))) (* (* c b) z)))
         (if (<= y 4.4e-17)
           (+
            (* (- (fma z y (- (* b (/ (- (* c z) (* i t)) x)))) (* a t)) x)
            (* j (* a c)))
           t_1))))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = fma(((c * a) - (i * y)), j, (((z * y) - (a * t)) * x));
	double tmp;
	if (y <= -4.5e+231) {
		tmp = fma(-i, j, (z * x)) * y;
	} else if (y <= -2.7e-24) {
		tmp = t_1;
	} else if (y <= 6.4e-295) {
		tmp = fma((j * c), a, ((i * ((b * t) - (j * y))) - ((c * b) * z)));
	} else if (y <= 4.4e-17) {
		tmp = ((fma(z, y, -(b * (((c * z) - (i * t)) / x))) - (a * t)) * x) + (j * (a * c));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = fma(Float64(Float64(c * a) - Float64(i * y)), j, Float64(Float64(Float64(z * y) - Float64(a * t)) * x))
	tmp = 0.0
	if (y <= -4.5e+231)
		tmp = Float64(fma(Float64(-i), j, Float64(z * x)) * y);
	elseif (y <= -2.7e-24)
		tmp = t_1;
	elseif (y <= 6.4e-295)
		tmp = fma(Float64(j * c), a, Float64(Float64(i * Float64(Float64(b * t) - Float64(j * y))) - Float64(Float64(c * b) * z)));
	elseif (y <= 4.4e-17)
		tmp = Float64(Float64(Float64(fma(z, y, Float64(-Float64(b * Float64(Float64(Float64(c * z) - Float64(i * t)) / x)))) - Float64(a * t)) * x) + Float64(j * Float64(a * c)));
	else
		tmp = t_1;
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j + N[(N[(N[(z * y), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision] * x), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[y, -4.5e+231], N[(N[((-i) * j + N[(z * x), $MachinePrecision]), $MachinePrecision] * y), $MachinePrecision], If[LessEqual[y, -2.7e-24], t$95$1, If[LessEqual[y, 6.4e-295], N[(N[(j * c), $MachinePrecision] * a + N[(N[(i * N[(N[(b * t), $MachinePrecision] - N[(j * y), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - N[(N[(c * b), $MachinePrecision] * z), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], If[LessEqual[y, 4.4e-17], N[(N[(N[(N[(z * y + (-N[(b * N[(N[(N[(c * z), $MachinePrecision] - N[(i * t), $MachinePrecision]), $MachinePrecision] / x), $MachinePrecision]), $MachinePrecision])), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision] * x), $MachinePrecision] + N[(j * N[(a * c), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], t$95$1]]]]]

Derivation

1. Split input into 4 regimes

2. if y < -4.49999999999999991e231

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{y \cdot \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot i\right) \cdot j + x \cdot z\right) \cdot y \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(i\right)\right) \cdot j + x \cdot z\right) \cdot y \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(i\right), j, x \cdot z\right) \cdot y \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-i, j, x \cdot z\right) \cdot y \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

      8. lower-*.f64 39.1

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

   4. Applied rewrites 39.1%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y} \]

   if -4.49999999999999991e231 < y < -2.70000000000000007e-24 or 4.4e-17 < y

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around 0

      \[\leadsto \color{blue}{j \cdot \left(a \cdot c - i \cdot y\right) + x \cdot \left(y \cdot z - a \cdot t\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j + \color{blue}{x} \cdot \left(y \cdot z - a \cdot t\right) \]

      2. *-commutative N/A

         \[\leadsto \left(c \cdot a - i \cdot y\right) \cdot j + x \cdot \left(y \cdot z - a \cdot t\right) \]

      3. *-commutative N/A

         \[\leadsto \left(c \cdot a - y \cdot i\right) \cdot j + x \cdot \left(y \cdot z - a \cdot t\right) \]

      4. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, \color{blue}{j}, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      5. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      6. lift--.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      8. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      9. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      10. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - t \cdot a\right) \cdot x\right) \]

      11. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - t \cdot a\right) \cdot x\right) \]

      12. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      13. lower--.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      14. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

      15. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

      16. lower-*.f64 59.9

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

   4. Applied rewrites 59.9%

      \[\leadsto \color{blue}{\mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right)} \]

   if -2.70000000000000007e-24 < y < 6.4e-295

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in i around 0

      \[\leadsto \color{blue}{\left(a \cdot \left(c \cdot j\right) + \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right)\right) - b \cdot \left(c \cdot z\right)} \]
   3. Step-by-step derivation

      1. associate--l+ N/A

         \[\leadsto a \cdot \left(c \cdot j\right) + \color{blue}{\left(\left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right)} \]

      2. *-commutative N/A

         \[\leadsto \left(c \cdot j\right) \cdot a + \left(\color{blue}{\left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right)} - b \cdot \left(c \cdot z\right)\right) \]

      3. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot j, \color{blue}{a}, \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right) \]

      4. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right) \]

      5. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right) \]

      6. lower--.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right) \]

   4. Applied rewrites 74.4%

      \[\leadsto \color{blue}{\mathsf{fma}\left(j \cdot c, a, \mathsf{fma}\left(\left(-j \cdot y\right) - \left(-b \cdot t\right), i, \left(z \cdot y - a \cdot t\right) \cdot x\right) - \left(c \cdot b\right) \cdot z\right)} \]

   5. Taylor expanded in x around 0

      \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

      2. lower--.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

      3. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

      4. lift-*.f64 58.6

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

   7. Applied rewrites 58.6%

      \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

   if 6.4e-295 < y < 4.4e-17

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(\left(-1 \cdot \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{x} + y \cdot z\right) - a \cdot t\right)} + j \cdot \left(c \cdot a - y \cdot i\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(-1 \cdot \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{x} + y \cdot z\right) - a \cdot t\right) \cdot \color{blue}{x} + j \cdot \left(c \cdot a - y \cdot i\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(-1 \cdot \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{x} + y \cdot z\right) - a \cdot t\right) \cdot \color{blue}{x} + j \cdot \left(c \cdot a - y \cdot i\right) \]

   4. Applied rewrites 68.0%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(z, y, -b \cdot \frac{c \cdot z - i \cdot t}{x}\right) - a \cdot t\right) \cdot x} + j \cdot \left(c \cdot a - y \cdot i\right) \]

   5. Taylor expanded in y around 0

      \[\leadsto \left(\mathsf{fma}\left(z, y, -b \cdot \frac{c \cdot z - i \cdot t}{x}\right) - a \cdot t\right) \cdot x + j \cdot \color{blue}{\left(a \cdot c\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 63.4

         \[\leadsto \left(\mathsf{fma}\left(z, y, -b \cdot \frac{c \cdot z - i \cdot t}{x}\right) - a \cdot t\right) \cdot x + j \cdot \left(a \cdot \color{blue}{c}\right) \]

   7. Applied rewrites 63.4%

      \[\leadsto \left(\mathsf{fma}\left(z, y, -b \cdot \frac{c \cdot z - i \cdot t}{x}\right) - a \cdot t\right) \cdot x + j \cdot \color{blue}{\left(a \cdot c\right)} \]
3. Recombined 4 regimes into one program.
4. Add Preprocessing

Alternative 3: 67.1% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right)\\ \mathbf{if}\;y \leq -4.5 \cdot 10^{+231}:\\ \;\;\;\;\mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y\\ \mathbf{elif}\;y \leq -9.6 \cdot 10^{-129}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;y \leq 6.8 \cdot 10^{-105}:\\ \;\;\;\;\mathsf{fma}\left(-a, t \cdot x, \left(j \cdot c\right) \cdot a\right) - \left(c \cdot z - i \cdot t\right) \cdot b\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (fma (- (* c a) (* i y)) j (* (- (* z y) (* a t)) x))))
   (if (<= y -4.5e+231)
     (* (fma (- i) j (* z x)) y)
     (if (<= y -9.6e-129)
       t_1
       (if (<= y 6.8e-105)
         (- (fma (- a) (* t x) (* (* j c) a)) (* (- (* c z) (* i t)) b))
         t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = fma(((c * a) - (i * y)), j, (((z * y) - (a * t)) * x));
	double tmp;
	if (y <= -4.5e+231) {
		tmp = fma(-i, j, (z * x)) * y;
	} else if (y <= -9.6e-129) {
		tmp = t_1;
	} else if (y <= 6.8e-105) {
		tmp = fma(-a, (t * x), ((j * c) * a)) - (((c * z) - (i * t)) * b);
	} else {
		tmp = t_1;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = fma(Float64(Float64(c * a) - Float64(i * y)), j, Float64(Float64(Float64(z * y) - Float64(a * t)) * x))
	tmp = 0.0
	if (y <= -4.5e+231)
		tmp = Float64(fma(Float64(-i), j, Float64(z * x)) * y);
	elseif (y <= -9.6e-129)
		tmp = t_1;
	elseif (y <= 6.8e-105)
		tmp = Float64(fma(Float64(-a), Float64(t * x), Float64(Float64(j * c) * a)) - Float64(Float64(Float64(c * z) - Float64(i * t)) * b));
	else
		tmp = t_1;
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j + N[(N[(N[(z * y), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision] * x), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[y, -4.5e+231], N[(N[((-i) * j + N[(z * x), $MachinePrecision]), $MachinePrecision] * y), $MachinePrecision], If[LessEqual[y, -9.6e-129], t$95$1, If[LessEqual[y, 6.8e-105], N[(N[((-a) * N[(t * x), $MachinePrecision] + N[(N[(j * c), $MachinePrecision] * a), $MachinePrecision]), $MachinePrecision] - N[(N[(N[(c * z), $MachinePrecision] - N[(i * t), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision]), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 3 regimes

2. if y < -4.49999999999999991e231

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{y \cdot \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot i\right) \cdot j + x \cdot z\right) \cdot y \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(i\right)\right) \cdot j + x \cdot z\right) \cdot y \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(i\right), j, x \cdot z\right) \cdot y \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-i, j, x \cdot z\right) \cdot y \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

      8. lower-*.f64 39.1

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

   4. Applied rewrites 39.1%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y} \]

   if -4.49999999999999991e231 < y < -9.59999999999999954e-129 or 6.79999999999999984e-105 < y

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around 0

      \[\leadsto \color{blue}{j \cdot \left(a \cdot c - i \cdot y\right) + x \cdot \left(y \cdot z - a \cdot t\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j + \color{blue}{x} \cdot \left(y \cdot z - a \cdot t\right) \]

      2. *-commutative N/A

         \[\leadsto \left(c \cdot a - i \cdot y\right) \cdot j + x \cdot \left(y \cdot z - a \cdot t\right) \]

      3. *-commutative N/A

         \[\leadsto \left(c \cdot a - y \cdot i\right) \cdot j + x \cdot \left(y \cdot z - a \cdot t\right) \]

      4. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, \color{blue}{j}, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      5. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      6. lift--.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      8. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      9. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      10. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - t \cdot a\right) \cdot x\right) \]

      11. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - t \cdot a\right) \cdot x\right) \]

      12. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      13. lower--.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      14. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

      15. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

      16. lower-*.f64 59.9

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

   4. Applied rewrites 59.9%

      \[\leadsto \color{blue}{\mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right)} \]

   if -9.59999999999999954e-129 < y < 6.79999999999999984e-105

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in y around 0

      \[\leadsto \color{blue}{\left(-1 \cdot \left(a \cdot \left(t \cdot x\right)\right) + a \cdot \left(c \cdot j\right)\right) - b \cdot \left(c \cdot z - i \cdot t\right)} \]
   3. Step-by-step derivation

      1. lower--.f64 N/A

         \[\leadsto \left(-1 \cdot \left(a \cdot \left(t \cdot x\right)\right) + a \cdot \left(c \cdot j\right)\right) - \color{blue}{b \cdot \left(c \cdot z - i \cdot t\right)} \]

      2. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot a\right) \cdot \left(t \cdot x\right) + a \cdot \left(c \cdot j\right)\right) - b \cdot \left(c \cdot z - i \cdot t\right) \]

      3. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(a\right)\right) \cdot \left(t \cdot x\right) + a \cdot \left(c \cdot j\right)\right) - b \cdot \left(c \cdot z - i \cdot t\right) \]

      4. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(a\right), t \cdot x, a \cdot \left(c \cdot j\right)\right) - \color{blue}{b} \cdot \left(c \cdot z - i \cdot t\right) \]

      5. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-a, t \cdot x, a \cdot \left(c \cdot j\right)\right) - b \cdot \left(c \cdot z - i \cdot t\right) \]

      6. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(-a, t \cdot x, a \cdot \left(c \cdot j\right)\right) - b \cdot \left(c \cdot z - i \cdot t\right) \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-a, t \cdot x, \left(c \cdot j\right) \cdot a\right) - b \cdot \left(c \cdot z - i \cdot t\right) \]

      8. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(-a, t \cdot x, \left(c \cdot j\right) \cdot a\right) - b \cdot \left(c \cdot z - i \cdot t\right) \]

      9. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-a, t \cdot x, \left(j \cdot c\right) \cdot a\right) - b \cdot \left(c \cdot z - i \cdot t\right) \]

      10. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(-a, t \cdot x, \left(j \cdot c\right) \cdot a\right) - b \cdot \left(c \cdot z - i \cdot t\right) \]

      11. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(-a, t \cdot x, \left(j \cdot c\right) \cdot a\right) - \left(c \cdot z - i \cdot t\right) \cdot \color{blue}{b} \]

      12. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(-a, t \cdot x, \left(j \cdot c\right) \cdot a\right) - \left(c \cdot z - t \cdot i\right) \cdot b \]

      13. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(-a, t \cdot x, \left(j \cdot c\right) \cdot a\right) - \left(c \cdot z - t \cdot i\right) \cdot \color{blue}{b} \]

   4. Applied rewrites 59.3%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-a, t \cdot x, \left(j \cdot c\right) \cdot a\right) - \left(c \cdot z - i \cdot t\right) \cdot b} \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 4: 65.5% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right)\\ \mathbf{if}\;y \leq -4.5 \cdot 10^{+231}:\\ \;\;\;\;\mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y\\ \mathbf{elif}\;y \leq -2.7 \cdot 10^{-24}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;y \leq 2.5 \cdot 10^{-106}:\\ \;\;\;\;\mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right)\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (fma (- (* c a) (* i y)) j (* (- (* z y) (* a t)) x))))
   (if (<= y -4.5e+231)
     (* (fma (- i) j (* z x)) y)
     (if (<= y -2.7e-24)
       t_1
       (if (<= y 2.5e-106)
         (fma (* j c) a (- (* i (- (* b t) (* j y))) (* (* c b) z)))
         t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = fma(((c * a) - (i * y)), j, (((z * y) - (a * t)) * x));
	double tmp;
	if (y <= -4.5e+231) {
		tmp = fma(-i, j, (z * x)) * y;
	} else if (y <= -2.7e-24) {
		tmp = t_1;
	} else if (y <= 2.5e-106) {
		tmp = fma((j * c), a, ((i * ((b * t) - (j * y))) - ((c * b) * z)));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = fma(Float64(Float64(c * a) - Float64(i * y)), j, Float64(Float64(Float64(z * y) - Float64(a * t)) * x))
	tmp = 0.0
	if (y <= -4.5e+231)
		tmp = Float64(fma(Float64(-i), j, Float64(z * x)) * y);
	elseif (y <= -2.7e-24)
		tmp = t_1;
	elseif (y <= 2.5e-106)
		tmp = fma(Float64(j * c), a, Float64(Float64(i * Float64(Float64(b * t) - Float64(j * y))) - Float64(Float64(c * b) * z)));
	else
		tmp = t_1;
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j + N[(N[(N[(z * y), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision] * x), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[y, -4.5e+231], N[(N[((-i) * j + N[(z * x), $MachinePrecision]), $MachinePrecision] * y), $MachinePrecision], If[LessEqual[y, -2.7e-24], t$95$1, If[LessEqual[y, 2.5e-106], N[(N[(j * c), $MachinePrecision] * a + N[(N[(i * N[(N[(b * t), $MachinePrecision] - N[(j * y), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - N[(N[(c * b), $MachinePrecision] * z), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 3 regimes

2. if y < -4.49999999999999991e231

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{y \cdot \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot i\right) \cdot j + x \cdot z\right) \cdot y \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(i\right)\right) \cdot j + x \cdot z\right) \cdot y \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(i\right), j, x \cdot z\right) \cdot y \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-i, j, x \cdot z\right) \cdot y \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

      8. lower-*.f64 39.1

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

   4. Applied rewrites 39.1%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y} \]

   if -4.49999999999999991e231 < y < -2.70000000000000007e-24 or 2.49999999999999991e-106 < y

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around 0

      \[\leadsto \color{blue}{j \cdot \left(a \cdot c - i \cdot y\right) + x \cdot \left(y \cdot z - a \cdot t\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j + \color{blue}{x} \cdot \left(y \cdot z - a \cdot t\right) \]

      2. *-commutative N/A

         \[\leadsto \left(c \cdot a - i \cdot y\right) \cdot j + x \cdot \left(y \cdot z - a \cdot t\right) \]

      3. *-commutative N/A

         \[\leadsto \left(c \cdot a - y \cdot i\right) \cdot j + x \cdot \left(y \cdot z - a \cdot t\right) \]

      4. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, \color{blue}{j}, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      5. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      6. lift--.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      8. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      9. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      10. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - t \cdot a\right) \cdot x\right) \]

      11. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - t \cdot a\right) \cdot x\right) \]

      12. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      13. lower--.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      14. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

      15. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

      16. lower-*.f64 59.9

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

   4. Applied rewrites 59.9%

      \[\leadsto \color{blue}{\mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right)} \]

   if -2.70000000000000007e-24 < y < 2.49999999999999991e-106

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in i around 0

      \[\leadsto \color{blue}{\left(a \cdot \left(c \cdot j\right) + \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right)\right) - b \cdot \left(c \cdot z\right)} \]
   3. Step-by-step derivation

      1. associate--l+ N/A

         \[\leadsto a \cdot \left(c \cdot j\right) + \color{blue}{\left(\left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right)} \]

      2. *-commutative N/A

         \[\leadsto \left(c \cdot j\right) \cdot a + \left(\color{blue}{\left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right)} - b \cdot \left(c \cdot z\right)\right) \]

      3. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot j, \color{blue}{a}, \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right) \]

      4. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right) \]

      5. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right) \]

      6. lower--.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, \left(i \cdot \left(-1 \cdot \left(j \cdot y\right) - -1 \cdot \left(b \cdot t\right)\right) + x \cdot \left(y \cdot z - a \cdot t\right)\right) - b \cdot \left(c \cdot z\right)\right) \]

   4. Applied rewrites 74.4%

      \[\leadsto \color{blue}{\mathsf{fma}\left(j \cdot c, a, \mathsf{fma}\left(\left(-j \cdot y\right) - \left(-b \cdot t\right), i, \left(z \cdot y - a \cdot t\right) \cdot x\right) - \left(c \cdot b\right) \cdot z\right)} \]

   5. Taylor expanded in x around 0

      \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

      2. lower--.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

      3. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

      4. lift-*.f64 58.6

         \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]

   7. Applied rewrites 58.6%

      \[\leadsto \mathsf{fma}\left(j \cdot c, a, i \cdot \left(b \cdot t - j \cdot y\right) - \left(c \cdot b\right) \cdot z\right) \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 5: 64.5% accurate, 1.2× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(i \cdot t - c \cdot z\right) \cdot b\\ \mathbf{if}\;b \leq -5.3 \cdot 10^{+147}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;b \leq 2.25 \cdot 10^{+184}:\\ \;\;\;\;\mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right)\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- (* i t) (* c z)) b)))
   (if (<= b -5.3e+147)
     t_1
     (if (<= b 2.25e+184)
       (fma (- (* c a) (* i y)) j (* (- (* z y) (* a t)) x))
       t_1))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((i * t) - (c * z)) * b;
	double tmp;
	if (b <= -5.3e+147) {
		tmp = t_1;
	} else if (b <= 2.25e+184) {
		tmp = fma(((c * a) - (i * y)), j, (((z * y) - (a * t)) * x));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(Float64(i * t) - Float64(c * z)) * b)
	tmp = 0.0
	if (b <= -5.3e+147)
		tmp = t_1;
	elseif (b <= 2.25e+184)
		tmp = fma(Float64(Float64(c * a) - Float64(i * y)), j, Float64(Float64(Float64(z * y) - Float64(a * t)) * x));
	else
		tmp = t_1;
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(N[(i * t), $MachinePrecision] - N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision]}, If[LessEqual[b, -5.3e+147], t$95$1, If[LessEqual[b, 2.25e+184], N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j + N[(N[(N[(z * y), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision] * x), $MachinePrecision]), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if b < -5.3000000000000002e147 or 2.25000000000000018e184 < b

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   if -5.3000000000000002e147 < b < 2.25000000000000018e184

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around 0

      \[\leadsto \color{blue}{j \cdot \left(a \cdot c - i \cdot y\right) + x \cdot \left(y \cdot z - a \cdot t\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j + \color{blue}{x} \cdot \left(y \cdot z - a \cdot t\right) \]

      2. *-commutative N/A

         \[\leadsto \left(c \cdot a - i \cdot y\right) \cdot j + x \cdot \left(y \cdot z - a \cdot t\right) \]

      3. *-commutative N/A

         \[\leadsto \left(c \cdot a - y \cdot i\right) \cdot j + x \cdot \left(y \cdot z - a \cdot t\right) \]

      4. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, \color{blue}{j}, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      5. lift-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      6. lift--.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - y \cdot i, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      8. lower-*.f64 N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, x \cdot \left(y \cdot z - a \cdot t\right)\right) \]

      9. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      10. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - t \cdot a\right) \cdot x\right) \]

      11. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - t \cdot a\right) \cdot x\right) \]

      12. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      13. lower--.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(y \cdot z - a \cdot t\right) \cdot x\right) \]

      14. *-commutative N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

      15. lower-*.f64 N/A

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

      16. lower-*.f64 59.9

          \[\leadsto \mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right) \]

   4. Applied rewrites 59.9%

      \[\leadsto \color{blue}{\mathsf{fma}\left(c \cdot a - i \cdot y, j, \left(z \cdot y - a \cdot t\right) \cdot x\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 6: 59.0% accurate, 1.2× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(-t\right) \cdot \left(a \cdot x - i \cdot b\right)\\ t_2 := j \cdot \left(c \cdot a - y \cdot i\right)\\ \mathbf{if}\;t \leq -1.6 \cdot 10^{+71}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;t \leq -1.2 \cdot 10^{-60}:\\ \;\;\;\;\left(z \cdot y\right) \cdot x + t\_2\\ \mathbf{elif}\;t \leq 9 \cdot 10^{+43}:\\ \;\;\;\;\left(-\left(c \cdot b\right) \cdot z\right) + t\_2\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- t) (- (* a x) (* i b)))) (t_2 (* j (- (* c a) (* y i)))))
   (if (<= t -1.6e+71)
     t_1
     (if (<= t -1.2e-60)
       (+ (* (* z y) x) t_2)
       (if (<= t 9e+43) (+ (- (* (* c b) z)) t_2) t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = -t * ((a * x) - (i * b));
	double t_2 = j * ((c * a) - (y * i));
	double tmp;
	if (t <= -1.6e+71) {
		tmp = t_1;
	} else if (t <= -1.2e-60) {
		tmp = ((z * y) * x) + t_2;
	} else if (t <= 9e+43) {
		tmp = -((c * b) * z) + t_2;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: t_2
    real(8) :: tmp
    t_1 = -t * ((a * x) - (i * b))
    t_2 = j * ((c * a) - (y * i))
    if (t <= (-1.6d+71)) then
        tmp = t_1
    else if (t <= (-1.2d-60)) then
        tmp = ((z * y) * x) + t_2
    else if (t <= 9d+43) then
        tmp = -((c * b) * z) + t_2
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = -t * ((a * x) - (i * b));
	double t_2 = j * ((c * a) - (y * i));
	double tmp;
	if (t <= -1.6e+71) {
		tmp = t_1;
	} else if (t <= -1.2e-60) {
		tmp = ((z * y) * x) + t_2;
	} else if (t <= 9e+43) {
		tmp = -((c * b) * z) + t_2;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = -t * ((a * x) - (i * b))
	t_2 = j * ((c * a) - (y * i))
	tmp = 0
	if t <= -1.6e+71:
		tmp = t_1
	elif t <= -1.2e-60:
		tmp = ((z * y) * x) + t_2
	elif t <= 9e+43:
		tmp = -((c * b) * z) + t_2
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(-t) * Float64(Float64(a * x) - Float64(i * b)))
	t_2 = Float64(j * Float64(Float64(c * a) - Float64(y * i)))
	tmp = 0.0
	if (t <= -1.6e+71)
		tmp = t_1;
	elseif (t <= -1.2e-60)
		tmp = Float64(Float64(Float64(z * y) * x) + t_2);
	elseif (t <= 9e+43)
		tmp = Float64(Float64(-Float64(Float64(c * b) * z)) + t_2);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = -t * ((a * x) - (i * b));
	t_2 = j * ((c * a) - (y * i));
	tmp = 0.0;
	if (t <= -1.6e+71)
		tmp = t_1;
	elseif (t <= -1.2e-60)
		tmp = ((z * y) * x) + t_2;
	elseif (t <= 9e+43)
		tmp = -((c * b) * z) + t_2;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[((-t) * N[(N[(a * x), $MachinePrecision] - N[(i * b), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]}, Block[{t$95$2 = N[(j * N[(N[(c * a), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[t, -1.6e+71], t$95$1, If[LessEqual[t, -1.2e-60], N[(N[(N[(z * y), $MachinePrecision] * x), $MachinePrecision] + t$95$2), $MachinePrecision], If[LessEqual[t, 9e+43], N[((-N[(N[(c * b), $MachinePrecision] * z), $MachinePrecision]) + t$95$2), $MachinePrecision], t$95$1]]]]]

Derivation

1. Split input into 3 regimes

2. if t < -1.60000000000000012e71 or 9e43 < t

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in t around -inf

      \[\leadsto \color{blue}{-1 \cdot \left(t \cdot \left(a \cdot x - b \cdot i\right)\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot t\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      2. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      3. lower-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      4. lower-neg.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      5. lower--.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b \cdot i}\right) \]

      6. lower-*.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b} \cdot i\right) \]

      7. *-commutative N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

      8. lower-*.f64 38.7

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

   4. Applied rewrites 38.7%

      \[\leadsto \color{blue}{\left(-t\right) \cdot \left(a \cdot x - i \cdot b\right)} \]

   if -1.60000000000000012e71 < t < -1.20000000000000005e-60

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z\right)} + j \cdot \left(c \cdot a - y \cdot i\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(y \cdot z\right) \cdot \color{blue}{x} + j \cdot \left(c \cdot a - y \cdot i\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(y \cdot z\right) \cdot \color{blue}{x} + j \cdot \left(c \cdot a - y \cdot i\right) \]

      3. *-commutative N/A

         \[\leadsto \left(z \cdot y\right) \cdot x + j \cdot \left(c \cdot a - y \cdot i\right) \]

      4. lower-*.f64 48.8

         \[\leadsto \left(z \cdot y\right) \cdot x + j \cdot \left(c \cdot a - y \cdot i\right) \]

   4. Applied rewrites 48.8%

      \[\leadsto \color{blue}{\left(z \cdot y\right) \cdot x} + j \cdot \left(c \cdot a - y \cdot i\right) \]

   if -1.20000000000000005e-60 < t < 9e43

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in c around inf

      \[\leadsto \color{blue}{-1 \cdot \left(b \cdot \left(c \cdot z\right)\right)} + j \cdot \left(c \cdot a - y \cdot i\right) \]
   3. Step-by-step derivation

      1. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(b \cdot \left(c \cdot z\right)\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

      2. lower-neg.f64 N/A

         \[\leadsto \left(-b \cdot \left(c \cdot z\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

      3. associate-*r* N/A

         \[\leadsto \left(-\left(b \cdot c\right) \cdot z\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

      4. lower-*.f64 N/A

         \[\leadsto \left(-\left(b \cdot c\right) \cdot z\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

      5. *-commutative N/A

         \[\leadsto \left(-\left(c \cdot b\right) \cdot z\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

      6. lower-*.f64 49.4

         \[\leadsto \left(-\left(c \cdot b\right) \cdot z\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   4. Applied rewrites 49.4%

      \[\leadsto \color{blue}{\left(-\left(c \cdot b\right) \cdot z\right)} + j \cdot \left(c \cdot a - y \cdot i\right) \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 7: 58.9% accurate, 1.4× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(-t\right) \cdot \left(a \cdot x - i \cdot b\right)\\ \mathbf{if}\;t \leq -1.6 \cdot 10^{+71}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;t \leq 1.32 \cdot 10^{+32}:\\ \;\;\;\;\left(z \cdot y\right) \cdot x + j \cdot \left(c \cdot a - y \cdot i\right)\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- t) (- (* a x) (* i b)))))
   (if (<= t -1.6e+71)
     t_1
     (if (<= t 1.32e+32) (+ (* (* z y) x) (* j (- (* c a) (* y i)))) t_1))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = -t * ((a * x) - (i * b));
	double tmp;
	if (t <= -1.6e+71) {
		tmp = t_1;
	} else if (t <= 1.32e+32) {
		tmp = ((z * y) * x) + (j * ((c * a) - (y * i)));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = -t * ((a * x) - (i * b))
    if (t <= (-1.6d+71)) then
        tmp = t_1
    else if (t <= 1.32d+32) then
        tmp = ((z * y) * x) + (j * ((c * a) - (y * i)))
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = -t * ((a * x) - (i * b));
	double tmp;
	if (t <= -1.6e+71) {
		tmp = t_1;
	} else if (t <= 1.32e+32) {
		tmp = ((z * y) * x) + (j * ((c * a) - (y * i)));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = -t * ((a * x) - (i * b))
	tmp = 0
	if t <= -1.6e+71:
		tmp = t_1
	elif t <= 1.32e+32:
		tmp = ((z * y) * x) + (j * ((c * a) - (y * i)))
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(-t) * Float64(Float64(a * x) - Float64(i * b)))
	tmp = 0.0
	if (t <= -1.6e+71)
		tmp = t_1;
	elseif (t <= 1.32e+32)
		tmp = Float64(Float64(Float64(z * y) * x) + Float64(j * Float64(Float64(c * a) - Float64(y * i))));
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = -t * ((a * x) - (i * b));
	tmp = 0.0;
	if (t <= -1.6e+71)
		tmp = t_1;
	elseif (t <= 1.32e+32)
		tmp = ((z * y) * x) + (j * ((c * a) - (y * i)));
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[((-t) * N[(N[(a * x), $MachinePrecision] - N[(i * b), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[t, -1.6e+71], t$95$1, If[LessEqual[t, 1.32e+32], N[(N[(N[(z * y), $MachinePrecision] * x), $MachinePrecision] + N[(j * N[(N[(c * a), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if t < -1.60000000000000012e71 or 1.31999999999999997e32 < t

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in t around -inf

      \[\leadsto \color{blue}{-1 \cdot \left(t \cdot \left(a \cdot x - b \cdot i\right)\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot t\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      2. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      3. lower-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      4. lower-neg.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      5. lower--.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b \cdot i}\right) \]

      6. lower-*.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b} \cdot i\right) \]

      7. *-commutative N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

      8. lower-*.f64 38.7

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

   4. Applied rewrites 38.7%

      \[\leadsto \color{blue}{\left(-t\right) \cdot \left(a \cdot x - i \cdot b\right)} \]

   if -1.60000000000000012e71 < t < 1.31999999999999997e32

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z\right)} + j \cdot \left(c \cdot a - y \cdot i\right) \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(y \cdot z\right) \cdot \color{blue}{x} + j \cdot \left(c \cdot a - y \cdot i\right) \]

      2. lower-*.f64 N/A

         \[\leadsto \left(y \cdot z\right) \cdot \color{blue}{x} + j \cdot \left(c \cdot a - y \cdot i\right) \]

      3. *-commutative N/A

         \[\leadsto \left(z \cdot y\right) \cdot x + j \cdot \left(c \cdot a - y \cdot i\right) \]

      4. lower-*.f64 48.8

         \[\leadsto \left(z \cdot y\right) \cdot x + j \cdot \left(c \cdot a - y \cdot i\right) \]

   4. Applied rewrites 48.8%

      \[\leadsto \color{blue}{\left(z \cdot y\right) \cdot x} + j \cdot \left(c \cdot a - y \cdot i\right) \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 8: 52.3% accurate, 1.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y\\ \mathbf{if}\;y \leq -2.1 \cdot 10^{-145}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;y \leq 7 \cdot 10^{-294}:\\ \;\;\;\;\left(i \cdot t - c \cdot z\right) \cdot b\\ \mathbf{elif}\;y \leq 8.5 \cdot 10^{-203}:\\ \;\;\;\;\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a\\ \mathbf{elif}\;y \leq 7.2 \cdot 10^{+80}:\\ \;\;\;\;\left(j \cdot a - b \cdot z\right) \cdot c\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (fma (- i) j (* z x)) y)))
   (if (<= y -2.1e-145)
     t_1
     (if (<= y 7e-294)
       (* (- (* i t) (* c z)) b)
       (if (<= y 8.5e-203)
         (* (fma (- t) x (* j c)) a)
         (if (<= y 7.2e+80) (* (- (* j a) (* b z)) c) t_1))))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = fma(-i, j, (z * x)) * y;
	double tmp;
	if (y <= -2.1e-145) {
		tmp = t_1;
	} else if (y <= 7e-294) {
		tmp = ((i * t) - (c * z)) * b;
	} else if (y <= 8.5e-203) {
		tmp = fma(-t, x, (j * c)) * a;
	} else if (y <= 7.2e+80) {
		tmp = ((j * a) - (b * z)) * c;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(fma(Float64(-i), j, Float64(z * x)) * y)
	tmp = 0.0
	if (y <= -2.1e-145)
		tmp = t_1;
	elseif (y <= 7e-294)
		tmp = Float64(Float64(Float64(i * t) - Float64(c * z)) * b);
	elseif (y <= 8.5e-203)
		tmp = Float64(fma(Float64(-t), x, Float64(j * c)) * a);
	elseif (y <= 7.2e+80)
		tmp = Float64(Float64(Float64(j * a) - Float64(b * z)) * c);
	else
		tmp = t_1;
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[((-i) * j + N[(z * x), $MachinePrecision]), $MachinePrecision] * y), $MachinePrecision]}, If[LessEqual[y, -2.1e-145], t$95$1, If[LessEqual[y, 7e-294], N[(N[(N[(i * t), $MachinePrecision] - N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision], If[LessEqual[y, 8.5e-203], N[(N[((-t) * x + N[(j * c), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision], If[LessEqual[y, 7.2e+80], N[(N[(N[(j * a), $MachinePrecision] - N[(b * z), $MachinePrecision]), $MachinePrecision] * c), $MachinePrecision], t$95$1]]]]]

Derivation

1. Split input into 4 regimes

2. if y < -2.09999999999999991e-145 or 7.1999999999999999e80 < y

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{y \cdot \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot i\right) \cdot j + x \cdot z\right) \cdot y \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(i\right)\right) \cdot j + x \cdot z\right) \cdot y \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(i\right), j, x \cdot z\right) \cdot y \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-i, j, x \cdot z\right) \cdot y \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

      8. lower-*.f64 39.1

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

   4. Applied rewrites 39.1%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y} \]

   if -2.09999999999999991e-145 < y < 7.00000000000000064e-294

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   if 7.00000000000000064e-294 < y < 8.50000000000000031e-203

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   if 8.50000000000000031e-203 < y < 7.1999999999999999e80

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in c around inf

      \[\leadsto \color{blue}{c \cdot \left(a \cdot j - b \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(a \cdot j - b \cdot z\right) \cdot \color{blue}{c} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(a \cdot j - b \cdot z\right) \cdot \color{blue}{c} \]

      3. lower--.f64 N/A

         \[\leadsto \left(a \cdot j - b \cdot z\right) \cdot c \]

      4. *-commutative N/A

         \[\leadsto \left(j \cdot a - b \cdot z\right) \cdot c \]

      5. lower-*.f64 N/A

         \[\leadsto \left(j \cdot a - b \cdot z\right) \cdot c \]

      6. lower-*.f64 40.1

         \[\leadsto \left(j \cdot a - b \cdot z\right) \cdot c \]

   4. Applied rewrites 40.1%

      \[\leadsto \color{blue}{\left(j \cdot a - b \cdot z\right) \cdot c} \]
3. Recombined 4 regimes into one program.
4. Add Preprocessing

Alternative 9: 50.1% accurate, 1.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y\\ \mathbf{if}\;y \leq -2.1 \cdot 10^{-145}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;y \leq 2 \cdot 10^{-293}:\\ \;\;\;\;\left(i \cdot t - c \cdot z\right) \cdot b\\ \mathbf{elif}\;y \leq 5.8 \cdot 10^{-203}:\\ \;\;\;\;x \cdot \left(y \cdot z - a \cdot t\right)\\ \mathbf{elif}\;y \leq 7.2 \cdot 10^{+80}:\\ \;\;\;\;\left(j \cdot a - b \cdot z\right) \cdot c\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (fma (- i) j (* z x)) y)))
   (if (<= y -2.1e-145)
     t_1
     (if (<= y 2e-293)
       (* (- (* i t) (* c z)) b)
       (if (<= y 5.8e-203)
         (* x (- (* y z) (* a t)))
         (if (<= y 7.2e+80) (* (- (* j a) (* b z)) c) t_1))))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = fma(-i, j, (z * x)) * y;
	double tmp;
	if (y <= -2.1e-145) {
		tmp = t_1;
	} else if (y <= 2e-293) {
		tmp = ((i * t) - (c * z)) * b;
	} else if (y <= 5.8e-203) {
		tmp = x * ((y * z) - (a * t));
	} else if (y <= 7.2e+80) {
		tmp = ((j * a) - (b * z)) * c;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(fma(Float64(-i), j, Float64(z * x)) * y)
	tmp = 0.0
	if (y <= -2.1e-145)
		tmp = t_1;
	elseif (y <= 2e-293)
		tmp = Float64(Float64(Float64(i * t) - Float64(c * z)) * b);
	elseif (y <= 5.8e-203)
		tmp = Float64(x * Float64(Float64(y * z) - Float64(a * t)));
	elseif (y <= 7.2e+80)
		tmp = Float64(Float64(Float64(j * a) - Float64(b * z)) * c);
	else
		tmp = t_1;
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[((-i) * j + N[(z * x), $MachinePrecision]), $MachinePrecision] * y), $MachinePrecision]}, If[LessEqual[y, -2.1e-145], t$95$1, If[LessEqual[y, 2e-293], N[(N[(N[(i * t), $MachinePrecision] - N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision], If[LessEqual[y, 5.8e-203], N[(x * N[(N[(y * z), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], If[LessEqual[y, 7.2e+80], N[(N[(N[(j * a), $MachinePrecision] - N[(b * z), $MachinePrecision]), $MachinePrecision] * c), $MachinePrecision], t$95$1]]]]]

Derivation

1. Split input into 4 regimes

2. if y < -2.09999999999999991e-145 or 7.1999999999999999e80 < y

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{y \cdot \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(i \cdot j\right) + x \cdot z\right) \cdot \color{blue}{y} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot i\right) \cdot j + x \cdot z\right) \cdot y \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(i\right)\right) \cdot j + x \cdot z\right) \cdot y \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(i\right), j, x \cdot z\right) \cdot y \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-i, j, x \cdot z\right) \cdot y \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

      8. lower-*.f64 39.1

         \[\leadsto \mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y \]

   4. Applied rewrites 39.1%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-i, j, z \cdot x\right) \cdot y} \]

   if -2.09999999999999991e-145 < y < 2.0000000000000001e-293

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   if 2.0000000000000001e-293 < y < 5.7999999999999998e-203

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]

   if 5.7999999999999998e-203 < y < 7.1999999999999999e80

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in c around inf

      \[\leadsto \color{blue}{c \cdot \left(a \cdot j - b \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(a \cdot j - b \cdot z\right) \cdot \color{blue}{c} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(a \cdot j - b \cdot z\right) \cdot \color{blue}{c} \]

      3. lower--.f64 N/A

         \[\leadsto \left(a \cdot j - b \cdot z\right) \cdot c \]

      4. *-commutative N/A

         \[\leadsto \left(j \cdot a - b \cdot z\right) \cdot c \]

      5. lower-*.f64 N/A

         \[\leadsto \left(j \cdot a - b \cdot z\right) \cdot c \]

      6. lower-*.f64 40.1

         \[\leadsto \left(j \cdot a - b \cdot z\right) \cdot c \]

   4. Applied rewrites 40.1%

      \[\leadsto \color{blue}{\left(j \cdot a - b \cdot z\right) \cdot c} \]
3. Recombined 4 regimes into one program.
4. Add Preprocessing

Alternative 10: 50.0% accurate, 2.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(y \cdot x - c \cdot b\right) \cdot z\\ \mathbf{if}\;z \leq -4.4 \cdot 10^{-28}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;z \leq 6.8 \cdot 10^{+46}:\\ \;\;\;\;\left(a \cdot c - i \cdot y\right) \cdot j\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- (* y x) (* c b)) z)))
   (if (<= z -4.4e-28) t_1 (if (<= z 6.8e+46) (* (- (* a c) (* i y)) j) t_1))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((y * x) - (c * b)) * z;
	double tmp;
	if (z <= -4.4e-28) {
		tmp = t_1;
	} else if (z <= 6.8e+46) {
		tmp = ((a * c) - (i * y)) * j;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = ((y * x) - (c * b)) * z
    if (z <= (-4.4d-28)) then
        tmp = t_1
    else if (z <= 6.8d+46) then
        tmp = ((a * c) - (i * y)) * j
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((y * x) - (c * b)) * z;
	double tmp;
	if (z <= -4.4e-28) {
		tmp = t_1;
	} else if (z <= 6.8e+46) {
		tmp = ((a * c) - (i * y)) * j;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = ((y * x) - (c * b)) * z
	tmp = 0
	if z <= -4.4e-28:
		tmp = t_1
	elif z <= 6.8e+46:
		tmp = ((a * c) - (i * y)) * j
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(Float64(y * x) - Float64(c * b)) * z)
	tmp = 0.0
	if (z <= -4.4e-28)
		tmp = t_1;
	elseif (z <= 6.8e+46)
		tmp = Float64(Float64(Float64(a * c) - Float64(i * y)) * j);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = ((y * x) - (c * b)) * z;
	tmp = 0.0;
	if (z <= -4.4e-28)
		tmp = t_1;
	elseif (z <= 6.8e+46)
		tmp = ((a * c) - (i * y)) * j;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(N[(y * x), $MachinePrecision] - N[(c * b), $MachinePrecision]), $MachinePrecision] * z), $MachinePrecision]}, If[LessEqual[z, -4.4e-28], t$95$1, If[LessEqual[z, 6.8e+46], N[(N[(N[(a * c), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if z < -4.39999999999999992e-28 or 6.7999999999999996e46 < z

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in z around inf

      \[\leadsto \color{blue}{z \cdot \left(x \cdot y - b \cdot c\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(x \cdot y - b \cdot c\right) \cdot \color{blue}{z} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(x \cdot y - b \cdot c\right) \cdot \color{blue}{z} \]

      3. lower--.f64 N/A

         \[\leadsto \left(x \cdot y - b \cdot c\right) \cdot z \]

      4. *-commutative N/A

         \[\leadsto \left(y \cdot x - b \cdot c\right) \cdot z \]

      5. lower-*.f64 N/A

         \[\leadsto \left(y \cdot x - b \cdot c\right) \cdot z \]

      6. *-commutative N/A

         \[\leadsto \left(y \cdot x - c \cdot b\right) \cdot z \]

      7. lower-*.f64 39.0

         \[\leadsto \left(y \cdot x - c \cdot b\right) \cdot z \]

   4. Applied rewrites 39.0%

      \[\leadsto \color{blue}{\left(y \cdot x - c \cdot b\right) \cdot z} \]

   if -4.39999999999999992e-28 < z < 6.7999999999999996e46

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \left(x \cdot \left(\frac{y \cdot z}{j} - \frac{a \cdot t}{j}\right)\right) \cdot j \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(x \cdot \left(\frac{y \cdot z}{j} - \frac{a \cdot t}{j}\right)\right) \cdot j \]

      2. sub-div N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      3. lower-/.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      4. lower--.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      5. lower-*.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      6. lift-*.f64 35.7

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

   7. Applied rewrites 35.7%

      \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

   8. Taylor expanded in j around inf

      \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]
   9. Step-by-step derivation

      1. lower--.f64 N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

      2. lower-*.f64 N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

      3. lower-*.f64 38.5

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

   10. Applied rewrites 38.5%

       \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 11: 49.8% accurate, 1.7× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(a \cdot c - i \cdot y\right) \cdot j\\ \mathbf{if}\;j \leq -2.4 \cdot 10^{+51}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;j \leq -9.2 \cdot 10^{-233}:\\ \;\;\;\;\left(i \cdot t - c \cdot z\right) \cdot b\\ \mathbf{elif}\;j \leq 1.2 \cdot 10^{+171}:\\ \;\;\;\;x \cdot \left(y \cdot z - a \cdot t\right)\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- (* a c) (* i y)) j)))
   (if (<= j -2.4e+51)
     t_1
     (if (<= j -9.2e-233)
       (* (- (* i t) (* c z)) b)
       (if (<= j 1.2e+171) (* x (- (* y z) (* a t))) t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((a * c) - (i * y)) * j;
	double tmp;
	if (j <= -2.4e+51) {
		tmp = t_1;
	} else if (j <= -9.2e-233) {
		tmp = ((i * t) - (c * z)) * b;
	} else if (j <= 1.2e+171) {
		tmp = x * ((y * z) - (a * t));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = ((a * c) - (i * y)) * j
    if (j <= (-2.4d+51)) then
        tmp = t_1
    else if (j <= (-9.2d-233)) then
        tmp = ((i * t) - (c * z)) * b
    else if (j <= 1.2d+171) then
        tmp = x * ((y * z) - (a * t))
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((a * c) - (i * y)) * j;
	double tmp;
	if (j <= -2.4e+51) {
		tmp = t_1;
	} else if (j <= -9.2e-233) {
		tmp = ((i * t) - (c * z)) * b;
	} else if (j <= 1.2e+171) {
		tmp = x * ((y * z) - (a * t));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = ((a * c) - (i * y)) * j
	tmp = 0
	if j <= -2.4e+51:
		tmp = t_1
	elif j <= -9.2e-233:
		tmp = ((i * t) - (c * z)) * b
	elif j <= 1.2e+171:
		tmp = x * ((y * z) - (a * t))
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(Float64(a * c) - Float64(i * y)) * j)
	tmp = 0.0
	if (j <= -2.4e+51)
		tmp = t_1;
	elseif (j <= -9.2e-233)
		tmp = Float64(Float64(Float64(i * t) - Float64(c * z)) * b);
	elseif (j <= 1.2e+171)
		tmp = Float64(x * Float64(Float64(y * z) - Float64(a * t)));
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = ((a * c) - (i * y)) * j;
	tmp = 0.0;
	if (j <= -2.4e+51)
		tmp = t_1;
	elseif (j <= -9.2e-233)
		tmp = ((i * t) - (c * z)) * b;
	elseif (j <= 1.2e+171)
		tmp = x * ((y * z) - (a * t));
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(N[(a * c), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j), $MachinePrecision]}, If[LessEqual[j, -2.4e+51], t$95$1, If[LessEqual[j, -9.2e-233], N[(N[(N[(i * t), $MachinePrecision] - N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision], If[LessEqual[j, 1.2e+171], N[(x * N[(N[(y * z), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 3 regimes

2. if j < -2.3999999999999999e51 or 1.19999999999999999e171 < j

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \left(x \cdot \left(\frac{y \cdot z}{j} - \frac{a \cdot t}{j}\right)\right) \cdot j \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(x \cdot \left(\frac{y \cdot z}{j} - \frac{a \cdot t}{j}\right)\right) \cdot j \]

      2. sub-div N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      3. lower-/.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      4. lower--.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      5. lower-*.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      6. lift-*.f64 35.7

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

   7. Applied rewrites 35.7%

      \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

   8. Taylor expanded in j around inf

      \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]
   9. Step-by-step derivation

      1. lower--.f64 N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

      2. lower-*.f64 N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

      3. lower-*.f64 38.5

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

   10. Applied rewrites 38.5%

       \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

   if -2.3999999999999999e51 < j < -9.2000000000000007e-233

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   if -9.2000000000000007e-233 < j < 1.19999999999999999e171

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 12: 48.5% accurate, 2.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(a \cdot c - i \cdot y\right) \cdot j\\ \mathbf{if}\;j \leq -4.7 \cdot 10^{-32}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;j \leq 1.2 \cdot 10^{+171}:\\ \;\;\;\;x \cdot \left(y \cdot z - a \cdot t\right)\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- (* a c) (* i y)) j)))
   (if (<= j -4.7e-32)
     t_1
     (if (<= j 1.2e+171) (* x (- (* y z) (* a t))) t_1))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((a * c) - (i * y)) * j;
	double tmp;
	if (j <= -4.7e-32) {
		tmp = t_1;
	} else if (j <= 1.2e+171) {
		tmp = x * ((y * z) - (a * t));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = ((a * c) - (i * y)) * j
    if (j <= (-4.7d-32)) then
        tmp = t_1
    else if (j <= 1.2d+171) then
        tmp = x * ((y * z) - (a * t))
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = ((a * c) - (i * y)) * j;
	double tmp;
	if (j <= -4.7e-32) {
		tmp = t_1;
	} else if (j <= 1.2e+171) {
		tmp = x * ((y * z) - (a * t));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = ((a * c) - (i * y)) * j
	tmp = 0
	if j <= -4.7e-32:
		tmp = t_1
	elif j <= 1.2e+171:
		tmp = x * ((y * z) - (a * t))
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(Float64(a * c) - Float64(i * y)) * j)
	tmp = 0.0
	if (j <= -4.7e-32)
		tmp = t_1;
	elseif (j <= 1.2e+171)
		tmp = Float64(x * Float64(Float64(y * z) - Float64(a * t)));
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = ((a * c) - (i * y)) * j;
	tmp = 0.0;
	if (j <= -4.7e-32)
		tmp = t_1;
	elseif (j <= 1.2e+171)
		tmp = x * ((y * z) - (a * t));
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(N[(a * c), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j), $MachinePrecision]}, If[LessEqual[j, -4.7e-32], t$95$1, If[LessEqual[j, 1.2e+171], N[(x * N[(N[(y * z), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if j < -4.70000000000000019e-32 or 1.19999999999999999e171 < j

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \left(x \cdot \left(\frac{y \cdot z}{j} - \frac{a \cdot t}{j}\right)\right) \cdot j \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(x \cdot \left(\frac{y \cdot z}{j} - \frac{a \cdot t}{j}\right)\right) \cdot j \]

      2. sub-div N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      3. lower-/.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      4. lower--.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      5. lower-*.f64 N/A

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

      6. lift-*.f64 35.7

         \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

   7. Applied rewrites 35.7%

      \[\leadsto \left(x \cdot \frac{y \cdot z - a \cdot t}{j}\right) \cdot j \]

   8. Taylor expanded in j around inf

      \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]
   9. Step-by-step derivation

      1. lower--.f64 N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

      2. lower-*.f64 N/A

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

      3. lower-*.f64 38.5

         \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

   10. Applied rewrites 38.5%

       \[\leadsto \left(a \cdot c - i \cdot y\right) \cdot j \]

   if -4.70000000000000019e-32 < j < 1.19999999999999999e171

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 13: 42.5% accurate, 1.7× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := x \cdot \left(y \cdot z - a \cdot t\right)\\ \mathbf{if}\;x \leq -3.3 \cdot 10^{-15}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;x \leq -2.7 \cdot 10^{-186}:\\ \;\;\;\;\left(c \cdot j\right) \cdot a\\ \mathbf{elif}\;x \leq 9.8 \cdot 10^{-139}:\\ \;\;\;\;\left(-t\right) \cdot \left(-1 \cdot \left(b \cdot i\right)\right)\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* x (- (* y z) (* a t)))))
   (if (<= x -3.3e-15)
     t_1
     (if (<= x -2.7e-186)
       (* (* c j) a)
       (if (<= x 9.8e-139) (* (- t) (* -1.0 (* b i))) t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = x * ((y * z) - (a * t));
	double tmp;
	if (x <= -3.3e-15) {
		tmp = t_1;
	} else if (x <= -2.7e-186) {
		tmp = (c * j) * a;
	} else if (x <= 9.8e-139) {
		tmp = -t * (-1.0 * (b * i));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = x * ((y * z) - (a * t))
    if (x <= (-3.3d-15)) then
        tmp = t_1
    else if (x <= (-2.7d-186)) then
        tmp = (c * j) * a
    else if (x <= 9.8d-139) then
        tmp = -t * ((-1.0d0) * (b * i))
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = x * ((y * z) - (a * t));
	double tmp;
	if (x <= -3.3e-15) {
		tmp = t_1;
	} else if (x <= -2.7e-186) {
		tmp = (c * j) * a;
	} else if (x <= 9.8e-139) {
		tmp = -t * (-1.0 * (b * i));
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = x * ((y * z) - (a * t))
	tmp = 0
	if x <= -3.3e-15:
		tmp = t_1
	elif x <= -2.7e-186:
		tmp = (c * j) * a
	elif x <= 9.8e-139:
		tmp = -t * (-1.0 * (b * i))
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(x * Float64(Float64(y * z) - Float64(a * t)))
	tmp = 0.0
	if (x <= -3.3e-15)
		tmp = t_1;
	elseif (x <= -2.7e-186)
		tmp = Float64(Float64(c * j) * a);
	elseif (x <= 9.8e-139)
		tmp = Float64(Float64(-t) * Float64(-1.0 * Float64(b * i)));
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = x * ((y * z) - (a * t));
	tmp = 0.0;
	if (x <= -3.3e-15)
		tmp = t_1;
	elseif (x <= -2.7e-186)
		tmp = (c * j) * a;
	elseif (x <= 9.8e-139)
		tmp = -t * (-1.0 * (b * i));
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(x * N[(N[(y * z), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[x, -3.3e-15], t$95$1, If[LessEqual[x, -2.7e-186], N[(N[(c * j), $MachinePrecision] * a), $MachinePrecision], If[LessEqual[x, 9.8e-139], N[((-t) * N[(-1.0 * N[(b * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 3 regimes

2. if x < -3.3e-15 or 9.80000000000000063e-139 < x

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]

   if -3.3e-15 < x < -2.6999999999999999e-186

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(c \cdot j\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 22.4

         \[\leadsto \left(c \cdot j\right) \cdot a \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(c \cdot j\right) \cdot a \]

   if -2.6999999999999999e-186 < x < 9.80000000000000063e-139

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in t around -inf

      \[\leadsto \color{blue}{-1 \cdot \left(t \cdot \left(a \cdot x - b \cdot i\right)\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot t\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      2. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      3. lower-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      4. lower-neg.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      5. lower--.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b \cdot i}\right) \]

      6. lower-*.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b} \cdot i\right) \]

      7. *-commutative N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

      8. lower-*.f64 38.7

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

   4. Applied rewrites 38.7%

      \[\leadsto \color{blue}{\left(-t\right) \cdot \left(a \cdot x - i \cdot b\right)} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(-t\right) \cdot \left(-1 \cdot \color{blue}{\left(b \cdot i\right)}\right) \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(-1 \cdot \left(b \cdot \color{blue}{i}\right)\right) \]

      2. lower-*.f64 22.7

         \[\leadsto \left(-t\right) \cdot \left(-1 \cdot \left(b \cdot i\right)\right) \]

   7. Applied rewrites 22.7%

      \[\leadsto \left(-t\right) \cdot \left(-1 \cdot \color{blue}{\left(b \cdot i\right)}\right) \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 14: 30.7% accurate, 1.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a\\ \mathbf{if}\;t \leq -1.8 \cdot 10^{+195}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;t \leq -9.2 \cdot 10^{-60}:\\ \;\;\;\;\left(-t\right) \cdot \left(-1 \cdot \left(b \cdot i\right)\right)\\ \mathbf{elif}\;t \leq 1.8 \cdot 10^{-19}:\\ \;\;\;\;\left(-1 \cdot \left(c \cdot z\right)\right) \cdot b\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* -1.0 (* t x)) a)))
   (if (<= t -1.8e+195)
     t_1
     (if (<= t -9.2e-60)
       (* (- t) (* -1.0 (* b i)))
       (if (<= t 1.8e-19) (* (* -1.0 (* c z)) b) t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (-1.0 * (t * x)) * a;
	double tmp;
	if (t <= -1.8e+195) {
		tmp = t_1;
	} else if (t <= -9.2e-60) {
		tmp = -t * (-1.0 * (b * i));
	} else if (t <= 1.8e-19) {
		tmp = (-1.0 * (c * z)) * b;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = ((-1.0d0) * (t * x)) * a
    if (t <= (-1.8d+195)) then
        tmp = t_1
    else if (t <= (-9.2d-60)) then
        tmp = -t * ((-1.0d0) * (b * i))
    else if (t <= 1.8d-19) then
        tmp = ((-1.0d0) * (c * z)) * b
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (-1.0 * (t * x)) * a;
	double tmp;
	if (t <= -1.8e+195) {
		tmp = t_1;
	} else if (t <= -9.2e-60) {
		tmp = -t * (-1.0 * (b * i));
	} else if (t <= 1.8e-19) {
		tmp = (-1.0 * (c * z)) * b;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (-1.0 * (t * x)) * a
	tmp = 0
	if t <= -1.8e+195:
		tmp = t_1
	elif t <= -9.2e-60:
		tmp = -t * (-1.0 * (b * i))
	elif t <= 1.8e-19:
		tmp = (-1.0 * (c * z)) * b
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(-1.0 * Float64(t * x)) * a)
	tmp = 0.0
	if (t <= -1.8e+195)
		tmp = t_1;
	elseif (t <= -9.2e-60)
		tmp = Float64(Float64(-t) * Float64(-1.0 * Float64(b * i)));
	elseif (t <= 1.8e-19)
		tmp = Float64(Float64(-1.0 * Float64(c * z)) * b);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = (-1.0 * (t * x)) * a;
	tmp = 0.0;
	if (t <= -1.8e+195)
		tmp = t_1;
	elseif (t <= -9.2e-60)
		tmp = -t * (-1.0 * (b * i));
	elseif (t <= 1.8e-19)
		tmp = (-1.0 * (c * z)) * b;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(-1.0 * N[(t * x), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision]}, If[LessEqual[t, -1.8e+195], t$95$1, If[LessEqual[t, -9.2e-60], N[((-t) * N[(-1.0 * N[(b * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], If[LessEqual[t, 1.8e-19], N[(N[(-1.0 * N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 3 regimes

2. if t < -1.7999999999999999e195 or 1.8000000000000001e-19 < t

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around inf

      \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

      2. lower-*.f64 22.1

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

   7. Applied rewrites 22.1%

      \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

   if -1.7999999999999999e195 < t < -9.2000000000000005e-60

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in t around -inf

      \[\leadsto \color{blue}{-1 \cdot \left(t \cdot \left(a \cdot x - b \cdot i\right)\right)} \]
   3. Step-by-step derivation

      1. associate-*r* N/A

         \[\leadsto \left(-1 \cdot t\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      2. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      3. lower-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(t\right)\right) \cdot \color{blue}{\left(a \cdot x - b \cdot i\right)} \]

      4. lower-neg.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(\color{blue}{a \cdot x} - b \cdot i\right) \]

      5. lower--.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b \cdot i}\right) \]

      6. lower-*.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - \color{blue}{b} \cdot i\right) \]

      7. *-commutative N/A

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

      8. lower-*.f64 38.7

         \[\leadsto \left(-t\right) \cdot \left(a \cdot x - i \cdot \color{blue}{b}\right) \]

   4. Applied rewrites 38.7%

      \[\leadsto \color{blue}{\left(-t\right) \cdot \left(a \cdot x - i \cdot b\right)} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(-t\right) \cdot \left(-1 \cdot \color{blue}{\left(b \cdot i\right)}\right) \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-t\right) \cdot \left(-1 \cdot \left(b \cdot \color{blue}{i}\right)\right) \]

      2. lower-*.f64 22.7

         \[\leadsto \left(-t\right) \cdot \left(-1 \cdot \left(b \cdot i\right)\right) \]

   7. Applied rewrites 22.7%

      \[\leadsto \left(-t\right) \cdot \left(-1 \cdot \color{blue}{\left(b \cdot i\right)}\right) \]

   if -9.2000000000000005e-60 < t < 1.8000000000000001e-19

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   5. Taylor expanded in z around inf

      \[\leadsto \left(-1 \cdot \left(c \cdot z\right)\right) \cdot b \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(c \cdot z\right)\right) \cdot b \]

      2. lift-*.f64 23.0

         \[\leadsto \left(-1 \cdot \left(c \cdot z\right)\right) \cdot b \]

   7. Applied rewrites 23.0%

      \[\leadsto \left(-1 \cdot \left(c \cdot z\right)\right) \cdot b \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 15: 30.7% accurate, 1.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;c \leq -1.02 \cdot 10^{-65}:\\ \;\;\;\;\left(c \cdot j\right) \cdot a\\ \mathbf{elif}\;c \leq 3 \cdot 10^{-249}:\\ \;\;\;\;\left(i \cdot t\right) \cdot b\\ \mathbf{elif}\;c \leq 9.5 \cdot 10^{+38}:\\ \;\;\;\;\left(-1 \cdot \left(t \cdot x\right)\right) \cdot a\\ \mathbf{else}:\\ \;\;\;\;\left(-1 \cdot \left(b \cdot c\right)\right) \cdot z\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (if (<= c -1.02e-65)
   (* (* c j) a)
   (if (<= c 3e-249)
     (* (* i t) b)
     (if (<= c 9.5e+38) (* (* -1.0 (* t x)) a) (* (* -1.0 (* b c)) z)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double tmp;
	if (c <= -1.02e-65) {
		tmp = (c * j) * a;
	} else if (c <= 3e-249) {
		tmp = (i * t) * b;
	} else if (c <= 9.5e+38) {
		tmp = (-1.0 * (t * x)) * a;
	} else {
		tmp = (-1.0 * (b * c)) * z;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: tmp
    if (c <= (-1.02d-65)) then
        tmp = (c * j) * a
    else if (c <= 3d-249) then
        tmp = (i * t) * b
    else if (c <= 9.5d+38) then
        tmp = ((-1.0d0) * (t * x)) * a
    else
        tmp = ((-1.0d0) * (b * c)) * z
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double tmp;
	if (c <= -1.02e-65) {
		tmp = (c * j) * a;
	} else if (c <= 3e-249) {
		tmp = (i * t) * b;
	} else if (c <= 9.5e+38) {
		tmp = (-1.0 * (t * x)) * a;
	} else {
		tmp = (-1.0 * (b * c)) * z;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	tmp = 0
	if c <= -1.02e-65:
		tmp = (c * j) * a
	elif c <= 3e-249:
		tmp = (i * t) * b
	elif c <= 9.5e+38:
		tmp = (-1.0 * (t * x)) * a
	else:
		tmp = (-1.0 * (b * c)) * z
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0
	if (c <= -1.02e-65)
		tmp = Float64(Float64(c * j) * a);
	elseif (c <= 3e-249)
		tmp = Float64(Float64(i * t) * b);
	elseif (c <= 9.5e+38)
		tmp = Float64(Float64(-1.0 * Float64(t * x)) * a);
	else
		tmp = Float64(Float64(-1.0 * Float64(b * c)) * z);
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0;
	if (c <= -1.02e-65)
		tmp = (c * j) * a;
	elseif (c <= 3e-249)
		tmp = (i * t) * b;
	elseif (c <= 9.5e+38)
		tmp = (-1.0 * (t * x)) * a;
	else
		tmp = (-1.0 * (b * c)) * z;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := If[LessEqual[c, -1.02e-65], N[(N[(c * j), $MachinePrecision] * a), $MachinePrecision], If[LessEqual[c, 3e-249], N[(N[(i * t), $MachinePrecision] * b), $MachinePrecision], If[LessEqual[c, 9.5e+38], N[(N[(-1.0 * N[(t * x), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision], N[(N[(-1.0 * N[(b * c), $MachinePrecision]), $MachinePrecision] * z), $MachinePrecision]]]]

Derivation

1. Split input into 4 regimes

2. if c < -1.02000000000000004e-65

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(c \cdot j\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 22.4

         \[\leadsto \left(c \cdot j\right) \cdot a \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(c \cdot j\right) \cdot a \]

   if -1.02000000000000004e-65 < c < 3.00000000000000004e-249

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   5. Taylor expanded in z around 0

      \[\leadsto \left(i \cdot t\right) \cdot b \]
   6. Step-by-step derivation

      1. lift-*.f64 22.4

         \[\leadsto \left(i \cdot t\right) \cdot b \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(i \cdot t\right) \cdot b \]

   if 3.00000000000000004e-249 < c < 9.4999999999999995e38

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around inf

      \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

      2. lower-*.f64 22.1

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

   7. Applied rewrites 22.1%

      \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

   if 9.4999999999999995e38 < c

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in z around inf

      \[\leadsto \color{blue}{z \cdot \left(x \cdot y - b \cdot c\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(x \cdot y - b \cdot c\right) \cdot \color{blue}{z} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(x \cdot y - b \cdot c\right) \cdot \color{blue}{z} \]

      3. lower--.f64 N/A

         \[\leadsto \left(x \cdot y - b \cdot c\right) \cdot z \]

      4. *-commutative N/A

         \[\leadsto \left(y \cdot x - b \cdot c\right) \cdot z \]

      5. lower-*.f64 N/A

         \[\leadsto \left(y \cdot x - b \cdot c\right) \cdot z \]

      6. *-commutative N/A

         \[\leadsto \left(y \cdot x - c \cdot b\right) \cdot z \]

      7. lower-*.f64 39.0

         \[\leadsto \left(y \cdot x - c \cdot b\right) \cdot z \]

   4. Applied rewrites 39.0%

      \[\leadsto \color{blue}{\left(y \cdot x - c \cdot b\right) \cdot z} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(-1 \cdot \left(b \cdot c\right)\right) \cdot z \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(b \cdot c\right)\right) \cdot z \]

      2. lower-*.f64 22.8

         \[\leadsto \left(-1 \cdot \left(b \cdot c\right)\right) \cdot z \]

   7. Applied rewrites 22.8%

      \[\leadsto \left(-1 \cdot \left(b \cdot c\right)\right) \cdot z \]
3. Recombined 4 regimes into one program.
4. Add Preprocessing

Alternative 16: 30.3% accurate, 1.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a\\ \mathbf{if}\;t \leq -2.4 \cdot 10^{+129}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;t \leq -6.4 \cdot 10^{-59}:\\ \;\;\;\;x \cdot \left(y \cdot z\right)\\ \mathbf{elif}\;t \leq 1.8 \cdot 10^{-19}:\\ \;\;\;\;\left(-1 \cdot \left(c \cdot z\right)\right) \cdot b\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* -1.0 (* t x)) a)))
   (if (<= t -2.4e+129)
     t_1
     (if (<= t -6.4e-59)
       (* x (* y z))
       (if (<= t 1.8e-19) (* (* -1.0 (* c z)) b) t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (-1.0 * (t * x)) * a;
	double tmp;
	if (t <= -2.4e+129) {
		tmp = t_1;
	} else if (t <= -6.4e-59) {
		tmp = x * (y * z);
	} else if (t <= 1.8e-19) {
		tmp = (-1.0 * (c * z)) * b;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = ((-1.0d0) * (t * x)) * a
    if (t <= (-2.4d+129)) then
        tmp = t_1
    else if (t <= (-6.4d-59)) then
        tmp = x * (y * z)
    else if (t <= 1.8d-19) then
        tmp = ((-1.0d0) * (c * z)) * b
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (-1.0 * (t * x)) * a;
	double tmp;
	if (t <= -2.4e+129) {
		tmp = t_1;
	} else if (t <= -6.4e-59) {
		tmp = x * (y * z);
	} else if (t <= 1.8e-19) {
		tmp = (-1.0 * (c * z)) * b;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (-1.0 * (t * x)) * a
	tmp = 0
	if t <= -2.4e+129:
		tmp = t_1
	elif t <= -6.4e-59:
		tmp = x * (y * z)
	elif t <= 1.8e-19:
		tmp = (-1.0 * (c * z)) * b
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(-1.0 * Float64(t * x)) * a)
	tmp = 0.0
	if (t <= -2.4e+129)
		tmp = t_1;
	elseif (t <= -6.4e-59)
		tmp = Float64(x * Float64(y * z));
	elseif (t <= 1.8e-19)
		tmp = Float64(Float64(-1.0 * Float64(c * z)) * b);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = (-1.0 * (t * x)) * a;
	tmp = 0.0;
	if (t <= -2.4e+129)
		tmp = t_1;
	elseif (t <= -6.4e-59)
		tmp = x * (y * z);
	elseif (t <= 1.8e-19)
		tmp = (-1.0 * (c * z)) * b;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(-1.0 * N[(t * x), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision]}, If[LessEqual[t, -2.4e+129], t$95$1, If[LessEqual[t, -6.4e-59], N[(x * N[(y * z), $MachinePrecision]), $MachinePrecision], If[LessEqual[t, 1.8e-19], N[(N[(-1.0 * N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 3 regimes

2. if t < -2.3999999999999999e129 or 1.8000000000000001e-19 < t

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around inf

      \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

      2. lower-*.f64 22.1

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

   7. Applied rewrites 22.1%

      \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

   if -2.3999999999999999e129 < t < -6.3999999999999998e-59

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]

   8. Taylor expanded in y around inf

      \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]
   9. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot \color{blue}{z}\right) \]

      2. lift-*.f64 22.2

         \[\leadsto x \cdot \left(y \cdot z\right) \]

   10. Applied rewrites 22.2%

       \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]

   if -6.3999999999999998e-59 < t < 1.8000000000000001e-19

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   5. Taylor expanded in z around inf

      \[\leadsto \left(-1 \cdot \left(c \cdot z\right)\right) \cdot b \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(c \cdot z\right)\right) \cdot b \]

      2. lift-*.f64 23.0

         \[\leadsto \left(-1 \cdot \left(c \cdot z\right)\right) \cdot b \]

   7. Applied rewrites 23.0%

      \[\leadsto \left(-1 \cdot \left(c \cdot z\right)\right) \cdot b \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 17: 29.9% accurate, 1.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(i \cdot t\right) \cdot b\\ \mathbf{if}\;i \leq -1.6 \cdot 10^{+53}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;i \leq -7.5 \cdot 10^{-117}:\\ \;\;\;\;x \cdot \left(-1 \cdot \left(a \cdot t\right)\right)\\ \mathbf{elif}\;i \leq -8.5 \cdot 10^{-288}:\\ \;\;\;\;x \cdot \left(y \cdot z\right)\\ \mathbf{elif}\;i \leq 1.56 \cdot 10^{-27}:\\ \;\;\;\;\left(c \cdot j\right) \cdot a\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* i t) b)))
   (if (<= i -1.6e+53)
     t_1
     (if (<= i -7.5e-117)
       (* x (* -1.0 (* a t)))
       (if (<= i -8.5e-288)
         (* x (* y z))
         (if (<= i 1.56e-27) (* (* c j) a) t_1))))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (i * t) * b;
	double tmp;
	if (i <= -1.6e+53) {
		tmp = t_1;
	} else if (i <= -7.5e-117) {
		tmp = x * (-1.0 * (a * t));
	} else if (i <= -8.5e-288) {
		tmp = x * (y * z);
	} else if (i <= 1.56e-27) {
		tmp = (c * j) * a;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = (i * t) * b
    if (i <= (-1.6d+53)) then
        tmp = t_1
    else if (i <= (-7.5d-117)) then
        tmp = x * ((-1.0d0) * (a * t))
    else if (i <= (-8.5d-288)) then
        tmp = x * (y * z)
    else if (i <= 1.56d-27) then
        tmp = (c * j) * a
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (i * t) * b;
	double tmp;
	if (i <= -1.6e+53) {
		tmp = t_1;
	} else if (i <= -7.5e-117) {
		tmp = x * (-1.0 * (a * t));
	} else if (i <= -8.5e-288) {
		tmp = x * (y * z);
	} else if (i <= 1.56e-27) {
		tmp = (c * j) * a;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (i * t) * b
	tmp = 0
	if i <= -1.6e+53:
		tmp = t_1
	elif i <= -7.5e-117:
		tmp = x * (-1.0 * (a * t))
	elif i <= -8.5e-288:
		tmp = x * (y * z)
	elif i <= 1.56e-27:
		tmp = (c * j) * a
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(i * t) * b)
	tmp = 0.0
	if (i <= -1.6e+53)
		tmp = t_1;
	elseif (i <= -7.5e-117)
		tmp = Float64(x * Float64(-1.0 * Float64(a * t)));
	elseif (i <= -8.5e-288)
		tmp = Float64(x * Float64(y * z));
	elseif (i <= 1.56e-27)
		tmp = Float64(Float64(c * j) * a);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = (i * t) * b;
	tmp = 0.0;
	if (i <= -1.6e+53)
		tmp = t_1;
	elseif (i <= -7.5e-117)
		tmp = x * (-1.0 * (a * t));
	elseif (i <= -8.5e-288)
		tmp = x * (y * z);
	elseif (i <= 1.56e-27)
		tmp = (c * j) * a;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(i * t), $MachinePrecision] * b), $MachinePrecision]}, If[LessEqual[i, -1.6e+53], t$95$1, If[LessEqual[i, -7.5e-117], N[(x * N[(-1.0 * N[(a * t), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], If[LessEqual[i, -8.5e-288], N[(x * N[(y * z), $MachinePrecision]), $MachinePrecision], If[LessEqual[i, 1.56e-27], N[(N[(c * j), $MachinePrecision] * a), $MachinePrecision], t$95$1]]]]]

Derivation

1. Split input into 4 regimes

2. if i < -1.6e53 or 1.56e-27 < i

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   5. Taylor expanded in z around 0

      \[\leadsto \left(i \cdot t\right) \cdot b \]
   6. Step-by-step derivation

      1. lift-*.f64 22.4

         \[\leadsto \left(i \cdot t\right) \cdot b \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(i \cdot t\right) \cdot b \]

   if -1.6e53 < i < -7.50000000000000066e-117

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]

   8. Taylor expanded in y around 0

      \[\leadsto x \cdot \left(-1 \cdot \color{blue}{\left(a \cdot t\right)}\right) \]
   9. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \left(-1 \cdot \left(a \cdot \color{blue}{t}\right)\right) \]

      2. lift-*.f64 21.8

         \[\leadsto x \cdot \left(-1 \cdot \left(a \cdot t\right)\right) \]

   10. Applied rewrites 21.8%

       \[\leadsto x \cdot \left(-1 \cdot \color{blue}{\left(a \cdot t\right)}\right) \]

   if -7.50000000000000066e-117 < i < -8.4999999999999997e-288

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]

   8. Taylor expanded in y around inf

      \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]
   9. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot \color{blue}{z}\right) \]

      2. lift-*.f64 22.2

         \[\leadsto x \cdot \left(y \cdot z\right) \]

   10. Applied rewrites 22.2%

       \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]

   if -8.4999999999999997e-288 < i < 1.56e-27

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(c \cdot j\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 22.4

         \[\leadsto \left(c \cdot j\right) \cdot a \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(c \cdot j\right) \cdot a \]
3. Recombined 4 regimes into one program.
4. Add Preprocessing

Alternative 18: 29.7% accurate, 1.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(c \cdot j\right) \cdot a\\ \mathbf{if}\;c \leq -1.02 \cdot 10^{-65}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;c \leq 3 \cdot 10^{-249}:\\ \;\;\;\;\left(i \cdot t\right) \cdot b\\ \mathbf{elif}\;c \leq 4.2 \cdot 10^{+52}:\\ \;\;\;\;\left(-1 \cdot \left(t \cdot x\right)\right) \cdot a\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* c j) a)))
   (if (<= c -1.02e-65)
     t_1
     (if (<= c 3e-249)
       (* (* i t) b)
       (if (<= c 4.2e+52) (* (* -1.0 (* t x)) a) t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (c * j) * a;
	double tmp;
	if (c <= -1.02e-65) {
		tmp = t_1;
	} else if (c <= 3e-249) {
		tmp = (i * t) * b;
	} else if (c <= 4.2e+52) {
		tmp = (-1.0 * (t * x)) * a;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = (c * j) * a
    if (c <= (-1.02d-65)) then
        tmp = t_1
    else if (c <= 3d-249) then
        tmp = (i * t) * b
    else if (c <= 4.2d+52) then
        tmp = ((-1.0d0) * (t * x)) * a
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (c * j) * a;
	double tmp;
	if (c <= -1.02e-65) {
		tmp = t_1;
	} else if (c <= 3e-249) {
		tmp = (i * t) * b;
	} else if (c <= 4.2e+52) {
		tmp = (-1.0 * (t * x)) * a;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (c * j) * a
	tmp = 0
	if c <= -1.02e-65:
		tmp = t_1
	elif c <= 3e-249:
		tmp = (i * t) * b
	elif c <= 4.2e+52:
		tmp = (-1.0 * (t * x)) * a
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(c * j) * a)
	tmp = 0.0
	if (c <= -1.02e-65)
		tmp = t_1;
	elseif (c <= 3e-249)
		tmp = Float64(Float64(i * t) * b);
	elseif (c <= 4.2e+52)
		tmp = Float64(Float64(-1.0 * Float64(t * x)) * a);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = (c * j) * a;
	tmp = 0.0;
	if (c <= -1.02e-65)
		tmp = t_1;
	elseif (c <= 3e-249)
		tmp = (i * t) * b;
	elseif (c <= 4.2e+52)
		tmp = (-1.0 * (t * x)) * a;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(c * j), $MachinePrecision] * a), $MachinePrecision]}, If[LessEqual[c, -1.02e-65], t$95$1, If[LessEqual[c, 3e-249], N[(N[(i * t), $MachinePrecision] * b), $MachinePrecision], If[LessEqual[c, 4.2e+52], N[(N[(-1.0 * N[(t * x), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 3 regimes

2. if c < -1.02000000000000004e-65 or 4.2e52 < c

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(c \cdot j\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 22.4

         \[\leadsto \left(c \cdot j\right) \cdot a \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(c \cdot j\right) \cdot a \]

   if -1.02000000000000004e-65 < c < 3.00000000000000004e-249

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   5. Taylor expanded in z around 0

      \[\leadsto \left(i \cdot t\right) \cdot b \]
   6. Step-by-step derivation

      1. lift-*.f64 22.4

         \[\leadsto \left(i \cdot t\right) \cdot b \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(i \cdot t\right) \cdot b \]

   if 3.00000000000000004e-249 < c < 4.2e52

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around inf

      \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

      2. lower-*.f64 22.1

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]

   7. Applied rewrites 22.1%

      \[\leadsto \left(-1 \cdot \left(t \cdot x\right)\right) \cdot a \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 19: 29.5% accurate, 2.2× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(i \cdot t\right) \cdot b\\ \mathbf{if}\;i \leq -1.05 \cdot 10^{+52}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;i \leq -8.5 \cdot 10^{-288}:\\ \;\;\;\;x \cdot \left(y \cdot z\right)\\ \mathbf{elif}\;i \leq 1.56 \cdot 10^{-27}:\\ \;\;\;\;\left(c \cdot j\right) \cdot a\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* i t) b)))
   (if (<= i -1.05e+52)
     t_1
     (if (<= i -8.5e-288)
       (* x (* y z))
       (if (<= i 1.56e-27) (* (* c j) a) t_1)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (i * t) * b;
	double tmp;
	if (i <= -1.05e+52) {
		tmp = t_1;
	} else if (i <= -8.5e-288) {
		tmp = x * (y * z);
	} else if (i <= 1.56e-27) {
		tmp = (c * j) * a;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = (i * t) * b
    if (i <= (-1.05d+52)) then
        tmp = t_1
    else if (i <= (-8.5d-288)) then
        tmp = x * (y * z)
    else if (i <= 1.56d-27) then
        tmp = (c * j) * a
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = (i * t) * b;
	double tmp;
	if (i <= -1.05e+52) {
		tmp = t_1;
	} else if (i <= -8.5e-288) {
		tmp = x * (y * z);
	} else if (i <= 1.56e-27) {
		tmp = (c * j) * a;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (i * t) * b
	tmp = 0
	if i <= -1.05e+52:
		tmp = t_1
	elif i <= -8.5e-288:
		tmp = x * (y * z)
	elif i <= 1.56e-27:
		tmp = (c * j) * a
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(i * t) * b)
	tmp = 0.0
	if (i <= -1.05e+52)
		tmp = t_1;
	elseif (i <= -8.5e-288)
		tmp = Float64(x * Float64(y * z));
	elseif (i <= 1.56e-27)
		tmp = Float64(Float64(c * j) * a);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = (i * t) * b;
	tmp = 0.0;
	if (i <= -1.05e+52)
		tmp = t_1;
	elseif (i <= -8.5e-288)
		tmp = x * (y * z);
	elseif (i <= 1.56e-27)
		tmp = (c * j) * a;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(N[(i * t), $MachinePrecision] * b), $MachinePrecision]}, If[LessEqual[i, -1.05e+52], t$95$1, If[LessEqual[i, -8.5e-288], N[(x * N[(y * z), $MachinePrecision]), $MachinePrecision], If[LessEqual[i, 1.56e-27], N[(N[(c * j), $MachinePrecision] * a), $MachinePrecision], t$95$1]]]]

Derivation

1. Split input into 3 regimes

2. if i < -1.05e52 or 1.56e-27 < i

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in b around inf

      \[\leadsto \color{blue}{b \cdot \left(i \cdot t - c \cdot z\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot \color{blue}{b} \]

      3. lower--.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      4. lower-*.f64 N/A

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

      5. lift-*.f64 39.9

         \[\leadsto \left(i \cdot t - c \cdot z\right) \cdot b \]

   4. Applied rewrites 39.9%

      \[\leadsto \color{blue}{\left(i \cdot t - c \cdot z\right) \cdot b} \]

   5. Taylor expanded in z around 0

      \[\leadsto \left(i \cdot t\right) \cdot b \]
   6. Step-by-step derivation

      1. lift-*.f64 22.4

         \[\leadsto \left(i \cdot t\right) \cdot b \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(i \cdot t\right) \cdot b \]

   if -1.05e52 < i < -8.4999999999999997e-288

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]

   8. Taylor expanded in y around inf

      \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]
   9. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot \color{blue}{z}\right) \]

      2. lift-*.f64 22.2

         \[\leadsto x \cdot \left(y \cdot z\right) \]

   10. Applied rewrites 22.2%

       \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]

   if -8.4999999999999997e-288 < i < 1.56e-27

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(c \cdot j\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 22.4

         \[\leadsto \left(c \cdot j\right) \cdot a \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(c \cdot j\right) \cdot a \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 20: 29.0% accurate, 2.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := x \cdot \left(y \cdot z\right)\\ \mathbf{if}\;z \leq -1.4 \cdot 10^{-47}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;z \leq 5.6 \cdot 10^{+118}:\\ \;\;\;\;\left(c \cdot j\right) \cdot a\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* x (* y z))))
   (if (<= z -1.4e-47) t_1 (if (<= z 5.6e+118) (* (* c j) a) t_1))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = x * (y * z);
	double tmp;
	if (z <= -1.4e-47) {
		tmp = t_1;
	} else if (z <= 5.6e+118) {
		tmp = (c * j) * a;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    real(8) :: t_1
    real(8) :: tmp
    t_1 = x * (y * z)
    if (z <= (-1.4d-47)) then
        tmp = t_1
    else if (z <= 5.6d+118) then
        tmp = (c * j) * a
    else
        tmp = t_1
    end if
    code = tmp
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double t_1 = x * (y * z);
	double tmp;
	if (z <= -1.4e-47) {
		tmp = t_1;
	} else if (z <= 5.6e+118) {
		tmp = (c * j) * a;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = x * (y * z)
	tmp = 0
	if z <= -1.4e-47:
		tmp = t_1
	elif z <= 5.6e+118:
		tmp = (c * j) * a
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(x * Float64(y * z))
	tmp = 0.0
	if (z <= -1.4e-47)
		tmp = t_1;
	elseif (z <= 5.6e+118)
		tmp = Float64(Float64(c * j) * a);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = x * (y * z);
	tmp = 0.0;
	if (z <= -1.4e-47)
		tmp = t_1;
	elseif (z <= 5.6e+118)
		tmp = (c * j) * a;
	else
		tmp = t_1;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := Block[{t$95$1 = N[(x * N[(y * z), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[z, -1.4e-47], t$95$1, If[LessEqual[z, 5.6e+118], N[(N[(c * j), $MachinePrecision] * a), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if z < -1.39999999999999996e-47 or 5.59999999999999972e118 < z

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in j around inf

      \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   4. Applied rewrites 67.1%

      \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

   5. Taylor expanded in x around inf

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
   6. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

      2. lower--.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

      4. lift-*.f64 38.4

         \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

   7. Applied rewrites 38.4%

      \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]

   8. Taylor expanded in y around inf

      \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]
   9. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto x \cdot \left(y \cdot \color{blue}{z}\right) \]

      2. lift-*.f64 22.2

         \[\leadsto x \cdot \left(y \cdot z\right) \]

   10. Applied rewrites 22.2%

       \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]

   if -1.39999999999999996e-47 < z < 5.59999999999999972e118

   1. Initial program 73.2%

      \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

   2. Taylor expanded in a around inf

      \[\leadsto \color{blue}{a \cdot \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right)} \]
   3. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      2. lower-*.f64 N/A

         \[\leadsto \left(-1 \cdot \left(t \cdot x\right) + c \cdot j\right) \cdot \color{blue}{a} \]

      3. associate-*r* N/A

         \[\leadsto \left(\left(-1 \cdot t\right) \cdot x + c \cdot j\right) \cdot a \]

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(t\right)\right) \cdot x + c \cdot j\right) \cdot a \]

      5. lower-fma.f64 N/A

         \[\leadsto \mathsf{fma}\left(\mathsf{neg}\left(t\right), x, c \cdot j\right) \cdot a \]

      6. lower-neg.f64 N/A

         \[\leadsto \mathsf{fma}\left(-t, x, c \cdot j\right) \cdot a \]

      7. *-commutative N/A

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

      8. lower-*.f64 39.7

         \[\leadsto \mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a \]

   4. Applied rewrites 39.7%

      \[\leadsto \color{blue}{\mathsf{fma}\left(-t, x, j \cdot c\right) \cdot a} \]

   5. Taylor expanded in x around 0

      \[\leadsto \left(c \cdot j\right) \cdot a \]
   6. Step-by-step derivation

      1. lower-*.f64 22.4

         \[\leadsto \left(c \cdot j\right) \cdot a \]

   7. Applied rewrites 22.4%

      \[\leadsto \left(c \cdot j\right) \cdot a \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 21: 22.2% accurate, 5.9× speedup

Math

\[\begin{array}{l} \\ x \cdot \left(y \cdot z\right) \end{array} \]

FPCore

(FPCore (x y z t a b c i j) :precision binary64 (* x (* y z)))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	return x * (y * z);
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

    interface fmax
        module procedure fmax88
        module procedure fmax44
        module procedure fmax84
        module procedure fmax48
    end interface
    interface fmin
        module procedure fmin88
        module procedure fmin44
        module procedure fmin84
        module procedure fmin48
    end interface
contains
    real(8) function fmax88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(4) function fmax44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, max(x, y), y /= y), x /= x)
    end function
    real(8) function fmax84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmax48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
    end function
    real(8) function fmin88(x, y) result (res)
        real(8), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(4) function fmin44(x, y) result (res)
        real(4), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(y, merge(x, min(x, y), y /= y), x /= x)
    end function
    real(8) function fmin84(x, y) result(res)
        real(8), intent (in) :: x
        real(4), intent (in) :: y
        res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
    end function
    real(8) function fmin48(x, y) result(res)
        real(4), intent (in) :: x
        real(8), intent (in) :: y
        res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
    end function
end module

real(8) function code(x, y, z, t, a, b, c, i, j)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    real(8), intent (in) :: t
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8), intent (in) :: c
    real(8), intent (in) :: i
    real(8), intent (in) :: j
    code = x * (y * z)
end function

Java

public static double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	return x * (y * z);
}

Python

def code(x, y, z, t, a, b, c, i, j):
	return x * (y * z)

Julia

function code(x, y, z, t, a, b, c, i, j)
	return Float64(x * Float64(y * z))
end

MATLAB

function tmp = code(x, y, z, t, a, b, c, i, j)
	tmp = x * (y * z);
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := N[(x * N[(y * z), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 73.2%

   \[\left(x \cdot \left(y \cdot z - t \cdot a\right) - b \cdot \left(c \cdot z - t \cdot i\right)\right) + j \cdot \left(c \cdot a - y \cdot i\right) \]

2. Taylor expanded in j around inf

   \[\leadsto \color{blue}{j \cdot \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right)} \]
3. Step-by-step derivation

   1. *-commutative N/A

      \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

   2. lower-*.f64 N/A

      \[\leadsto \left(\left(a \cdot c + \frac{x \cdot \left(y \cdot z - a \cdot t\right)}{j}\right) - \left(i \cdot y + \frac{b \cdot \left(c \cdot z - i \cdot t\right)}{j}\right)\right) \cdot \color{blue}{j} \]

4. Applied rewrites 67.1%

   \[\leadsto \color{blue}{\left(\mathsf{fma}\left(c, a, x \cdot \frac{z \cdot y - a \cdot t}{j}\right) - \mathsf{fma}\left(i, y, b \cdot \frac{c \cdot z - i \cdot t}{j}\right)\right) \cdot j} \]

5. Taylor expanded in x around inf

   \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]
6. Step-by-step derivation

   1. lower-*.f64 N/A

      \[\leadsto x \cdot \color{blue}{\left(y \cdot z - a \cdot t\right)} \]

   2. lower--.f64 N/A

      \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a \cdot t}\right) \]

   3. lower-*.f64 N/A

      \[\leadsto x \cdot \left(y \cdot z - \color{blue}{a} \cdot t\right) \]

   4. lift-*.f64 38.4

      \[\leadsto x \cdot \left(y \cdot z - a \cdot \color{blue}{t}\right) \]

7. Applied rewrites 38.4%

   \[\leadsto \color{blue}{x \cdot \left(y \cdot z - a \cdot t\right)} \]

8. Taylor expanded in y around inf

   \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]
9. Step-by-step derivation

   1. lower-*.f64 N/A

      \[\leadsto x \cdot \left(y \cdot \color{blue}{z}\right) \]

   2. lift-*.f64 22.2

      \[\leadsto x \cdot \left(y \cdot z\right) \]

10. Applied rewrites 22.2%

    \[\leadsto x \cdot \color{blue}{\left(y \cdot z\right)} \]
11. Add Preprocessing

Reproduce

herbie shell --seed 2025139 
(FPCore (x y z t a b c i j)
  :name "Data.Colour.Matrix:determinant from colour-2.3.3, A"
  :precision binary64
  (+ (- (* x (- (* y z) (* t a))) (* b (- (* c z) (* t i)))) (* j (- (* c a) (* y i)))))