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

Percentage Accurate: 73.1% → 82.5%

Time: 26.4s

Alternatives: 18

Speedup: 0.5×

• 58.0% 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: 73.1% 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: 82.5% accurate, 0.5× speedup

Math

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

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
     (*
      (-
       (fma y x (/ (fma (- (* c a) (* i y)) j (* (fma (- x) a (* i b)) t)) z))
       (* c b))
      z))))

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 = (fma(y, x, (fma(((c * a) - (i * y)), j, (fma(-x, a, (i * b)) * t)) / z)) - (c * b)) * z;
	}
	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(fma(y, x, Float64(fma(Float64(Float64(c * a) - Float64(i * y)), j, Float64(fma(Float64(-x), a, Float64(i * b)) * t)) / z)) - Float64(c * b)) * z);
	end
	return 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[(N[(N[(y * x + N[(N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j + N[(N[((-x) * a + N[(i * b), $MachinePrecision]), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision] / z), $MachinePrecision]), $MachinePrecision] - N[(c * b), $MachinePrecision]), $MachinePrecision] * z), $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 91.2%

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

   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 0.0%

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

   2. Taylor expanded in z around inf

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

   4. Applied rewrites 45.7%

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

Alternative 2: 82.2% accurate, 0.5× speedup

Math

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

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
     (* (fma t (/ (- (* b i) (* a x)) j) (- (* a c) (* y i))) j))))

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 = fma(t, (((b * i) - (a * x)) / j), ((a * c) - (y * i))) * j;
	}
	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(fma(t, Float64(Float64(Float64(b * i) - Float64(a * x)) / j), Float64(Float64(a * c) - Float64(y * i))) * j);
	end
	return 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[(N[(t * N[(N[(N[(b * i), $MachinePrecision] - N[(a * x), $MachinePrecision]), $MachinePrecision] / j), $MachinePrecision] + N[(N[(a * c), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] * j), $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 91.2%

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

   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 0.0%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 43.5%

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

   5. Taylor expanded in y around inf

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

      1. associate-*r* N/A

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

      2. mul-1-neg N/A

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

      3. lower-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(i\right)\right) \cdot \left(j \cdot \color{blue}{y}\right) \]

      4. lower-neg.f64 N/A

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

      5. lower-*.f64 24.9

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

   7. Applied rewrites 24.9%

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

   8. Taylor expanded in j around inf

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

      1. *-commutative N/A

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

      2. +-commutative N/A

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

      3. *-lft-identity N/A

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

      4. metadata-eval N/A

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

      5. distribute-lft-neg-in N/A

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

      6. associate--l+ N/A

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

      7. *-commutative N/A

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

      8. distribute-lft-neg-in N/A

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

      9. metadata-eval N/A

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

      10. *-lft-identity N/A

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

      11. lower-*.f64 N/A

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

   10. Applied rewrites 47.4%

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

Alternative 3: 66.3% accurate, 1.4× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;z \leq 1.18 \cdot 10^{+85}:\\ \;\;\;\;\mathsf{fma}\left(c \cdot a - i \cdot y, j, \mathsf{fma}\left(-x, a, i \cdot b\right) \cdot t\right)\\ \mathbf{else}:\\ \;\;\;\;\left(\mathsf{fma}\left(y, x, \frac{\left(j \cdot c\right) \cdot a}{z}\right) - c \cdot b\right) \cdot z\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (if (<= z 1.18e+85)
   (fma (- (* c a) (* i y)) j (* (fma (- x) a (* i b)) t))
   (* (- (fma y x (/ (* (* j c) a) z)) (* c b)) 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 (z <= 1.18e+85) {
		tmp = fma(((c * a) - (i * y)), j, (fma(-x, a, (i * b)) * t));
	} else {
		tmp = (fma(y, x, (((j * c) * a) / z)) - (c * b)) * z;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0
	if (z <= 1.18e+85)
		tmp = fma(Float64(Float64(c * a) - Float64(i * y)), j, Float64(fma(Float64(-x), a, Float64(i * b)) * t));
	else
		tmp = Float64(Float64(fma(y, x, Float64(Float64(Float64(j * c) * a) / z)) - Float64(c * b)) * z);
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := If[LessEqual[z, 1.18e+85], N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j + N[(N[((-x) * a + N[(i * b), $MachinePrecision]), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision], N[(N[(N[(y * x + N[(N[(N[(j * c), $MachinePrecision] * a), $MachinePrecision] / z), $MachinePrecision]), $MachinePrecision] - N[(c * b), $MachinePrecision]), $MachinePrecision] * z), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if z < 1.17999999999999997e85

   1. Initial program 75.5%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 65.4%

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

   if 1.17999999999999997e85 < z

   1. Initial program 62.5%

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

   2. Taylor expanded in z around inf

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

   4. Applied rewrites 79.4%

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

   5. Taylor expanded in c around inf

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

      1. *-commutative N/A

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

      2. lower-*.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 70.1

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

   7. Applied rewrites 70.1%

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

Alternative 4: 63.3% accurate, 1.1× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;i \leq -4.5 \cdot 10^{+234}:\\ \;\;\;\;\mathsf{fma}\left(-j, y, b \cdot t\right) \cdot i\\ \mathbf{elif}\;i \leq -3.9 \cdot 10^{-262}:\\ \;\;\;\;\mathsf{fma}\left(j \cdot c - t \cdot x, a, \left(i \cdot t - c \cdot z\right) \cdot b\right)\\ \mathbf{elif}\;i \leq 7.2 \cdot 10^{+15}:\\ \;\;\;\;\mathsf{fma}\left(z \cdot y - a \cdot t, x, \left(j \cdot a - b \cdot z\right) \cdot c\right)\\ \mathbf{else}:\\ \;\;\;\;\left(y \cdot x\right) \cdot z + j \cdot \left(c \cdot a - y \cdot i\right)\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (if (<= i -4.5e+234)
   (* (fma (- j) y (* b t)) i)
   (if (<= i -3.9e-262)
     (fma (- (* j c) (* t x)) a (* (- (* i t) (* c z)) b))
     (if (<= i 7.2e+15)
       (fma (- (* z y) (* a t)) x (* (- (* j a) (* b z)) c))
       (+ (* (* y x) z) (* 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) {
	double tmp;
	if (i <= -4.5e+234) {
		tmp = fma(-j, y, (b * t)) * i;
	} else if (i <= -3.9e-262) {
		tmp = fma(((j * c) - (t * x)), a, (((i * t) - (c * z)) * b));
	} else if (i <= 7.2e+15) {
		tmp = fma(((z * y) - (a * t)), x, (((j * a) - (b * z)) * c));
	} else {
		tmp = ((y * x) * z) + (j * ((c * a) - (y * i)));
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0
	if (i <= -4.5e+234)
		tmp = Float64(fma(Float64(-j), y, Float64(b * t)) * i);
	elseif (i <= -3.9e-262)
		tmp = fma(Float64(Float64(j * c) - Float64(t * x)), a, Float64(Float64(Float64(i * t) - Float64(c * z)) * b));
	elseif (i <= 7.2e+15)
		tmp = fma(Float64(Float64(z * y) - Float64(a * t)), x, Float64(Float64(Float64(j * a) - Float64(b * z)) * c));
	else
		tmp = Float64(Float64(Float64(y * x) * z) + Float64(j * Float64(Float64(c * a) - Float64(y * i))));
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := If[LessEqual[i, -4.5e+234], N[(N[((-j) * y + N[(b * t), $MachinePrecision]), $MachinePrecision] * i), $MachinePrecision], If[LessEqual[i, -3.9e-262], N[(N[(N[(j * c), $MachinePrecision] - N[(t * x), $MachinePrecision]), $MachinePrecision] * a + N[(N[(N[(i * t), $MachinePrecision] - N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision]), $MachinePrecision], If[LessEqual[i, 7.2e+15], N[(N[(N[(z * y), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision] * x + N[(N[(N[(j * a), $MachinePrecision] - N[(b * z), $MachinePrecision]), $MachinePrecision] * c), $MachinePrecision]), $MachinePrecision], N[(N[(N[(y * x), $MachinePrecision] * z), $MachinePrecision] + N[(j * N[(N[(c * a), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]]]

Derivation

1. Split input into 4 regimes

2. if i < -4.49999999999999982e234

   1. Initial program 59.3%

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

   2. Taylor expanded in i around inf

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

      1. *-commutative N/A

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

      2. lower-*.f64 N/A

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

      3. fp-cancel-sub-sign-inv N/A

         \[\leadsto \left(-1 \cdot \left(j \cdot y\right) + \left(\mathsf{neg}\left(-1\right)\right) \cdot \left(b \cdot t\right)\right) \cdot i \]

      4. metadata-eval N/A

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

      5. *-lft-identity N/A

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

      6. mul-1-neg N/A

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

      7. distribute-lft-neg-out N/A

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

      8. lower-fma.f64 N/A

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

      9. lower-neg.f64 N/A

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

      10. lower-*.f64 76.3

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

   4. Applied rewrites 76.3%

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

   if -4.49999999999999982e234 < i < -3.89999999999999984e-262

   1. Initial program 75.2%

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

   2. Taylor expanded in y around 0

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

   4. Applied rewrites 59.8%

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

   if -3.89999999999999984e-262 < i < 7.2e15

   1. Initial program 79.8%

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

   2. Taylor expanded in i around 0

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

      1. +-commutative N/A

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

      2. associate--l+ N/A

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

      3. *-commutative N/A

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

      4. *-commutative N/A

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

      5. associate-*r* N/A

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

      6. *-commutative N/A

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

      7. associate-*l* N/A

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

      8. *-commutative N/A

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

      9. associate-*l* N/A

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

      10. *-commutative N/A

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

      11. distribute-lft-out-- N/A

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

      12. lower-fma.f64 N/A

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

   4. Applied rewrites 73.2%

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

   if 7.2e15 < i

   1. Initial program 64.9%

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

   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. associate-*r* N/A

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

      2. lower-*.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 47.6

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

   4. Applied rewrites 47.6%

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

Alternative 5: 62.7% accurate, 1.2× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;y \leq -1.12 \cdot 10^{+37}:\\ \;\;\;\;\left(y \cdot x\right) \cdot z + j \cdot \left(c \cdot a - y \cdot i\right)\\ \mathbf{elif}\;y \leq 1.3 \cdot 10^{+150}:\\ \;\;\;\;\mathsf{fma}\left(j \cdot c - t \cdot x, a, \left(i \cdot t - c \cdot z\right) \cdot b\right)\\ \mathbf{else}:\\ \;\;\;\;\left(z \cdot y - a \cdot t\right) \cdot x\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (if (<= y -1.12e+37)
   (+ (* (* y x) z) (* j (- (* c a) (* y i))))
   (if (<= y 1.3e+150)
     (fma (- (* j c) (* t x)) a (* (- (* i t) (* c z)) b))
     (* (- (* z y) (* a t)) x))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double tmp;
	if (y <= -1.12e+37) {
		tmp = ((y * x) * z) + (j * ((c * a) - (y * i)));
	} else if (y <= 1.3e+150) {
		tmp = fma(((j * c) - (t * x)), a, (((i * t) - (c * z)) * b));
	} else {
		tmp = ((z * y) - (a * t)) * x;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0
	if (y <= -1.12e+37)
		tmp = Float64(Float64(Float64(y * x) * z) + Float64(j * Float64(Float64(c * a) - Float64(y * i))));
	elseif (y <= 1.3e+150)
		tmp = fma(Float64(Float64(j * c) - Float64(t * x)), a, Float64(Float64(Float64(i * t) - Float64(c * z)) * b));
	else
		tmp = Float64(Float64(Float64(z * y) - Float64(a * t)) * x);
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := If[LessEqual[y, -1.12e+37], N[(N[(N[(y * x), $MachinePrecision] * z), $MachinePrecision] + N[(j * N[(N[(c * a), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], If[LessEqual[y, 1.3e+150], N[(N[(N[(j * c), $MachinePrecision] - N[(t * x), $MachinePrecision]), $MachinePrecision] * a + N[(N[(N[(i * t), $MachinePrecision] - N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision]), $MachinePrecision], N[(N[(N[(z * y), $MachinePrecision] - N[(a * t), $MachinePrecision]), $MachinePrecision] * x), $MachinePrecision]]]

Derivation

1. Split input into 3 regimes

2. if y < -1.12e37

   1. Initial program 64.2%

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

   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. associate-*r* N/A

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

      2. lower-*.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 59.0

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

   4. Applied rewrites 59.0%

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

   if -1.12e37 < y < 1.30000000000000003e150

   1. Initial program 78.7%

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

   2. Taylor expanded in y around 0

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

   4. Applied rewrites 67.6%

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

   if 1.30000000000000003e150 < y

   1. Initial program 59.8%

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

   2. Taylor expanded in x around inf

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

      1. *-commutative N/A

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

      2. *-commutative N/A

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

      3. lower-*.f64 N/A

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

      4. lift--.f64 N/A

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

      5. *-commutative N/A

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

      6. lower-*.f64 N/A

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

      7. *-commutative N/A

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

      8. lower-*.f64 48.2

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

   4. Applied rewrites 48.2%

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

Alternative 6: 61.8% accurate, 1.4× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;t \leq -4.5 \cdot 10^{+39}:\\ \;\;\;\;\mathsf{fma}\left(b \cdot i - a \cdot x, t, \left(c \cdot j\right) \cdot a\right)\\ \mathbf{elif}\;t \leq 1.35 \cdot 10^{-73}:\\ \;\;\;\;\left(y \cdot x\right) \cdot z + j \cdot \left(c \cdot a - y \cdot i\right)\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(j \cdot c, a, \left(i \cdot b - a \cdot x\right) \cdot t\right)\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (if (<= t -4.5e+39)
   (fma (- (* b i) (* a x)) t (* (* c j) a))
   (if (<= t 1.35e-73)
     (+ (* (* y x) z) (* j (- (* c a) (* y i))))
     (fma (* j c) a (* (- (* i b) (* a x)) t)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double tmp;
	if (t <= -4.5e+39) {
		tmp = fma(((b * i) - (a * x)), t, ((c * j) * a));
	} else if (t <= 1.35e-73) {
		tmp = ((y * x) * z) + (j * ((c * a) - (y * i)));
	} else {
		tmp = fma((j * c), a, (((i * b) - (a * x)) * t));
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0
	if (t <= -4.5e+39)
		tmp = fma(Float64(Float64(b * i) - Float64(a * x)), t, Float64(Float64(c * j) * a));
	elseif (t <= 1.35e-73)
		tmp = Float64(Float64(Float64(y * x) * z) + Float64(j * Float64(Float64(c * a) - Float64(y * i))));
	else
		tmp = fma(Float64(j * c), a, Float64(Float64(Float64(i * b) - Float64(a * x)) * t));
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := If[LessEqual[t, -4.5e+39], N[(N[(N[(b * i), $MachinePrecision] - N[(a * x), $MachinePrecision]), $MachinePrecision] * t + N[(N[(c * j), $MachinePrecision] * a), $MachinePrecision]), $MachinePrecision], If[LessEqual[t, 1.35e-73], N[(N[(N[(y * x), $MachinePrecision] * z), $MachinePrecision] + N[(j * N[(N[(c * a), $MachinePrecision] - N[(y * i), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(N[(j * c), $MachinePrecision] * a + N[(N[(N[(i * b), $MachinePrecision] - N[(a * x), $MachinePrecision]), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 3 regimes

2. if t < -4.49999999999999996e39

   1. Initial program 65.1%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 72.7%

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

   5. Taylor expanded in y around inf

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

      1. associate-*r* N/A

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

      2. mul-1-neg N/A

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

      3. lower-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(i\right)\right) \cdot \left(j \cdot \color{blue}{y}\right) \]

      4. lower-neg.f64 N/A

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

      5. lower-*.f64 16.7

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

   7. Applied rewrites 16.7%

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

   8. Taylor expanded in y around 0

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

      1. +-commutative N/A

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

      2. *-commutative N/A

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

      3. lower-fma.f64 N/A

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

      4. +-commutative N/A

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

      5. associate-*r* N/A

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

      6. mul-1-neg N/A

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

      7. fp-cancel-sub-sign N/A

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

      8. *-lft-identity N/A

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

      9. metadata-eval N/A

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

      10. distribute-lft-neg-in N/A

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

      11. lower--.f64 N/A

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

      12. lower-*.f64 N/A

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

      13. distribute-lft-neg-in N/A

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

      14. metadata-eval N/A

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

      15. *-lft-identity N/A

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

      16. lift-*.f64 N/A

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

      17. *-commutative N/A

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

      18. lower-*.f64 N/A

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

      19. lower-*.f64 69.3

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

   10. Applied rewrites 69.3%

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

   if -4.49999999999999996e39 < t < 1.34999999999999997e-73

   1. Initial program 79.9%

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

   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. associate-*r* N/A

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

      2. lower-*.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 58.7

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

   4. Applied rewrites 58.7%

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

   if 1.34999999999999997e-73 < t

   1. Initial program 68.2%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 69.5%

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

   5. Taylor expanded in y around 0

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

      1. *-commutative N/A

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

      2. lower-fma.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. *-commutative N/A

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

      6. mul-1-neg N/A

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

      7. *-commutative N/A

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

      8. distribute-lft-neg-out N/A

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

      9. *-commutative N/A

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

      10. lift-neg.f64 N/A

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

      11. lift-fma.f64 N/A

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

      12. lift-*.f64 N/A

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

      13. lift-*.f64 64.8

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

   7. Applied rewrites 64.3%

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

Alternative 7: 57.2% accurate, 1.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;t \leq -2.9 \cdot 10^{-17}:\\ \;\;\;\;\mathsf{fma}\left(b \cdot i - a \cdot x, t, \left(c \cdot j\right) \cdot a\right)\\ \mathbf{elif}\;t \leq 1.12 \cdot 10^{-33}:\\ \;\;\;\;\left(c \cdot a - i \cdot y\right) \cdot j\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(j \cdot c, a, \left(i \cdot b - a \cdot x\right) \cdot t\right)\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (if (<= t -2.9e-17)
   (fma (- (* b i) (* a x)) t (* (* c j) a))
   (if (<= t 1.12e-33)
     (* (- (* c a) (* i y)) j)
     (fma (* j c) a (* (- (* i b) (* a x)) t)))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double tmp;
	if (t <= -2.9e-17) {
		tmp = fma(((b * i) - (a * x)), t, ((c * j) * a));
	} else if (t <= 1.12e-33) {
		tmp = ((c * a) - (i * y)) * j;
	} else {
		tmp = fma((j * c), a, (((i * b) - (a * x)) * t));
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0
	if (t <= -2.9e-17)
		tmp = fma(Float64(Float64(b * i) - Float64(a * x)), t, Float64(Float64(c * j) * a));
	elseif (t <= 1.12e-33)
		tmp = Float64(Float64(Float64(c * a) - Float64(i * y)) * j);
	else
		tmp = fma(Float64(j * c), a, Float64(Float64(Float64(i * b) - Float64(a * x)) * t));
	end
	return tmp
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := If[LessEqual[t, -2.9e-17], N[(N[(N[(b * i), $MachinePrecision] - N[(a * x), $MachinePrecision]), $MachinePrecision] * t + N[(N[(c * j), $MachinePrecision] * a), $MachinePrecision]), $MachinePrecision], If[LessEqual[t, 1.12e-33], N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j), $MachinePrecision], N[(N[(j * c), $MachinePrecision] * a + N[(N[(N[(i * b), $MachinePrecision] - N[(a * x), $MachinePrecision]), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 3 regimes

2. if t < -2.9000000000000003e-17

   1. Initial program 66.8%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 70.4%

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

   5. Taylor expanded in y around inf

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

      1. associate-*r* N/A

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

      2. mul-1-neg N/A

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

      3. lower-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(i\right)\right) \cdot \left(j \cdot \color{blue}{y}\right) \]

      4. lower-neg.f64 N/A

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

      5. lower-*.f64 17.1

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

   7. Applied rewrites 17.1%

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

   8. Taylor expanded in y around 0

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

      1. +-commutative N/A

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

      2. *-commutative N/A

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

      3. lower-fma.f64 N/A

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

      4. +-commutative N/A

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

      5. associate-*r* N/A

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

      6. mul-1-neg N/A

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

      7. fp-cancel-sub-sign N/A

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

      8. *-lft-identity N/A

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

      9. metadata-eval N/A

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

      10. distribute-lft-neg-in N/A

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

      11. lower--.f64 N/A

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

      12. lower-*.f64 N/A

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

      13. distribute-lft-neg-in N/A

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

      14. metadata-eval N/A

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

      15. *-lft-identity N/A

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

      16. lift-*.f64 N/A

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

      17. *-commutative N/A

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

      18. lower-*.f64 N/A

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

      19. lower-*.f64 65.9

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

   10. Applied rewrites 65.9%

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

   if -2.9000000000000003e-17 < t < 1.11999999999999999e-33

   1. Initial program 80.3%

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

   2. Taylor expanded in j around inf

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

      1. *-commutative N/A

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

      2. *-commutative N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. lift--.f64 N/A

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

      6. lift-*.f64 N/A

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

      7. *-commutative N/A

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

      8. lower-*.f64 46.6

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

   4. Applied rewrites 46.6%

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

   if 1.11999999999999999e-33 < t

   1. Initial program 66.8%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 71.4%

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

   5. Taylor expanded in y around 0

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

      1. *-commutative N/A

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

      2. lower-fma.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. *-commutative N/A

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

      6. mul-1-neg N/A

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

      7. *-commutative N/A

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

      8. distribute-lft-neg-out N/A

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

      9. *-commutative N/A

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

      10. lift-neg.f64 N/A

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

      11. lift-fma.f64 N/A

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

      12. lift-*.f64 N/A

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

      13. lift-*.f64 67.4

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

   7. Applied rewrites 67.0%

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

Alternative 8: 57.2% accurate, 1.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \mathsf{fma}\left(j \cdot c, a, \left(i \cdot b - a \cdot x\right) \cdot t\right)\\ \mathbf{if}\;t \leq -2.9 \cdot 10^{-17}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;t \leq 1.12 \cdot 10^{-33}:\\ \;\;\;\;\left(c \cdot a - i \cdot y\right) \cdot j\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (fma (* j c) a (* (- (* i b) (* a x)) t))))
   (if (<= t -2.9e-17)
     t_1
     (if (<= t 1.12e-33) (* (- (* c a) (* 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 = fma((j * c), a, (((i * b) - (a * x)) * t));
	double tmp;
	if (t <= -2.9e-17) {
		tmp = t_1;
	} else if (t <= 1.12e-33) {
		tmp = ((c * a) - (i * y)) * j;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = fma(Float64(j * c), a, Float64(Float64(Float64(i * b) - Float64(a * x)) * t))
	tmp = 0.0
	if (t <= -2.9e-17)
		tmp = t_1;
	elseif (t <= 1.12e-33)
		tmp = Float64(Float64(Float64(c * a) - Float64(i * y)) * j);
	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[(j * c), $MachinePrecision] * a + N[(N[(N[(i * b), $MachinePrecision] - N[(a * x), $MachinePrecision]), $MachinePrecision] * t), $MachinePrecision]), $MachinePrecision]}, If[LessEqual[t, -2.9e-17], t$95$1, If[LessEqual[t, 1.12e-33], N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if t < -2.9000000000000003e-17 or 1.11999999999999999e-33 < t

   1. Initial program 66.8%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 70.9%

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

   5. Taylor expanded in y around 0

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

      1. *-commutative N/A

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

      2. lower-fma.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. *-commutative N/A

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

      6. mul-1-neg N/A

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

      7. *-commutative N/A

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

      8. distribute-lft-neg-out N/A

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

      9. *-commutative N/A

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

      10. lift-neg.f64 N/A

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

      11. lift-fma.f64 N/A

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

      12. lift-*.f64 N/A

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

      13. lift-*.f64 66.9

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

   7. Applied rewrites 66.5%

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

   if -2.9000000000000003e-17 < t < 1.11999999999999999e-33

   1. Initial program 80.3%

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

   2. Taylor expanded in j around inf

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

      1. *-commutative N/A

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

      2. *-commutative N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. lift--.f64 N/A

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

      6. lift-*.f64 N/A

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

      7. *-commutative N/A

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

      8. lower-*.f64 46.6

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

   4. Applied rewrites 46.6%

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

Alternative 9: 52.8% accurate, 1.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(j \cdot c - t \cdot x\right) \cdot a\\ \mathbf{if}\;a \leq -3.8 \cdot 10^{+129}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;a \leq 3.7 \cdot 10^{-258}:\\ \;\;\;\;\left(b \cdot t - y \cdot j\right) \cdot i\\ \mathbf{elif}\;a \leq 1.25 \cdot 10^{-48}:\\ \;\;\;\;\left(i \cdot t - c \cdot z\right) \cdot b\\ \mathbf{elif}\;a \leq 62:\\ \;\;\;\;\left(c \cdot a - i \cdot y\right) \cdot j\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- (* j c) (* t x)) a)))
   (if (<= a -3.8e+129)
     t_1
     (if (<= a 3.7e-258)
       (* (- (* b t) (* y j)) i)
       (if (<= a 1.25e-48)
         (* (- (* i t) (* c z)) b)
         (if (<= a 62.0) (* (- (* c a) (* 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 = ((j * c) - (t * x)) * a;
	double tmp;
	if (a <= -3.8e+129) {
		tmp = t_1;
	} else if (a <= 3.7e-258) {
		tmp = ((b * t) - (y * j)) * i;
	} else if (a <= 1.25e-48) {
		tmp = ((i * t) - (c * z)) * b;
	} else if (a <= 62.0) {
		tmp = ((c * a) - (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 = ((j * c) - (t * x)) * a
    if (a <= (-3.8d+129)) then
        tmp = t_1
    else if (a <= 3.7d-258) then
        tmp = ((b * t) - (y * j)) * i
    else if (a <= 1.25d-48) then
        tmp = ((i * t) - (c * z)) * b
    else if (a <= 62.0d0) then
        tmp = ((c * a) - (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 = ((j * c) - (t * x)) * a;
	double tmp;
	if (a <= -3.8e+129) {
		tmp = t_1;
	} else if (a <= 3.7e-258) {
		tmp = ((b * t) - (y * j)) * i;
	} else if (a <= 1.25e-48) {
		tmp = ((i * t) - (c * z)) * b;
	} else if (a <= 62.0) {
		tmp = ((c * a) - (i * y)) * j;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = ((j * c) - (t * x)) * a
	tmp = 0
	if a <= -3.8e+129:
		tmp = t_1
	elif a <= 3.7e-258:
		tmp = ((b * t) - (y * j)) * i
	elif a <= 1.25e-48:
		tmp = ((i * t) - (c * z)) * b
	elif a <= 62.0:
		tmp = ((c * a) - (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(j * c) - Float64(t * x)) * a)
	tmp = 0.0
	if (a <= -3.8e+129)
		tmp = t_1;
	elseif (a <= 3.7e-258)
		tmp = Float64(Float64(Float64(b * t) - Float64(y * j)) * i);
	elseif (a <= 1.25e-48)
		tmp = Float64(Float64(Float64(i * t) - Float64(c * z)) * b);
	elseif (a <= 62.0)
		tmp = Float64(Float64(Float64(c * a) - 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 = ((j * c) - (t * x)) * a;
	tmp = 0.0;
	if (a <= -3.8e+129)
		tmp = t_1;
	elseif (a <= 3.7e-258)
		tmp = ((b * t) - (y * j)) * i;
	elseif (a <= 1.25e-48)
		tmp = ((i * t) - (c * z)) * b;
	elseif (a <= 62.0)
		tmp = ((c * a) - (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[(j * c), $MachinePrecision] - N[(t * x), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision]}, If[LessEqual[a, -3.8e+129], t$95$1, If[LessEqual[a, 3.7e-258], N[(N[(N[(b * t), $MachinePrecision] - N[(y * j), $MachinePrecision]), $MachinePrecision] * i), $MachinePrecision], If[LessEqual[a, 1.25e-48], N[(N[(N[(i * t), $MachinePrecision] - N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision], If[LessEqual[a, 62.0], N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j), $MachinePrecision], t$95$1]]]]]

Derivation

1. Split input into 4 regimes

2. if a < -3.80000000000000005e129 or 62 < a

   1. Initial program 64.1%

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

   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. +-commutative N/A

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

      4. associate-*r* N/A

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

      5. mul-1-neg N/A

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

      6. fp-cancel-sub-sign N/A

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

      7. lower--.f64 N/A

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

      8. *-commutative N/A

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

      9. lower-*.f64 N/A

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

      10. lower-*.f64 62.5

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

   4. Applied rewrites 62.5%

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

   if -3.80000000000000005e129 < a < 3.7e-258

   1. Initial program 77.6%

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

   3. Applied rewrites 74.4%

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

   4. Taylor expanded in i around inf

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

      1. *-commutative N/A

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

      2. associate-*r* N/A

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

      3. mul-1-neg N/A

         \[\leadsto \left(-1 \cdot \left(j \cdot y\right) - \left(\mathsf{neg}\left(b\right)\right) \cdot t\right) \cdot i \]

      4. fp-cancel-sign-sub-inv N/A

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

      5. lower-*.f64 N/A

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

      6. +-commutative N/A

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

      7. associate-*r* N/A

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

      8. mul-1-neg N/A

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

      9. fp-cancel-sub-sign N/A

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

      10. lower--.f64 N/A

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

      11. lower-*.f64 N/A

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

      12. *-commutative N/A

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

      13. lower-*.f64 43.2

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

   6. Applied rewrites 43.2%

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

   if 3.7e-258 < a < 1.25e-48

   1. Initial program 81.8%

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

   2. Taylor expanded in b around inf

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

   4. Applied rewrites 46.0%

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

   if 1.25e-48 < a < 62

   1. Initial program 79.8%

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

   2. Taylor expanded in j around inf

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

      1. *-commutative N/A

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

      2. *-commutative N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. lift--.f64 N/A

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

      6. lift-*.f64 N/A

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

      7. *-commutative N/A

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

      8. lower-*.f64 31.3

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

   4. Applied rewrites 31.3%

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

Alternative 10: 50.8% accurate, 1.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(c \cdot a - i \cdot y\right) \cdot j\\ t_2 := \left(j \cdot c - t \cdot x\right) \cdot a\\ \mathbf{if}\;a \leq -6.2 \cdot 10^{+139}:\\ \;\;\;\;t\_2\\ \mathbf{elif}\;a \leq -2.2 \cdot 10^{-142}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;a \leq 1.25 \cdot 10^{-48}:\\ \;\;\;\;\left(i \cdot t - c \cdot z\right) \cdot b\\ \mathbf{elif}\;a \leq 62:\\ \;\;\;\;t\_1\\ \mathbf{else}:\\ \;\;\;\;t\_2\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- (* c a) (* i y)) j)) (t_2 (* (- (* j c) (* t x)) a)))
   (if (<= a -6.2e+139)
     t_2
     (if (<= a -2.2e-142)
       t_1
       (if (<= a 1.25e-48)
         (* (- (* i t) (* c z)) b)
         (if (<= a 62.0) t_1 t_2))))))

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 * a) - (i * y)) * j;
	double t_2 = ((j * c) - (t * x)) * a;
	double tmp;
	if (a <= -6.2e+139) {
		tmp = t_2;
	} else if (a <= -2.2e-142) {
		tmp = t_1;
	} else if (a <= 1.25e-48) {
		tmp = ((i * t) - (c * z)) * b;
	} else if (a <= 62.0) {
		tmp = t_1;
	} else {
		tmp = t_2;
	}
	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 = ((c * a) - (i * y)) * j
    t_2 = ((j * c) - (t * x)) * a
    if (a <= (-6.2d+139)) then
        tmp = t_2
    else if (a <= (-2.2d-142)) then
        tmp = t_1
    else if (a <= 1.25d-48) then
        tmp = ((i * t) - (c * z)) * b
    else if (a <= 62.0d0) then
        tmp = t_1
    else
        tmp = t_2
    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 * a) - (i * y)) * j;
	double t_2 = ((j * c) - (t * x)) * a;
	double tmp;
	if (a <= -6.2e+139) {
		tmp = t_2;
	} else if (a <= -2.2e-142) {
		tmp = t_1;
	} else if (a <= 1.25e-48) {
		tmp = ((i * t) - (c * z)) * b;
	} else if (a <= 62.0) {
		tmp = t_1;
	} else {
		tmp = t_2;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = ((c * a) - (i * y)) * j
	t_2 = ((j * c) - (t * x)) * a
	tmp = 0
	if a <= -6.2e+139:
		tmp = t_2
	elif a <= -2.2e-142:
		tmp = t_1
	elif a <= 1.25e-48:
		tmp = ((i * t) - (c * z)) * b
	elif a <= 62.0:
		tmp = t_1
	else:
		tmp = t_2
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(Float64(c * a) - Float64(i * y)) * j)
	t_2 = Float64(Float64(Float64(j * c) - Float64(t * x)) * a)
	tmp = 0.0
	if (a <= -6.2e+139)
		tmp = t_2;
	elseif (a <= -2.2e-142)
		tmp = t_1;
	elseif (a <= 1.25e-48)
		tmp = Float64(Float64(Float64(i * t) - Float64(c * z)) * b);
	elseif (a <= 62.0)
		tmp = t_1;
	else
		tmp = t_2;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = ((c * a) - (i * y)) * j;
	t_2 = ((j * c) - (t * x)) * a;
	tmp = 0.0;
	if (a <= -6.2e+139)
		tmp = t_2;
	elseif (a <= -2.2e-142)
		tmp = t_1;
	elseif (a <= 1.25e-48)
		tmp = ((i * t) - (c * z)) * b;
	elseif (a <= 62.0)
		tmp = t_1;
	else
		tmp = t_2;
	end
	tmp_2 = 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), $MachinePrecision]}, Block[{t$95$2 = N[(N[(N[(j * c), $MachinePrecision] - N[(t * x), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision]}, If[LessEqual[a, -6.2e+139], t$95$2, If[LessEqual[a, -2.2e-142], t$95$1, If[LessEqual[a, 1.25e-48], N[(N[(N[(i * t), $MachinePrecision] - N[(c * z), $MachinePrecision]), $MachinePrecision] * b), $MachinePrecision], If[LessEqual[a, 62.0], t$95$1, t$95$2]]]]]]

Derivation

1. Split input into 3 regimes

2. if a < -6.2e139 or 62 < a

   1. Initial program 63.9%

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

   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. +-commutative N/A

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

      4. associate-*r* N/A

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

      5. mul-1-neg N/A

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

      6. fp-cancel-sub-sign N/A

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

      7. lower--.f64 N/A

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

      8. *-commutative N/A

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

      9. lower-*.f64 N/A

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

      10. lower-*.f64 62.5

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

   4. Applied rewrites 62.5%

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

   if -6.2e139 < a < -2.20000000000000016e-142 or 1.25e-48 < a < 62

   1. Initial program 76.0%

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

   2. Taylor expanded in j around inf

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

      1. *-commutative N/A

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

      2. *-commutative N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. lift--.f64 N/A

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

      6. lift-*.f64 N/A

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

      7. *-commutative N/A

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

      8. lower-*.f64 38.0

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

   4. Applied rewrites 38.0%

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

   if -2.20000000000000016e-142 < a < 1.25e-48

   1. Initial program 80.9%

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

   2. Taylor expanded in b around inf

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

   4. Applied rewrites 46.5%

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

Alternative 11: 50.4% accurate, 2.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(i \cdot b - a \cdot x\right) \cdot t\\ \mathbf{if}\;t \leq -8.2 \cdot 10^{+41}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;t \leq 8.2 \cdot 10^{-33}:\\ \;\;\;\;\left(c \cdot a - i \cdot y\right) \cdot j\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (- (* i b) (* a x)) t)))
   (if (<= t -8.2e+41) t_1 (if (<= t 8.2e-33) (* (- (* c a) (* 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 = ((i * b) - (a * x)) * t;
	double tmp;
	if (t <= -8.2e+41) {
		tmp = t_1;
	} else if (t <= 8.2e-33) {
		tmp = ((c * a) - (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 = ((i * b) - (a * x)) * t
    if (t <= (-8.2d+41)) then
        tmp = t_1
    else if (t <= 8.2d-33) then
        tmp = ((c * a) - (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 = ((i * b) - (a * x)) * t;
	double tmp;
	if (t <= -8.2e+41) {
		tmp = t_1;
	} else if (t <= 8.2e-33) {
		tmp = ((c * a) - (i * y)) * j;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = ((i * b) - (a * x)) * t
	tmp = 0
	if t <= -8.2e+41:
		tmp = t_1
	elif t <= 8.2e-33:
		tmp = ((c * a) - (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(i * b) - Float64(a * x)) * t)
	tmp = 0.0
	if (t <= -8.2e+41)
		tmp = t_1;
	elseif (t <= 8.2e-33)
		tmp = Float64(Float64(Float64(c * a) - 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 = ((i * b) - (a * x)) * t;
	tmp = 0.0;
	if (t <= -8.2e+41)
		tmp = t_1;
	elseif (t <= 8.2e-33)
		tmp = ((c * a) - (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[(i * b), $MachinePrecision] - N[(a * x), $MachinePrecision]), $MachinePrecision] * t), $MachinePrecision]}, If[LessEqual[t, -8.2e+41], t$95$1, If[LessEqual[t, 8.2e-33], N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if t < -8.2000000000000007e41 or 8.2e-33 < t

   1. Initial program 66.1%

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

   2. Taylor expanded in t around inf

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

      1. *-commutative N/A

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

      2. distribute-lft-out-- N/A

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

      3. associate-*l* N/A

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

      4. *-commutative N/A

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

      5. *-commutative N/A

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

      6. associate-*l* N/A

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

      7. distribute-lft-out-- N/A

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

      8. lower-*.f64 N/A

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

   4. Applied rewrites 60.8%

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

      1. lift-*.f64 N/A

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

      2. *-commutative N/A

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

      3. lower-fma.f64 N/A

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

      4. +-commutative N/A

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

      5. lift-neg.f64 N/A

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

      6. fp-cancel-sub-sign N/A

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

      7. *-commutative N/A

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

      8. remove-double-neg N/A

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

      9. mul-1-neg N/A

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

      10. lower--.f64 N/A

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

      11. *-commutative N/A

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

      12. lift-*.f64 N/A

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

      13. mul-1-neg N/A

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

      14. remove-double-neg N/A

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

      15. lower-*.f64 60.5

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

   6. Applied rewrites 60.5%

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

   if -8.2000000000000007e41 < t < 8.2e-33

   1. Initial program 79.8%

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

   2. Taylor expanded in j around inf

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

      1. *-commutative N/A

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

      2. *-commutative N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. lift--.f64 N/A

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

      6. lift-*.f64 N/A

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

      7. *-commutative N/A

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

      8. lower-*.f64 45.5

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

   4. Applied rewrites 45.5%

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

Alternative 12: 41.7% accurate, 2.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;b \leq -4.6 \cdot 10^{+148}:\\ \;\;\;\;\left(\left(-c\right) \cdot b\right) \cdot z\\ \mathbf{elif}\;b \leq 5.8 \cdot 10^{+127}:\\ \;\;\;\;\left(c \cdot a - i \cdot y\right) \cdot j\\ \mathbf{else}:\\ \;\;\;\;\left(i \cdot t\right) \cdot b\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (if (<= b -4.6e+148)
   (* (* (- c) b) z)
   (if (<= b 5.8e+127) (* (- (* c a) (* i y)) j) (* (* i t) b))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double tmp;
	if (b <= -4.6e+148) {
		tmp = (-c * b) * z;
	} else if (b <= 5.8e+127) {
		tmp = ((c * a) - (i * y)) * j;
	} else {
		tmp = (i * t) * b;
	}
	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 (b <= (-4.6d+148)) then
        tmp = (-c * b) * z
    else if (b <= 5.8d+127) then
        tmp = ((c * a) - (i * y)) * j
    else
        tmp = (i * t) * b
    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 (b <= -4.6e+148) {
		tmp = (-c * b) * z;
	} else if (b <= 5.8e+127) {
		tmp = ((c * a) - (i * y)) * j;
	} else {
		tmp = (i * t) * b;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	tmp = 0
	if b <= -4.6e+148:
		tmp = (-c * b) * z
	elif b <= 5.8e+127:
		tmp = ((c * a) - (i * y)) * j
	else:
		tmp = (i * t) * b
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0
	if (b <= -4.6e+148)
		tmp = Float64(Float64(Float64(-c) * b) * z);
	elseif (b <= 5.8e+127)
		tmp = Float64(Float64(Float64(c * a) - Float64(i * y)) * j);
	else
		tmp = Float64(Float64(i * t) * b);
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0;
	if (b <= -4.6e+148)
		tmp = (-c * b) * z;
	elseif (b <= 5.8e+127)
		tmp = ((c * a) - (i * y)) * j;
	else
		tmp = (i * t) * b;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := If[LessEqual[b, -4.6e+148], N[(N[((-c) * b), $MachinePrecision] * z), $MachinePrecision], If[LessEqual[b, 5.8e+127], N[(N[(N[(c * a), $MachinePrecision] - N[(i * y), $MachinePrecision]), $MachinePrecision] * j), $MachinePrecision], N[(N[(i * t), $MachinePrecision] * b), $MachinePrecision]]]

Derivation

1. Split input into 3 regimes

2. if b < -4.6000000000000001e148

   1. Initial program 69.8%

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

   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 48.1

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

   4. Applied rewrites 48.1%

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

   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. mul-1-neg N/A

         \[\leadsto \left(\mathsf{neg}\left(b \cdot c\right)\right) \cdot z \]

      2. *-commutative N/A

         \[\leadsto \left(\mathsf{neg}\left(c \cdot b\right)\right) \cdot z \]

      3. distribute-lft-neg-out N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(c\right)\right) \cdot b\right) \cdot z \]

      4. lower-*.f64 N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(c\right)\right) \cdot b\right) \cdot z \]

      5. lower-neg.f64 40.2

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

   7. Applied rewrites 40.2%

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

   if -4.6000000000000001e148 < b < 5.8000000000000004e127

   1. Initial program 74.3%

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

   2. Taylor expanded in j around inf

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

      1. *-commutative N/A

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

      2. *-commutative N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. lift--.f64 N/A

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

      6. lift-*.f64 N/A

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

      7. *-commutative N/A

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

      8. lower-*.f64 42.5

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

   4. Applied rewrites 42.5%

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

   if 5.8000000000000004e127 < b

   1. Initial program 70.3%

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

   3. Applied rewrites 71.9%

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

   4. Taylor expanded in b around inf

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

      1. *-commutative N/A

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

      2. fp-cancel-sub-sign-inv N/A

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

      3. remove-double-neg N/A

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

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(-1 \cdot \left(i \cdot t\right)\right)\right) + \left(\mathsf{neg}\left(c\right)\right) \cdot z\right) \cdot b \]

      5. distribute-lft-neg-out N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(-1 \cdot \left(i \cdot t\right)\right)\right) + \left(\mathsf{neg}\left(c \cdot z\right)\right)\right) \cdot b \]

      6. distribute-neg-in N/A

         \[\leadsto \left(\mathsf{neg}\left(\left(-1 \cdot \left(i \cdot t\right) + c \cdot z\right)\right)\right) \cdot b \]

      7. +-commutative N/A

         \[\leadsto \left(\mathsf{neg}\left(\left(c \cdot z + -1 \cdot \left(i \cdot t\right)\right)\right)\right) \cdot b \]

      8. associate-*r* N/A

         \[\leadsto \left(\mathsf{neg}\left(\left(c \cdot z + \left(-1 \cdot i\right) \cdot t\right)\right)\right) \cdot b \]

      9. mul-1-neg N/A

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

      10. fp-cancel-sub-sign-inv N/A

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

      11. mul-1-neg N/A

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

      12. lower-*.f64 N/A

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

   6. Applied rewrites 70.0%

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

   7. Taylor expanded in z around 0

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

      1. lower-*.f64 38.8

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

   9. Applied rewrites 38.8%

      \[\leadsto \left(i \cdot t\right) \cdot b \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 13: 29.9% accurate, 2.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(a \cdot j\right) \cdot c\\ \mathbf{if}\;a \leq -7.5 \cdot 10^{+76}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;a \leq 9 \cdot 10^{-51}:\\ \;\;\;\;\left(i \cdot t\right) \cdot b\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* a j) c)))
   (if (<= a -7.5e+76) t_1 (if (<= a 9e-51) (* (* 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 = (a * j) * c;
	double tmp;
	if (a <= -7.5e+76) {
		tmp = t_1;
	} else if (a <= 9e-51) {
		tmp = (i * t) * 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 = (a * j) * c
    if (a <= (-7.5d+76)) then
        tmp = t_1
    else if (a <= 9d-51) then
        tmp = (i * t) * 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 = (a * j) * c;
	double tmp;
	if (a <= -7.5e+76) {
		tmp = t_1;
	} else if (a <= 9e-51) {
		tmp = (i * t) * b;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (a * j) * c
	tmp = 0
	if a <= -7.5e+76:
		tmp = t_1
	elif a <= 9e-51:
		tmp = (i * t) * b
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(a * j) * c)
	tmp = 0.0
	if (a <= -7.5e+76)
		tmp = t_1;
	elseif (a <= 9e-51)
		tmp = Float64(Float64(i * t) * 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 = (a * j) * c;
	tmp = 0.0;
	if (a <= -7.5e+76)
		tmp = t_1;
	elseif (a <= 9e-51)
		tmp = (i * t) * 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[(a * j), $MachinePrecision] * c), $MachinePrecision]}, If[LessEqual[a, -7.5e+76], t$95$1, If[LessEqual[a, 9e-51], N[(N[(i * t), $MachinePrecision] * b), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if a < -7.4999999999999995e76 or 8.99999999999999948e-51 < a

   1. Initial program 65.9%

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

   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 45.6

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

   4. Applied rewrites 45.6%

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

   5. Taylor expanded in z around 0

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

      1. lower-*.f64 34.4

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

   7. Applied rewrites 34.4%

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

   if -7.4999999999999995e76 < a < 8.99999999999999948e-51

   1. Initial program 79.7%

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

   3. Applied rewrites 76.1%

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

   4. Taylor expanded in b around inf

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

      1. *-commutative N/A

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

      2. fp-cancel-sub-sign-inv N/A

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

      3. remove-double-neg N/A

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

      4. mul-1-neg N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(-1 \cdot \left(i \cdot t\right)\right)\right) + \left(\mathsf{neg}\left(c\right)\right) \cdot z\right) \cdot b \]

      5. distribute-lft-neg-out N/A

         \[\leadsto \left(\left(\mathsf{neg}\left(-1 \cdot \left(i \cdot t\right)\right)\right) + \left(\mathsf{neg}\left(c \cdot z\right)\right)\right) \cdot b \]

      6. distribute-neg-in N/A

         \[\leadsto \left(\mathsf{neg}\left(\left(-1 \cdot \left(i \cdot t\right) + c \cdot z\right)\right)\right) \cdot b \]

      7. +-commutative N/A

         \[\leadsto \left(\mathsf{neg}\left(\left(c \cdot z + -1 \cdot \left(i \cdot t\right)\right)\right)\right) \cdot b \]

      8. associate-*r* N/A

         \[\leadsto \left(\mathsf{neg}\left(\left(c \cdot z + \left(-1 \cdot i\right) \cdot t\right)\right)\right) \cdot b \]

      9. mul-1-neg N/A

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

      10. fp-cancel-sub-sign-inv N/A

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

      11. mul-1-neg N/A

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

      12. lower-*.f64 N/A

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

   6. Applied rewrites 44.3%

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

   7. Taylor expanded in z around 0

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

      1. lower-*.f64 24.7

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

   9. Applied rewrites 24.7%

      \[\leadsto \left(i \cdot t\right) \cdot b \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 14: 29.4% accurate, 2.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(a \cdot j\right) \cdot c\\ \mathbf{if}\;a \leq -3.8 \cdot 10^{+47}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;a \leq 9 \cdot 10^{-51}:\\ \;\;\;\;\left(i \cdot b\right) \cdot t\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* a j) c)))
   (if (<= a -3.8e+47) t_1 (if (<= a 9e-51) (* (* i b) 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 * j) * c;
	double tmp;
	if (a <= -3.8e+47) {
		tmp = t_1;
	} else if (a <= 9e-51) {
		tmp = (i * b) * 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 * j) * c
    if (a <= (-3.8d+47)) then
        tmp = t_1
    else if (a <= 9d-51) then
        tmp = (i * b) * 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 * j) * c;
	double tmp;
	if (a <= -3.8e+47) {
		tmp = t_1;
	} else if (a <= 9e-51) {
		tmp = (i * b) * t;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (a * j) * c
	tmp = 0
	if a <= -3.8e+47:
		tmp = t_1
	elif a <= 9e-51:
		tmp = (i * b) * t
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(a * j) * c)
	tmp = 0.0
	if (a <= -3.8e+47)
		tmp = t_1;
	elseif (a <= 9e-51)
		tmp = Float64(Float64(i * b) * 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 * j) * c;
	tmp = 0.0;
	if (a <= -3.8e+47)
		tmp = t_1;
	elseif (a <= 9e-51)
		tmp = (i * b) * 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[(a * j), $MachinePrecision] * c), $MachinePrecision]}, If[LessEqual[a, -3.8e+47], t$95$1, If[LessEqual[a, 9e-51], N[(N[(i * b), $MachinePrecision] * t), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if a < -3.8000000000000003e47 or 8.99999999999999948e-51 < a

   1. Initial program 66.1%

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

   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 45.4

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

   4. Applied rewrites 45.4%

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

   5. Taylor expanded in z around 0

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

      1. lower-*.f64 34.1

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

   7. Applied rewrites 34.1%

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

   if -3.8000000000000003e47 < a < 8.99999999999999948e-51

   1. Initial program 80.2%

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

   2. Taylor expanded in t around inf

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

      1. *-commutative N/A

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

      2. distribute-lft-out-- N/A

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

      3. associate-*l* N/A

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

      4. *-commutative N/A

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

      5. *-commutative N/A

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

      6. associate-*l* N/A

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

      7. distribute-lft-out-- N/A

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

      8. lower-*.f64 N/A

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

   4. Applied rewrites 32.8%

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

   5. Taylor expanded in x around 0

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

      1. *-commutative N/A

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

      2. lift-*.f64 25.6

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

   7. Applied rewrites 25.6%

      \[\leadsto \left(i \cdot b\right) \cdot t \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 15: 29.1% accurate, 2.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;a \leq -1.95 \cdot 10^{+117}:\\ \;\;\;\;\left(c \cdot a\right) \cdot j\\ \mathbf{elif}\;a \leq 9 \cdot 10^{-51}:\\ \;\;\;\;\left(i \cdot b\right) \cdot t\\ \mathbf{else}:\\ \;\;\;\;\left(j \cdot c\right) \cdot a\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (if (<= a -1.95e+117)
   (* (* c a) j)
   (if (<= a 9e-51) (* (* i b) t) (* (* j c) a))))

C

double code(double x, double y, double z, double t, double a, double b, double c, double i, double j) {
	double tmp;
	if (a <= -1.95e+117) {
		tmp = (c * a) * j;
	} else if (a <= 9e-51) {
		tmp = (i * b) * t;
	} else {
		tmp = (j * c) * a;
	}
	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 (a <= (-1.95d+117)) then
        tmp = (c * a) * j
    else if (a <= 9d-51) then
        tmp = (i * b) * t
    else
        tmp = (j * c) * a
    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 (a <= -1.95e+117) {
		tmp = (c * a) * j;
	} else if (a <= 9e-51) {
		tmp = (i * b) * t;
	} else {
		tmp = (j * c) * a;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	tmp = 0
	if a <= -1.95e+117:
		tmp = (c * a) * j
	elif a <= 9e-51:
		tmp = (i * b) * t
	else:
		tmp = (j * c) * a
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0
	if (a <= -1.95e+117)
		tmp = Float64(Float64(c * a) * j);
	elseif (a <= 9e-51)
		tmp = Float64(Float64(i * b) * t);
	else
		tmp = Float64(Float64(j * c) * a);
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	tmp = 0.0;
	if (a <= -1.95e+117)
		tmp = (c * a) * j;
	elseif (a <= 9e-51)
		tmp = (i * b) * t;
	else
		tmp = (j * c) * a;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := If[LessEqual[a, -1.95e+117], N[(N[(c * a), $MachinePrecision] * j), $MachinePrecision], If[LessEqual[a, 9e-51], N[(N[(i * b), $MachinePrecision] * t), $MachinePrecision], N[(N[(j * c), $MachinePrecision] * a), $MachinePrecision]]]

Derivation

1. Split input into 3 regimes

2. if a < -1.94999999999999995e117

   1. Initial program 60.5%

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

   2. Taylor expanded in j around inf

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

      1. *-commutative N/A

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

      2. *-commutative N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 N/A

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

      5. lift--.f64 N/A

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

      6. lift-*.f64 N/A

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

      7. *-commutative N/A

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

      8. lower-*.f64 48.3

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

   4. Applied rewrites 48.3%

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

   5. Taylor expanded in y around 0

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

      1. *-commutative N/A

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

      2. lift-*.f64 41.8

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

   7. Applied rewrites 41.8%

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

   if -1.94999999999999995e117 < a < 8.99999999999999948e-51

   1. Initial program 79.1%

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

   2. Taylor expanded in t around inf

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

      1. *-commutative N/A

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

      2. distribute-lft-out-- N/A

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

      3. associate-*l* N/A

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

      4. *-commutative N/A

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

      5. *-commutative N/A

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

      6. associate-*l* N/A

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

      7. distribute-lft-out-- N/A

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

      8. lower-*.f64 N/A

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

   4. Applied rewrites 34.0%

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

   5. Taylor expanded in x around 0

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

      1. *-commutative N/A

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

      2. lift-*.f64 25.1

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

   7. Applied rewrites 25.1%

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

   if 8.99999999999999948e-51 < a

   1. Initial program 68.2%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 66.4%

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

   5. Taylor expanded in c around inf

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

      1. *-commutative N/A

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

      2. lower-*.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 29.7

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

   7. Applied rewrites 29.7%

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

Alternative 16: 29.0% accurate, 2.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(j \cdot c\right) \cdot a\\ \mathbf{if}\;a \leq -7.2 \cdot 10^{+116}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;a \leq 9 \cdot 10^{-51}:\\ \;\;\;\;\left(i \cdot b\right) \cdot t\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* j c) a)))
   (if (<= a -7.2e+116) t_1 (if (<= a 9e-51) (* (* i b) 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 = (j * c) * a;
	double tmp;
	if (a <= -7.2e+116) {
		tmp = t_1;
	} else if (a <= 9e-51) {
		tmp = (i * b) * 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 = (j * c) * a
    if (a <= (-7.2d+116)) then
        tmp = t_1
    else if (a <= 9d-51) then
        tmp = (i * b) * 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 = (j * c) * a;
	double tmp;
	if (a <= -7.2e+116) {
		tmp = t_1;
	} else if (a <= 9e-51) {
		tmp = (i * b) * t;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (j * c) * a
	tmp = 0
	if a <= -7.2e+116:
		tmp = t_1
	elif a <= 9e-51:
		tmp = (i * b) * t
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(j * c) * a)
	tmp = 0.0
	if (a <= -7.2e+116)
		tmp = t_1;
	elseif (a <= 9e-51)
		tmp = Float64(Float64(i * b) * 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 = (j * c) * a;
	tmp = 0.0;
	if (a <= -7.2e+116)
		tmp = t_1;
	elseif (a <= 9e-51)
		tmp = (i * b) * 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[(j * c), $MachinePrecision] * a), $MachinePrecision]}, If[LessEqual[a, -7.2e+116], t$95$1, If[LessEqual[a, 9e-51], N[(N[(i * b), $MachinePrecision] * t), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if a < -7.19999999999999941e116 or 8.99999999999999948e-51 < a

   1. Initial program 65.6%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 66.1%

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

   5. Taylor expanded in c around inf

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

      1. *-commutative N/A

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

      2. lower-*.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 33.5

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

   7. Applied rewrites 33.5%

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

   if -7.19999999999999941e116 < a < 8.99999999999999948e-51

   1. Initial program 79.1%

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

   2. Taylor expanded in t around inf

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

      1. *-commutative N/A

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

      2. distribute-lft-out-- N/A

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

      3. associate-*l* N/A

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

      4. *-commutative N/A

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

      5. *-commutative N/A

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

      6. associate-*l* N/A

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

      7. distribute-lft-out-- N/A

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

      8. lower-*.f64 N/A

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

   4. Applied rewrites 34.0%

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

   5. Taylor expanded in x around 0

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

      1. *-commutative N/A

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

      2. lift-*.f64 25.1

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

   7. Applied rewrites 25.1%

      \[\leadsto \left(i \cdot b\right) \cdot t \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 17: 28.9% accurate, 2.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_1 := \left(j \cdot c\right) \cdot a\\ \mathbf{if}\;a \leq -5.8 \cdot 10^{+50}:\\ \;\;\;\;t\_1\\ \mathbf{elif}\;a \leq 8.6 \cdot 10^{-50}:\\ \;\;\;\;\left(z \cdot y\right) \cdot x\\ \mathbf{else}:\\ \;\;\;\;t\_1\\ \end{array} \end{array} \]

FPCore

(FPCore (x y z t a b c i j)
 :precision binary64
 (let* ((t_1 (* (* j c) a)))
   (if (<= a -5.8e+50) t_1 (if (<= a 8.6e-50) (* (* z y) 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 = (j * c) * a;
	double tmp;
	if (a <= -5.8e+50) {
		tmp = t_1;
	} else if (a <= 8.6e-50) {
		tmp = (z * y) * x;
	} 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 = (j * c) * a
    if (a <= (-5.8d+50)) then
        tmp = t_1
    else if (a <= 8.6d-50) then
        tmp = (z * y) * x
    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 = (j * c) * a;
	double tmp;
	if (a <= -5.8e+50) {
		tmp = t_1;
	} else if (a <= 8.6e-50) {
		tmp = (z * y) * x;
	} else {
		tmp = t_1;
	}
	return tmp;
}

Python

def code(x, y, z, t, a, b, c, i, j):
	t_1 = (j * c) * a
	tmp = 0
	if a <= -5.8e+50:
		tmp = t_1
	elif a <= 8.6e-50:
		tmp = (z * y) * x
	else:
		tmp = t_1
	return tmp

Julia

function code(x, y, z, t, a, b, c, i, j)
	t_1 = Float64(Float64(j * c) * a)
	tmp = 0.0
	if (a <= -5.8e+50)
		tmp = t_1;
	elseif (a <= 8.6e-50)
		tmp = Float64(Float64(z * y) * x);
	else
		tmp = t_1;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y, z, t, a, b, c, i, j)
	t_1 = (j * c) * a;
	tmp = 0.0;
	if (a <= -5.8e+50)
		tmp = t_1;
	elseif (a <= 8.6e-50)
		tmp = (z * y) * x;
	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[(j * c), $MachinePrecision] * a), $MachinePrecision]}, If[LessEqual[a, -5.8e+50], t$95$1, If[LessEqual[a, 8.6e-50], N[(N[(z * y), $MachinePrecision] * x), $MachinePrecision], t$95$1]]]

Derivation

1. Split input into 2 regimes

2. if a < -5.8e50 or 8.59999999999999995e-50 < a

   1. Initial program 66.0%

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

   2. Taylor expanded in z around 0

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

   4. Applied rewrites 66.5%

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

   5. Taylor expanded in c around inf

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

      1. *-commutative N/A

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

      2. lower-*.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 32.7

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

   7. Applied rewrites 32.7%

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

   if -5.8e50 < a < 8.59999999999999995e-50

   1. Initial program 80.2%

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

   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 45.2

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

   4. Applied rewrites 45.2%

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

   5. Taylor expanded in x around inf

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

      1. *-commutative N/A

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

      2. lower-*.f64 N/A

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

      3. *-commutative N/A

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

      4. lower-*.f64 25.4

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

   7. Applied rewrites 25.4%

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

Alternative 18: 22.1% accurate, 5.9× speedup

Math

\[\begin{array}{l} \\ \left(z \cdot y\right) \cdot x \end{array} \]

FPCore

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

C

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

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 = (z * y) * x
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 (z * y) * x;
}

Python

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

Julia

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

MATLAB

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

Wolfram

code[x_, y_, z_, t_, a_, b_, c_, i_, j_] := N[(N[(z * y), $MachinePrecision] * x), $MachinePrecision]

Derivation

1. Initial program 73.1%

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

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 39.3

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

4. Applied rewrites 39.3%

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

5. Taylor expanded in x around inf

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

   1. *-commutative N/A

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

   2. lower-*.f64 N/A

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

   3. *-commutative N/A

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

   4. lower-*.f64 22.1

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

7. Applied rewrites 22.1%

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

Reproduce

herbie shell --seed 2025130 
(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)))))