Diagrams.TwoD.Apollonian:initialConfig from diagrams-contrib-1.3.0.5, A

Percentage Accurate: 68.0% → 97.9%

Time: 3.6s

Alternatives: 9

Speedup: 1.1×

• 41.0% of points produce a very large (infinite) output. You may want to add a precondition.

Specification

Math

\[\begin{array}{l} \\ \frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \end{array} \]

FPCore

(FPCore (x y z)
 :precision binary64
 (/ (- (+ (* x x) (* y y)) (* z z)) (* y 2.0)))

C

double code(double x, double y, double z) {
	return (((x * x) + (y * y)) - (z * z)) / (y * 2.0);
}

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)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    code = (((x * x) + (y * y)) - (z * z)) / (y * 2.0d0)
end function

Java

public static double code(double x, double y, double z) {
	return (((x * x) + (y * y)) - (z * z)) / (y * 2.0);
}

Python

def code(x, y, z):
	return (((x * x) + (y * y)) - (z * z)) / (y * 2.0)

Julia

function code(x, y, z)
	return Float64(Float64(Float64(Float64(x * x) + Float64(y * y)) - Float64(z * z)) / Float64(y * 2.0))
end

MATLAB

function tmp = code(x, y, z)
	tmp = (((x * x) + (y * y)) - (z * z)) / (y * 2.0);
end

Wolfram

code[x_, y_, z_] := N[(N[(N[(N[(x * x), $MachinePrecision] + N[(y * y), $MachinePrecision]), $MachinePrecision] - N[(z * z), $MachinePrecision]), $MachinePrecision] / N[(y * 2.0), $MachinePrecision]), $MachinePrecision]

Initial Program: 68.0% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \end{array} \]

FPCore

(FPCore (x y z)
 :precision binary64
 (/ (- (+ (* x x) (* y y)) (* z z)) (* y 2.0)))

C

double code(double x, double y, double z) {
	return (((x * x) + (y * y)) - (z * z)) / (y * 2.0);
}

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)
use fmin_fmax_functions
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8), intent (in) :: z
    code = (((x * x) + (y * y)) - (z * z)) / (y * 2.0d0)
end function

Java

public static double code(double x, double y, double z) {
	return (((x * x) + (y * y)) - (z * z)) / (y * 2.0);
}

Python

def code(x, y, z):
	return (((x * x) + (y * y)) - (z * z)) / (y * 2.0)

Julia

function code(x, y, z)
	return Float64(Float64(Float64(Float64(x * x) + Float64(y * y)) - Float64(z * z)) / Float64(y * 2.0))
end

MATLAB

function tmp = code(x, y, z)
	tmp = (((x * x) + (y * y)) - (z * z)) / (y * 2.0);
end

Wolfram

code[x_, y_, z_] := N[(N[(N[(N[(x * x), $MachinePrecision] + N[(y * y), $MachinePrecision]), $MachinePrecision] - N[(z * z), $MachinePrecision]), $MachinePrecision] / N[(y * 2.0), $MachinePrecision]), $MachinePrecision]

Alternative 1: 97.9% accurate, 0.8× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \begin{array}{l} \mathbf{if}\;y\_m \leq 1.49 \cdot 10^{+61}:\\ \;\;\;\;\left(y\_m + \frac{\left(x + z\right) \cdot \left(x - z\right)}{y\_m}\right) \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(\mathsf{fma}\left(\frac{x}{y\_m}, x, y\_m\right), 0.5, \left(z \cdot -0.5\right) \cdot \frac{z}{y\_m}\right)\\ \end{array} \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z)
 :precision binary64
 (*
  y_s
  (if (<= y_m 1.49e+61)
    (* (+ y_m (/ (* (+ x z) (- x z)) y_m)) 0.5)
    (fma (fma (/ x y_m) x y_m) 0.5 (* (* z -0.5) (/ z y_m))))))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if (y_m <= 1.49e+61) {
		tmp = (y_m + (((x + z) * (x - z)) / y_m)) * 0.5;
	} else {
		tmp = fma(fma((x / y_m), x, y_m), 0.5, ((z * -0.5) * (z / y_m)));
	}
	return y_s * tmp;
}

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	tmp = 0.0
	if (y_m <= 1.49e+61)
		tmp = Float64(Float64(y_m + Float64(Float64(Float64(x + z) * Float64(x - z)) / y_m)) * 0.5);
	else
		tmp = fma(fma(Float64(x / y_m), x, y_m), 0.5, Float64(Float64(z * -0.5) * Float64(z / y_m)));
	end
	return Float64(y_s * tmp)
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * If[LessEqual[y$95$m, 1.49e+61], N[(N[(y$95$m + N[(N[(N[(x + z), $MachinePrecision] * N[(x - z), $MachinePrecision]), $MachinePrecision] / y$95$m), $MachinePrecision]), $MachinePrecision] * 0.5), $MachinePrecision], N[(N[(N[(x / y$95$m), $MachinePrecision] * x + y$95$m), $MachinePrecision] * 0.5 + N[(N[(z * -0.5), $MachinePrecision] * N[(z / y$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if y < 1.4900000000000001e61

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

      1. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2}} \]

      2. lift-*.f64 N/A

         \[\leadsto \frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{\color{blue}{y \cdot 2}} \]

      3. associate-/r* N/A

         \[\leadsto \color{blue}{\frac{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y}}{2}} \]

      4. mult-flip N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y} \cdot \frac{1}{2}} \]

      5. lower-*.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y} \cdot \frac{1}{2}} \]

   3. Applied rewrites 89.1%

      \[\leadsto \color{blue}{\left(y + \frac{\left(x + z\right) \cdot \left(x - z\right)}{y}\right) \cdot 0.5} \]

   if 1.4900000000000001e61 < y

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

   3. Applied rewrites 73.7%

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

   5. Applied rewrites 91.4%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\mathsf{fma}\left(\frac{x}{y}, x, y\right), 0.5, \left(z \cdot -0.5\right) \cdot \frac{z}{y}\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 2: 95.6% accurate, 0.7× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \begin{array}{l} \mathbf{if}\;y\_m \leq 2.9 \cdot 10^{-31}:\\ \;\;\;\;\left(y\_m + \frac{\left(x + z\right) \cdot \left(x - z\right)}{y\_m}\right) \cdot 0.5\\ \mathbf{elif}\;y\_m \leq 8 \cdot 10^{+150}:\\ \;\;\;\;\mathsf{fma}\left(\left(x - z\right) \cdot \frac{z + x}{y\_m \cdot y\_m}, 0.5, 0.5\right) \cdot y\_m\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(\frac{z}{y\_m} \cdot z, -0.5, y\_m \cdot 0.5\right)\\ \end{array} \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z)
 :precision binary64
 (*
  y_s
  (if (<= y_m 2.9e-31)
    (* (+ y_m (/ (* (+ x z) (- x z)) y_m)) 0.5)
    (if (<= y_m 8e+150)
      (* (fma (* (- x z) (/ (+ z x) (* y_m y_m))) 0.5 0.5) y_m)
      (fma (* (/ z y_m) z) -0.5 (* y_m 0.5))))))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if (y_m <= 2.9e-31) {
		tmp = (y_m + (((x + z) * (x - z)) / y_m)) * 0.5;
	} else if (y_m <= 8e+150) {
		tmp = fma(((x - z) * ((z + x) / (y_m * y_m))), 0.5, 0.5) * y_m;
	} else {
		tmp = fma(((z / y_m) * z), -0.5, (y_m * 0.5));
	}
	return y_s * tmp;
}

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	tmp = 0.0
	if (y_m <= 2.9e-31)
		tmp = Float64(Float64(y_m + Float64(Float64(Float64(x + z) * Float64(x - z)) / y_m)) * 0.5);
	elseif (y_m <= 8e+150)
		tmp = Float64(fma(Float64(Float64(x - z) * Float64(Float64(z + x) / Float64(y_m * y_m))), 0.5, 0.5) * y_m);
	else
		tmp = fma(Float64(Float64(z / y_m) * z), -0.5, Float64(y_m * 0.5));
	end
	return Float64(y_s * tmp)
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * If[LessEqual[y$95$m, 2.9e-31], N[(N[(y$95$m + N[(N[(N[(x + z), $MachinePrecision] * N[(x - z), $MachinePrecision]), $MachinePrecision] / y$95$m), $MachinePrecision]), $MachinePrecision] * 0.5), $MachinePrecision], If[LessEqual[y$95$m, 8e+150], N[(N[(N[(N[(x - z), $MachinePrecision] * N[(N[(z + x), $MachinePrecision] / N[(y$95$m * y$95$m), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] * 0.5 + 0.5), $MachinePrecision] * y$95$m), $MachinePrecision], N[(N[(N[(z / y$95$m), $MachinePrecision] * z), $MachinePrecision] * -0.5 + N[(y$95$m * 0.5), $MachinePrecision]), $MachinePrecision]]]), $MachinePrecision]

Derivation

1. Split input into 3 regimes

2. if y < 2.9000000000000001e-31

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

      1. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2}} \]

      2. lift-*.f64 N/A

         \[\leadsto \frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{\color{blue}{y \cdot 2}} \]

      3. associate-/r* N/A

         \[\leadsto \color{blue}{\frac{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y}}{2}} \]

      4. mult-flip N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y} \cdot \frac{1}{2}} \]

      5. lower-*.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y} \cdot \frac{1}{2}} \]

   3. Applied rewrites 89.1%

      \[\leadsto \color{blue}{\left(y + \frac{\left(x + z\right) \cdot \left(x - z\right)}{y}\right) \cdot 0.5} \]

   if 2.9000000000000001e-31 < y < 7.99999999999999985e150

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{y \cdot \left(\frac{1}{2} + \frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}\right)} \]
   3. Step-by-step derivation

      1. lower-*.f64 N/A

         \[\leadsto y \cdot \color{blue}{\left(\frac{1}{2} + \frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}\right)} \]

      2. lower-+.f64 N/A

         \[\leadsto y \cdot \left(\frac{1}{2} + \color{blue}{\frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}}\right) \]

      3. lower-*.f64 N/A

         \[\leadsto y \cdot \left(\frac{1}{2} + \frac{1}{2} \cdot \color{blue}{\frac{{x}^{2} - {z}^{2}}{{y}^{2}}}\right) \]

      4. lower-/.f64 N/A

         \[\leadsto y \cdot \left(\frac{1}{2} + \frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{\color{blue}{{y}^{2}}}\right) \]

      5. lower--.f64 N/A

         \[\leadsto y \cdot \left(\frac{1}{2} + \frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{{\color{blue}{y}}^{2}}\right) \]

      6. lower-pow.f64 N/A

         \[\leadsto y \cdot \left(\frac{1}{2} + \frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}\right) \]

      7. lower-pow.f64 N/A

         \[\leadsto y \cdot \left(\frac{1}{2} + \frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}\right) \]

      8. lower-pow.f64 71.6

         \[\leadsto y \cdot \left(0.5 + 0.5 \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{\color{blue}{2}}}\right) \]

   4. Applied rewrites 71.6%

      \[\leadsto \color{blue}{y \cdot \left(0.5 + 0.5 \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}\right)} \]
   5. Step-by-step derivation

      1. lift-*.f64 N/A

         \[\leadsto y \cdot \color{blue}{\left(\frac{1}{2} + \frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}\right)} \]

      2. *-commutative N/A

         \[\leadsto \left(\frac{1}{2} + \frac{1}{2} \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}\right) \cdot \color{blue}{y} \]

      3. lower-*.f64 71.6

         \[\leadsto \left(0.5 + 0.5 \cdot \frac{{x}^{2} - {z}^{2}}{{y}^{2}}\right) \cdot \color{blue}{y} \]

   6. Applied rewrites 84.1%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\left(x - z\right) \cdot \frac{z + x}{y \cdot y}, 0.5, 0.5\right) \cdot y} \]

   if 7.99999999999999985e150 < y

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

      1. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2}} \]

      2. frac-2neg N/A

         \[\leadsto \color{blue}{\frac{\mathsf{neg}\left(\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      3. lift--.f64 N/A

         \[\leadsto \frac{\mathsf{neg}\left(\color{blue}{\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)}\right)}{\mathsf{neg}\left(y \cdot 2\right)} \]

      4. sub-negate-rev N/A

         \[\leadsto \frac{\color{blue}{z \cdot z - \left(x \cdot x + y \cdot y\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      5. sub-flip N/A

         \[\leadsto \frac{\color{blue}{z \cdot z + \left(\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      6. div-add N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \frac{\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      7. frac-2neg N/A

         \[\leadsto \frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \color{blue}{\frac{x \cdot x + y \cdot y}{y \cdot 2}} \]

      8. distribute-neg-frac2 N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\frac{z \cdot z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      9. lift-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(\frac{\color{blue}{z \cdot z}}{y \cdot 2}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      10. associate-/l* N/A

          \[\leadsto \left(\mathsf{neg}\left(\color{blue}{z \cdot \frac{z}{y \cdot 2}}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      11. distribute-rgt-neg-in N/A

          \[\leadsto \color{blue}{z \cdot \left(\mathsf{neg}\left(\frac{z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      12. distribute-frac-neg N/A

          \[\leadsto z \cdot \color{blue}{\frac{\mathsf{neg}\left(z\right)}{y \cdot 2}} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      13. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{\mathsf{neg}\left(z\right)}{y \cdot 2}, \frac{x \cdot x + y \cdot y}{y \cdot 2}\right)} \]

   3. Applied rewrites 87.2%

      \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \left(y + \frac{x \cdot x}{y}\right) \cdot 0.5\right)} \]

   4. Taylor expanded in x around 0

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot \frac{1}{2}\right) \]
   5. Step-by-step derivation

   6. Applied rewrites 68.0%

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot 0.5\right) \]
   7. Step-by-step derivation

      1. lift-fma.f64 N/A

         \[\leadsto \color{blue}{z \cdot \frac{z}{-2 \cdot y} + y \cdot \frac{1}{2}} \]

      2. lift-/.f64 N/A

         \[\leadsto z \cdot \color{blue}{\frac{z}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      3. associate-*r/ N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      4. lift-*.f64 N/A

         \[\leadsto \frac{z \cdot z}{\color{blue}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      5. *-commutative N/A

         \[\leadsto \frac{z \cdot z}{\color{blue}{y \cdot -2}} + y \cdot \frac{1}{2} \]

      6. times-frac N/A

         \[\leadsto \color{blue}{\frac{z}{y} \cdot \frac{z}{-2}} + y \cdot \frac{1}{2} \]

      7. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{z}{y}} \cdot \frac{z}{-2} + y \cdot \frac{1}{2} \]

      8. mult-flip N/A

         \[\leadsto \frac{z}{y} \cdot \color{blue}{\left(z \cdot \frac{1}{-2}\right)} + y \cdot \frac{1}{2} \]

      9. metadata-eval N/A

         \[\leadsto \frac{z}{y} \cdot \left(z \cdot \color{blue}{\frac{-1}{2}}\right) + y \cdot \frac{1}{2} \]

      10. associate-*r* N/A

          \[\leadsto \color{blue}{\left(\frac{z}{y} \cdot z\right) \cdot \frac{-1}{2}} + y \cdot \frac{1}{2} \]

      11. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{z}{y} \cdot z, \frac{-1}{2}, y \cdot \frac{1}{2}\right)} \]

      12. lower-*.f64 68.0

          \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{z}{y} \cdot z}, -0.5, y \cdot 0.5\right) \]

   8. Applied rewrites 68.0%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{z}{y} \cdot z, -0.5, y \cdot 0.5\right)} \]
3. Recombined 3 regimes into one program.
4. Add Preprocessing

Alternative 3: 93.5% accurate, 1.0× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \begin{array}{l} \mathbf{if}\;y\_m \leq 5.8 \cdot 10^{+127}:\\ \;\;\;\;\left(y\_m + \frac{\left(x + z\right) \cdot \left(x - z\right)}{y\_m}\right) \cdot 0.5\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(\frac{z}{y\_m} \cdot z, -0.5, y\_m \cdot 0.5\right)\\ \end{array} \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z)
 :precision binary64
 (*
  y_s
  (if (<= y_m 5.8e+127)
    (* (+ y_m (/ (* (+ x z) (- x z)) y_m)) 0.5)
    (fma (* (/ z y_m) z) -0.5 (* y_m 0.5)))))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if (y_m <= 5.8e+127) {
		tmp = (y_m + (((x + z) * (x - z)) / y_m)) * 0.5;
	} else {
		tmp = fma(((z / y_m) * z), -0.5, (y_m * 0.5));
	}
	return y_s * tmp;
}

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	tmp = 0.0
	if (y_m <= 5.8e+127)
		tmp = Float64(Float64(y_m + Float64(Float64(Float64(x + z) * Float64(x - z)) / y_m)) * 0.5);
	else
		tmp = fma(Float64(Float64(z / y_m) * z), -0.5, Float64(y_m * 0.5));
	end
	return Float64(y_s * tmp)
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * If[LessEqual[y$95$m, 5.8e+127], N[(N[(y$95$m + N[(N[(N[(x + z), $MachinePrecision] * N[(x - z), $MachinePrecision]), $MachinePrecision] / y$95$m), $MachinePrecision]), $MachinePrecision] * 0.5), $MachinePrecision], N[(N[(N[(z / y$95$m), $MachinePrecision] * z), $MachinePrecision] * -0.5 + N[(y$95$m * 0.5), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if y < 5.8000000000000004e127

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

      1. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2}} \]

      2. lift-*.f64 N/A

         \[\leadsto \frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{\color{blue}{y \cdot 2}} \]

      3. associate-/r* N/A

         \[\leadsto \color{blue}{\frac{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y}}{2}} \]

      4. mult-flip N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y} \cdot \frac{1}{2}} \]

      5. lower-*.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y} \cdot \frac{1}{2}} \]

   3. Applied rewrites 89.1%

      \[\leadsto \color{blue}{\left(y + \frac{\left(x + z\right) \cdot \left(x - z\right)}{y}\right) \cdot 0.5} \]

   if 5.8000000000000004e127 < y

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

      1. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2}} \]

      2. frac-2neg N/A

         \[\leadsto \color{blue}{\frac{\mathsf{neg}\left(\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      3. lift--.f64 N/A

         \[\leadsto \frac{\mathsf{neg}\left(\color{blue}{\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)}\right)}{\mathsf{neg}\left(y \cdot 2\right)} \]

      4. sub-negate-rev N/A

         \[\leadsto \frac{\color{blue}{z \cdot z - \left(x \cdot x + y \cdot y\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      5. sub-flip N/A

         \[\leadsto \frac{\color{blue}{z \cdot z + \left(\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      6. div-add N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \frac{\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      7. frac-2neg N/A

         \[\leadsto \frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \color{blue}{\frac{x \cdot x + y \cdot y}{y \cdot 2}} \]

      8. distribute-neg-frac2 N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\frac{z \cdot z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      9. lift-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(\frac{\color{blue}{z \cdot z}}{y \cdot 2}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      10. associate-/l* N/A

          \[\leadsto \left(\mathsf{neg}\left(\color{blue}{z \cdot \frac{z}{y \cdot 2}}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      11. distribute-rgt-neg-in N/A

          \[\leadsto \color{blue}{z \cdot \left(\mathsf{neg}\left(\frac{z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      12. distribute-frac-neg N/A

          \[\leadsto z \cdot \color{blue}{\frac{\mathsf{neg}\left(z\right)}{y \cdot 2}} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      13. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{\mathsf{neg}\left(z\right)}{y \cdot 2}, \frac{x \cdot x + y \cdot y}{y \cdot 2}\right)} \]

   3. Applied rewrites 87.2%

      \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \left(y + \frac{x \cdot x}{y}\right) \cdot 0.5\right)} \]

   4. Taylor expanded in x around 0

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot \frac{1}{2}\right) \]
   5. Step-by-step derivation

   6. Applied rewrites 68.0%

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot 0.5\right) \]
   7. Step-by-step derivation

      1. lift-fma.f64 N/A

         \[\leadsto \color{blue}{z \cdot \frac{z}{-2 \cdot y} + y \cdot \frac{1}{2}} \]

      2. lift-/.f64 N/A

         \[\leadsto z \cdot \color{blue}{\frac{z}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      3. associate-*r/ N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      4. lift-*.f64 N/A

         \[\leadsto \frac{z \cdot z}{\color{blue}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      5. *-commutative N/A

         \[\leadsto \frac{z \cdot z}{\color{blue}{y \cdot -2}} + y \cdot \frac{1}{2} \]

      6. times-frac N/A

         \[\leadsto \color{blue}{\frac{z}{y} \cdot \frac{z}{-2}} + y \cdot \frac{1}{2} \]

      7. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{z}{y}} \cdot \frac{z}{-2} + y \cdot \frac{1}{2} \]

      8. mult-flip N/A

         \[\leadsto \frac{z}{y} \cdot \color{blue}{\left(z \cdot \frac{1}{-2}\right)} + y \cdot \frac{1}{2} \]

      9. metadata-eval N/A

         \[\leadsto \frac{z}{y} \cdot \left(z \cdot \color{blue}{\frac{-1}{2}}\right) + y \cdot \frac{1}{2} \]

      10. associate-*r* N/A

          \[\leadsto \color{blue}{\left(\frac{z}{y} \cdot z\right) \cdot \frac{-1}{2}} + y \cdot \frac{1}{2} \]

      11. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{z}{y} \cdot z, \frac{-1}{2}, y \cdot \frac{1}{2}\right)} \]

      12. lower-*.f64 68.0

          \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{z}{y} \cdot z}, -0.5, y \cdot 0.5\right) \]

   8. Applied rewrites 68.0%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{z}{y} \cdot z, -0.5, y \cdot 0.5\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 4: 84.6% accurate, 1.1× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \begin{array}{l} \mathbf{if}\;y\_m \leq 1.9 \cdot 10^{+76}:\\ \;\;\;\;\frac{\left(x + z\right) \cdot \left(x - z\right)}{y\_m + y\_m}\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(\frac{z}{y\_m} \cdot z, -0.5, y\_m \cdot 0.5\right)\\ \end{array} \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z)
 :precision binary64
 (*
  y_s
  (if (<= y_m 1.9e+76)
    (/ (* (+ x z) (- x z)) (+ y_m y_m))
    (fma (* (/ z y_m) z) -0.5 (* y_m 0.5)))))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if (y_m <= 1.9e+76) {
		tmp = ((x + z) * (x - z)) / (y_m + y_m);
	} else {
		tmp = fma(((z / y_m) * z), -0.5, (y_m * 0.5));
	}
	return y_s * tmp;
}

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	tmp = 0.0
	if (y_m <= 1.9e+76)
		tmp = Float64(Float64(Float64(x + z) * Float64(x - z)) / Float64(y_m + y_m));
	else
		tmp = fma(Float64(Float64(z / y_m) * z), -0.5, Float64(y_m * 0.5));
	end
	return Float64(y_s * tmp)
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * If[LessEqual[y$95$m, 1.9e+76], N[(N[(N[(x + z), $MachinePrecision] * N[(x - z), $MachinePrecision]), $MachinePrecision] / N[(y$95$m + y$95$m), $MachinePrecision]), $MachinePrecision], N[(N[(N[(z / y$95$m), $MachinePrecision] * z), $MachinePrecision] * -0.5 + N[(y$95$m * 0.5), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if y < 1.90000000000000012e76

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

   3. Applied rewrites 73.7%

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

   4. Taylor expanded in y around 0

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

      1. lower-*.f64 N/A

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

      2. lower-+.f64 N/A

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

      3. lower--.f64 61.4

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

   6. Applied rewrites 61.4%

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

   if 1.90000000000000012e76 < y

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

      1. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2}} \]

      2. frac-2neg N/A

         \[\leadsto \color{blue}{\frac{\mathsf{neg}\left(\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      3. lift--.f64 N/A

         \[\leadsto \frac{\mathsf{neg}\left(\color{blue}{\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)}\right)}{\mathsf{neg}\left(y \cdot 2\right)} \]

      4. sub-negate-rev N/A

         \[\leadsto \frac{\color{blue}{z \cdot z - \left(x \cdot x + y \cdot y\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      5. sub-flip N/A

         \[\leadsto \frac{\color{blue}{z \cdot z + \left(\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      6. div-add N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \frac{\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      7. frac-2neg N/A

         \[\leadsto \frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \color{blue}{\frac{x \cdot x + y \cdot y}{y \cdot 2}} \]

      8. distribute-neg-frac2 N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\frac{z \cdot z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      9. lift-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(\frac{\color{blue}{z \cdot z}}{y \cdot 2}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      10. associate-/l* N/A

          \[\leadsto \left(\mathsf{neg}\left(\color{blue}{z \cdot \frac{z}{y \cdot 2}}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      11. distribute-rgt-neg-in N/A

          \[\leadsto \color{blue}{z \cdot \left(\mathsf{neg}\left(\frac{z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      12. distribute-frac-neg N/A

          \[\leadsto z \cdot \color{blue}{\frac{\mathsf{neg}\left(z\right)}{y \cdot 2}} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      13. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{\mathsf{neg}\left(z\right)}{y \cdot 2}, \frac{x \cdot x + y \cdot y}{y \cdot 2}\right)} \]

   3. Applied rewrites 87.2%

      \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \left(y + \frac{x \cdot x}{y}\right) \cdot 0.5\right)} \]

   4. Taylor expanded in x around 0

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot \frac{1}{2}\right) \]
   5. Step-by-step derivation

   6. Applied rewrites 68.0%

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot 0.5\right) \]
   7. Step-by-step derivation

      1. lift-fma.f64 N/A

         \[\leadsto \color{blue}{z \cdot \frac{z}{-2 \cdot y} + y \cdot \frac{1}{2}} \]

      2. lift-/.f64 N/A

         \[\leadsto z \cdot \color{blue}{\frac{z}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      3. associate-*r/ N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      4. lift-*.f64 N/A

         \[\leadsto \frac{z \cdot z}{\color{blue}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      5. *-commutative N/A

         \[\leadsto \frac{z \cdot z}{\color{blue}{y \cdot -2}} + y \cdot \frac{1}{2} \]

      6. times-frac N/A

         \[\leadsto \color{blue}{\frac{z}{y} \cdot \frac{z}{-2}} + y \cdot \frac{1}{2} \]

      7. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{z}{y}} \cdot \frac{z}{-2} + y \cdot \frac{1}{2} \]

      8. mult-flip N/A

         \[\leadsto \frac{z}{y} \cdot \color{blue}{\left(z \cdot \frac{1}{-2}\right)} + y \cdot \frac{1}{2} \]

      9. metadata-eval N/A

         \[\leadsto \frac{z}{y} \cdot \left(z \cdot \color{blue}{\frac{-1}{2}}\right) + y \cdot \frac{1}{2} \]

      10. associate-*r* N/A

          \[\leadsto \color{blue}{\left(\frac{z}{y} \cdot z\right) \cdot \frac{-1}{2}} + y \cdot \frac{1}{2} \]

      11. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{z}{y} \cdot z, \frac{-1}{2}, y \cdot \frac{1}{2}\right)} \]

      12. lower-*.f64 68.0

          \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{z}{y} \cdot z}, -0.5, y \cdot 0.5\right) \]

   8. Applied rewrites 68.0%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{z}{y} \cdot z, -0.5, y \cdot 0.5\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 5: 68.6% accurate, 1.0× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \begin{array}{l} \mathbf{if}\;x \cdot x \leq 1.4 \cdot 10^{+288}:\\ \;\;\;\;\mathsf{fma}\left(\frac{z}{y\_m} \cdot z, -0.5, y\_m \cdot 0.5\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{\frac{\left(x - z\right) \cdot z}{2}}{y\_m}\\ \end{array} \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z)
 :precision binary64
 (*
  y_s
  (if (<= (* x x) 1.4e+288)
    (fma (* (/ z y_m) z) -0.5 (* y_m 0.5))
    (/ (/ (* (- x z) z) 2.0) y_m))))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if ((x * x) <= 1.4e+288) {
		tmp = fma(((z / y_m) * z), -0.5, (y_m * 0.5));
	} else {
		tmp = (((x - z) * z) / 2.0) / y_m;
	}
	return y_s * tmp;
}

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	tmp = 0.0
	if (Float64(x * x) <= 1.4e+288)
		tmp = fma(Float64(Float64(z / y_m) * z), -0.5, Float64(y_m * 0.5));
	else
		tmp = Float64(Float64(Float64(Float64(x - z) * z) / 2.0) / y_m);
	end
	return Float64(y_s * tmp)
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * If[LessEqual[N[(x * x), $MachinePrecision], 1.4e+288], N[(N[(N[(z / y$95$m), $MachinePrecision] * z), $MachinePrecision] * -0.5 + N[(y$95$m * 0.5), $MachinePrecision]), $MachinePrecision], N[(N[(N[(N[(x - z), $MachinePrecision] * z), $MachinePrecision] / 2.0), $MachinePrecision] / y$95$m), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 x x) < 1.3999999999999999e288

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

      1. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2}} \]

      2. frac-2neg N/A

         \[\leadsto \color{blue}{\frac{\mathsf{neg}\left(\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      3. lift--.f64 N/A

         \[\leadsto \frac{\mathsf{neg}\left(\color{blue}{\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)}\right)}{\mathsf{neg}\left(y \cdot 2\right)} \]

      4. sub-negate-rev N/A

         \[\leadsto \frac{\color{blue}{z \cdot z - \left(x \cdot x + y \cdot y\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      5. sub-flip N/A

         \[\leadsto \frac{\color{blue}{z \cdot z + \left(\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      6. div-add N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \frac{\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      7. frac-2neg N/A

         \[\leadsto \frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \color{blue}{\frac{x \cdot x + y \cdot y}{y \cdot 2}} \]

      8. distribute-neg-frac2 N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\frac{z \cdot z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      9. lift-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(\frac{\color{blue}{z \cdot z}}{y \cdot 2}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      10. associate-/l* N/A

          \[\leadsto \left(\mathsf{neg}\left(\color{blue}{z \cdot \frac{z}{y \cdot 2}}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      11. distribute-rgt-neg-in N/A

          \[\leadsto \color{blue}{z \cdot \left(\mathsf{neg}\left(\frac{z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      12. distribute-frac-neg N/A

          \[\leadsto z \cdot \color{blue}{\frac{\mathsf{neg}\left(z\right)}{y \cdot 2}} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      13. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{\mathsf{neg}\left(z\right)}{y \cdot 2}, \frac{x \cdot x + y \cdot y}{y \cdot 2}\right)} \]

   3. Applied rewrites 87.2%

      \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \left(y + \frac{x \cdot x}{y}\right) \cdot 0.5\right)} \]

   4. Taylor expanded in x around 0

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot \frac{1}{2}\right) \]
   5. Step-by-step derivation

   6. Applied rewrites 68.0%

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot 0.5\right) \]
   7. Step-by-step derivation

      1. lift-fma.f64 N/A

         \[\leadsto \color{blue}{z \cdot \frac{z}{-2 \cdot y} + y \cdot \frac{1}{2}} \]

      2. lift-/.f64 N/A

         \[\leadsto z \cdot \color{blue}{\frac{z}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      3. associate-*r/ N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      4. lift-*.f64 N/A

         \[\leadsto \frac{z \cdot z}{\color{blue}{-2 \cdot y}} + y \cdot \frac{1}{2} \]

      5. *-commutative N/A

         \[\leadsto \frac{z \cdot z}{\color{blue}{y \cdot -2}} + y \cdot \frac{1}{2} \]

      6. times-frac N/A

         \[\leadsto \color{blue}{\frac{z}{y} \cdot \frac{z}{-2}} + y \cdot \frac{1}{2} \]

      7. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{z}{y}} \cdot \frac{z}{-2} + y \cdot \frac{1}{2} \]

      8. mult-flip N/A

         \[\leadsto \frac{z}{y} \cdot \color{blue}{\left(z \cdot \frac{1}{-2}\right)} + y \cdot \frac{1}{2} \]

      9. metadata-eval N/A

         \[\leadsto \frac{z}{y} \cdot \left(z \cdot \color{blue}{\frac{-1}{2}}\right) + y \cdot \frac{1}{2} \]

      10. associate-*r* N/A

          \[\leadsto \color{blue}{\left(\frac{z}{y} \cdot z\right) \cdot \frac{-1}{2}} + y \cdot \frac{1}{2} \]

      11. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{z}{y} \cdot z, \frac{-1}{2}, y \cdot \frac{1}{2}\right)} \]

      12. lower-*.f64 68.0

          \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{z}{y} \cdot z}, -0.5, y \cdot 0.5\right) \]

   8. Applied rewrites 68.0%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{z}{y} \cdot z, -0.5, y \cdot 0.5\right)} \]

   if 1.3999999999999999e288 < (*.f64 x x)

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

   3. Applied rewrites 73.7%

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

   4. Taylor expanded in y around 0

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

      1. lower-*.f64 N/A

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

      2. lower-+.f64 N/A

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

      3. lower--.f64 61.4

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

   6. Applied rewrites 61.4%

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

   7. Taylor expanded in x around 0

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

   9. Applied rewrites 36.9%

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

       1. lift-/.f64 N/A

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

       2. lift-+.f64 N/A

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

       3. count-2 N/A

          \[\leadsto \frac{z \cdot \left(x - z\right)}{\color{blue}{2 \cdot y}} \]

       4. associate-/r* N/A

          \[\leadsto \color{blue}{\frac{\frac{z \cdot \left(x - z\right)}{2}}{y}} \]

       5. lower-/.f64 N/A

          \[\leadsto \color{blue}{\frac{\frac{z \cdot \left(x - z\right)}{2}}{y}} \]

       6. lower-/.f64 36.9

          \[\leadsto \frac{\color{blue}{\frac{z \cdot \left(x - z\right)}{2}}}{y} \]

       7. lift-*.f64 N/A

          \[\leadsto \frac{\frac{z \cdot \color{blue}{\left(x - z\right)}}{2}}{y} \]

       8. *-commutative N/A

          \[\leadsto \frac{\frac{\left(x - z\right) \cdot \color{blue}{z}}{2}}{y} \]

       9. lower-*.f64 36.9

          \[\leadsto \frac{\frac{\left(x - z\right) \cdot \color{blue}{z}}{2}}{y} \]

   11. Applied rewrites 36.9%

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

Alternative 6: 68.6% accurate, 1.0× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \begin{array}{l} \mathbf{if}\;x \cdot x \leq 1.4 \cdot 10^{+288}:\\ \;\;\;\;\mathsf{fma}\left(z, \frac{z}{-2 \cdot y\_m}, y\_m \cdot 0.5\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{\frac{\left(x - z\right) \cdot z}{2}}{y\_m}\\ \end{array} \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z)
 :precision binary64
 (*
  y_s
  (if (<= (* x x) 1.4e+288)
    (fma z (/ z (* -2.0 y_m)) (* y_m 0.5))
    (/ (/ (* (- x z) z) 2.0) y_m))))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if ((x * x) <= 1.4e+288) {
		tmp = fma(z, (z / (-2.0 * y_m)), (y_m * 0.5));
	} else {
		tmp = (((x - z) * z) / 2.0) / y_m;
	}
	return y_s * tmp;
}

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	tmp = 0.0
	if (Float64(x * x) <= 1.4e+288)
		tmp = fma(z, Float64(z / Float64(-2.0 * y_m)), Float64(y_m * 0.5));
	else
		tmp = Float64(Float64(Float64(Float64(x - z) * z) / 2.0) / y_m);
	end
	return Float64(y_s * tmp)
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * If[LessEqual[N[(x * x), $MachinePrecision], 1.4e+288], N[(z * N[(z / N[(-2.0 * y$95$m), $MachinePrecision]), $MachinePrecision] + N[(y$95$m * 0.5), $MachinePrecision]), $MachinePrecision], N[(N[(N[(N[(x - z), $MachinePrecision] * z), $MachinePrecision] / 2.0), $MachinePrecision] / y$95$m), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if (*.f64 x x) < 1.3999999999999999e288

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

      1. lift-/.f64 N/A

         \[\leadsto \color{blue}{\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2}} \]

      2. frac-2neg N/A

         \[\leadsto \color{blue}{\frac{\mathsf{neg}\left(\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      3. lift--.f64 N/A

         \[\leadsto \frac{\mathsf{neg}\left(\color{blue}{\left(\left(x \cdot x + y \cdot y\right) - z \cdot z\right)}\right)}{\mathsf{neg}\left(y \cdot 2\right)} \]

      4. sub-negate-rev N/A

         \[\leadsto \frac{\color{blue}{z \cdot z - \left(x \cdot x + y \cdot y\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      5. sub-flip N/A

         \[\leadsto \frac{\color{blue}{z \cdot z + \left(\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)\right)}}{\mathsf{neg}\left(y \cdot 2\right)} \]

      6. div-add N/A

         \[\leadsto \color{blue}{\frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \frac{\mathsf{neg}\left(\left(x \cdot x + y \cdot y\right)\right)}{\mathsf{neg}\left(y \cdot 2\right)}} \]

      7. frac-2neg N/A

         \[\leadsto \frac{z \cdot z}{\mathsf{neg}\left(y \cdot 2\right)} + \color{blue}{\frac{x \cdot x + y \cdot y}{y \cdot 2}} \]

      8. distribute-neg-frac2 N/A

         \[\leadsto \color{blue}{\left(\mathsf{neg}\left(\frac{z \cdot z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      9. lift-*.f64 N/A

         \[\leadsto \left(\mathsf{neg}\left(\frac{\color{blue}{z \cdot z}}{y \cdot 2}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      10. associate-/l* N/A

          \[\leadsto \left(\mathsf{neg}\left(\color{blue}{z \cdot \frac{z}{y \cdot 2}}\right)\right) + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      11. distribute-rgt-neg-in N/A

          \[\leadsto \color{blue}{z \cdot \left(\mathsf{neg}\left(\frac{z}{y \cdot 2}\right)\right)} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      12. distribute-frac-neg N/A

          \[\leadsto z \cdot \color{blue}{\frac{\mathsf{neg}\left(z\right)}{y \cdot 2}} + \frac{x \cdot x + y \cdot y}{y \cdot 2} \]

      13. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{\mathsf{neg}\left(z\right)}{y \cdot 2}, \frac{x \cdot x + y \cdot y}{y \cdot 2}\right)} \]

   3. Applied rewrites 87.2%

      \[\leadsto \color{blue}{\mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \left(y + \frac{x \cdot x}{y}\right) \cdot 0.5\right)} \]

   4. Taylor expanded in x around 0

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot \frac{1}{2}\right) \]
   5. Step-by-step derivation

   6. Applied rewrites 68.0%

      \[\leadsto \mathsf{fma}\left(z, \frac{z}{-2 \cdot y}, \color{blue}{y} \cdot 0.5\right) \]

   if 1.3999999999999999e288 < (*.f64 x x)

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

   3. Applied rewrites 73.7%

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

   4. Taylor expanded in y around 0

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

      1. lower-*.f64 N/A

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

      2. lower-+.f64 N/A

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

      3. lower--.f64 61.4

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

   6. Applied rewrites 61.4%

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

   7. Taylor expanded in x around 0

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

   9. Applied rewrites 36.9%

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

       1. lift-/.f64 N/A

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

       2. lift-+.f64 N/A

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

       3. count-2 N/A

          \[\leadsto \frac{z \cdot \left(x - z\right)}{\color{blue}{2 \cdot y}} \]

       4. associate-/r* N/A

          \[\leadsto \color{blue}{\frac{\frac{z \cdot \left(x - z\right)}{2}}{y}} \]

       5. lower-/.f64 N/A

          \[\leadsto \color{blue}{\frac{\frac{z \cdot \left(x - z\right)}{2}}{y}} \]

       6. lower-/.f64 36.9

          \[\leadsto \frac{\color{blue}{\frac{z \cdot \left(x - z\right)}{2}}}{y} \]

       7. lift-*.f64 N/A

          \[\leadsto \frac{\frac{z \cdot \color{blue}{\left(x - z\right)}}{2}}{y} \]

       8. *-commutative N/A

          \[\leadsto \frac{\frac{\left(x - z\right) \cdot \color{blue}{z}}{2}}{y} \]

       9. lower-*.f64 36.9

          \[\leadsto \frac{\frac{\left(x - z\right) \cdot \color{blue}{z}}{2}}{y} \]

   11. Applied rewrites 36.9%

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

Alternative 7: 45.4% accurate, 1.3× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \begin{array}{l} \mathbf{if}\;z \leq 17.5:\\ \;\;\;\;0.5 \cdot y\_m\\ \mathbf{else}:\\ \;\;\;\;\frac{0.5}{y\_m} \cdot \left(\left(x - z\right) \cdot z\right)\\ \end{array} \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z)
 :precision binary64
 (* y_s (if (<= z 17.5) (* 0.5 y_m) (* (/ 0.5 y_m) (* (- x z) z)))))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if (z <= 17.5) {
		tmp = 0.5 * y_m;
	} else {
		tmp = (0.5 / y_m) * ((x - z) * z);
	}
	return y_s * tmp;
}

Fortran

y\_m =     private
y\_s =     private
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(y_s, x, y_m, z)
use fmin_fmax_functions
    real(8), intent (in) :: y_s
    real(8), intent (in) :: x
    real(8), intent (in) :: y_m
    real(8), intent (in) :: z
    real(8) :: tmp
    if (z <= 17.5d0) then
        tmp = 0.5d0 * y_m
    else
        tmp = (0.5d0 / y_m) * ((x - z) * z)
    end if
    code = y_s * tmp
end function

Java

y\_m = Math.abs(y);
y\_s = Math.copySign(1.0, y);
public static double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if (z <= 17.5) {
		tmp = 0.5 * y_m;
	} else {
		tmp = (0.5 / y_m) * ((x - z) * z);
	}
	return y_s * tmp;
}

Python

y\_m = math.fabs(y)
y\_s = math.copysign(1.0, y)
def code(y_s, x, y_m, z):
	tmp = 0
	if z <= 17.5:
		tmp = 0.5 * y_m
	else:
		tmp = (0.5 / y_m) * ((x - z) * z)
	return y_s * tmp

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	tmp = 0.0
	if (z <= 17.5)
		tmp = Float64(0.5 * y_m);
	else
		tmp = Float64(Float64(0.5 / y_m) * Float64(Float64(x - z) * z));
	end
	return Float64(y_s * tmp)
end

MATLAB

y\_m = abs(y);
y\_s = sign(y) * abs(1.0);
function tmp_2 = code(y_s, x, y_m, z)
	tmp = 0.0;
	if (z <= 17.5)
		tmp = 0.5 * y_m;
	else
		tmp = (0.5 / y_m) * ((x - z) * z);
	end
	tmp_2 = y_s * tmp;
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * If[LessEqual[z, 17.5], N[(0.5 * y$95$m), $MachinePrecision], N[(N[(0.5 / y$95$m), $MachinePrecision] * N[(N[(x - z), $MachinePrecision] * z), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if z < 17.5

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{\frac{1}{2} \cdot y} \]
   3. Step-by-step derivation

      1. lower-*.f64 34.7

         \[\leadsto 0.5 \cdot \color{blue}{y} \]

   4. Applied rewrites 34.7%

      \[\leadsto \color{blue}{0.5 \cdot y} \]

   if 17.5 < z

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

   3. Applied rewrites 73.7%

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

   4. Taylor expanded in y around 0

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

      1. lower-*.f64 N/A

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

      2. lower-+.f64 N/A

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

      3. lower--.f64 61.4

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

   6. Applied rewrites 61.4%

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

   7. Taylor expanded in x around 0

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

   9. Applied rewrites 36.9%

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

       1. lift-/.f64 N/A

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

       2. mult-flip N/A

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

       3. lift-+.f64 N/A

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

       4. count-2 N/A

          \[\leadsto \left(z \cdot \left(x - z\right)\right) \cdot \frac{1}{\color{blue}{2 \cdot y}} \]

       5. *-commutative N/A

          \[\leadsto \left(z \cdot \left(x - z\right)\right) \cdot \frac{1}{\color{blue}{y \cdot 2}} \]

       6. *-commutative N/A

          \[\leadsto \color{blue}{\frac{1}{y \cdot 2} \cdot \left(z \cdot \left(x - z\right)\right)} \]

       7. lower-*.f64 N/A

          \[\leadsto \color{blue}{\frac{1}{y \cdot 2} \cdot \left(z \cdot \left(x - z\right)\right)} \]

       8. *-commutative N/A

          \[\leadsto \frac{1}{\color{blue}{2 \cdot y}} \cdot \left(z \cdot \left(x - z\right)\right) \]

       9. count-2 N/A

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

       10. lift-+.f64 N/A

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

       11. lift-+.f64 N/A

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

       12. count-2 N/A

           \[\leadsto \frac{1}{\color{blue}{2 \cdot y}} \cdot \left(z \cdot \left(x - z\right)\right) \]

       13. associate-/r* N/A

           \[\leadsto \color{blue}{\frac{\frac{1}{2}}{y}} \cdot \left(z \cdot \left(x - z\right)\right) \]

       14. metadata-eval N/A

           \[\leadsto \frac{\color{blue}{\frac{1}{2}}}{y} \cdot \left(z \cdot \left(x - z\right)\right) \]

       15. lift-/.f64 36.8

           \[\leadsto \color{blue}{\frac{0.5}{y}} \cdot \left(z \cdot \left(x - z\right)\right) \]

       16. lift-*.f64 N/A

           \[\leadsto \frac{\frac{1}{2}}{y} \cdot \left(z \cdot \color{blue}{\left(x - z\right)}\right) \]

       17. *-commutative N/A

           \[\leadsto \frac{\frac{1}{2}}{y} \cdot \left(\left(x - z\right) \cdot \color{blue}{z}\right) \]

       18. lower-*.f64 36.8

           \[\leadsto \frac{0.5}{y} \cdot \left(\left(x - z\right) \cdot \color{blue}{z}\right) \]

   11. Applied rewrites 36.8%

       \[\leadsto \color{blue}{\frac{0.5}{y} \cdot \left(\left(x - z\right) \cdot z\right)} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 8: 45.4% accurate, 1.3× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \begin{array}{l} \mathbf{if}\;z \leq 17.5:\\ \;\;\;\;0.5 \cdot y\_m\\ \mathbf{else}:\\ \;\;\;\;\frac{z \cdot \left(x - z\right)}{y\_m + y\_m}\\ \end{array} \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z)
 :precision binary64
 (* y_s (if (<= z 17.5) (* 0.5 y_m) (/ (* z (- x z)) (+ y_m y_m)))))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if (z <= 17.5) {
		tmp = 0.5 * y_m;
	} else {
		tmp = (z * (x - z)) / (y_m + y_m);
	}
	return y_s * tmp;
}

Fortran

y\_m =     private
y\_s =     private
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(y_s, x, y_m, z)
use fmin_fmax_functions
    real(8), intent (in) :: y_s
    real(8), intent (in) :: x
    real(8), intent (in) :: y_m
    real(8), intent (in) :: z
    real(8) :: tmp
    if (z <= 17.5d0) then
        tmp = 0.5d0 * y_m
    else
        tmp = (z * (x - z)) / (y_m + y_m)
    end if
    code = y_s * tmp
end function

Java

y\_m = Math.abs(y);
y\_s = Math.copySign(1.0, y);
public static double code(double y_s, double x, double y_m, double z) {
	double tmp;
	if (z <= 17.5) {
		tmp = 0.5 * y_m;
	} else {
		tmp = (z * (x - z)) / (y_m + y_m);
	}
	return y_s * tmp;
}

Python

y\_m = math.fabs(y)
y\_s = math.copysign(1.0, y)
def code(y_s, x, y_m, z):
	tmp = 0
	if z <= 17.5:
		tmp = 0.5 * y_m
	else:
		tmp = (z * (x - z)) / (y_m + y_m)
	return y_s * tmp

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	tmp = 0.0
	if (z <= 17.5)
		tmp = Float64(0.5 * y_m);
	else
		tmp = Float64(Float64(z * Float64(x - z)) / Float64(y_m + y_m));
	end
	return Float64(y_s * tmp)
end

MATLAB

y\_m = abs(y);
y\_s = sign(y) * abs(1.0);
function tmp_2 = code(y_s, x, y_m, z)
	tmp = 0.0;
	if (z <= 17.5)
		tmp = 0.5 * y_m;
	else
		tmp = (z * (x - z)) / (y_m + y_m);
	end
	tmp_2 = y_s * tmp;
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * If[LessEqual[z, 17.5], N[(0.5 * y$95$m), $MachinePrecision], N[(N[(z * N[(x - z), $MachinePrecision]), $MachinePrecision] / N[(y$95$m + y$95$m), $MachinePrecision]), $MachinePrecision]]), $MachinePrecision]

Derivation

1. Split input into 2 regimes

2. if z < 17.5

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]

   2. Taylor expanded in y around inf

      \[\leadsto \color{blue}{\frac{1}{2} \cdot y} \]
   3. Step-by-step derivation

      1. lower-*.f64 34.7

         \[\leadsto 0.5 \cdot \color{blue}{y} \]

   4. Applied rewrites 34.7%

      \[\leadsto \color{blue}{0.5 \cdot y} \]

   if 17.5 < z

   1. Initial program 68.0%

      \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]
   2. Step-by-step derivation

   3. Applied rewrites 73.7%

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

   4. Taylor expanded in y around 0

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

      1. lower-*.f64 N/A

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

      2. lower-+.f64 N/A

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

      3. lower--.f64 61.4

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

   6. Applied rewrites 61.4%

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

   7. Taylor expanded in x around 0

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

   9. Applied rewrites 36.9%

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

Alternative 9: 34.7% accurate, 5.4× speedup

Math

\[\begin{array}{l} y\_m = \left|y\right| \\ y\_s = \mathsf{copysign}\left(1, y\right) \\ y\_s \cdot \left(0.5 \cdot y\_m\right) \end{array} \]

FPCore

y\_m = (fabs.f64 y)
y\_s = (copysign.f64 #s(literal 1 binary64) y)
(FPCore (y_s x y_m z) :precision binary64 (* y_s (* 0.5 y_m)))

C

y\_m = fabs(y);
y\_s = copysign(1.0, y);
double code(double y_s, double x, double y_m, double z) {
	return y_s * (0.5 * y_m);
}

Fortran

y\_m =     private
y\_s =     private
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(y_s, x, y_m, z)
use fmin_fmax_functions
    real(8), intent (in) :: y_s
    real(8), intent (in) :: x
    real(8), intent (in) :: y_m
    real(8), intent (in) :: z
    code = y_s * (0.5d0 * y_m)
end function

Java

y\_m = Math.abs(y);
y\_s = Math.copySign(1.0, y);
public static double code(double y_s, double x, double y_m, double z) {
	return y_s * (0.5 * y_m);
}

Python

y\_m = math.fabs(y)
y\_s = math.copysign(1.0, y)
def code(y_s, x, y_m, z):
	return y_s * (0.5 * y_m)

Julia

y\_m = abs(y)
y\_s = copysign(1.0, y)
function code(y_s, x, y_m, z)
	return Float64(y_s * Float64(0.5 * y_m))
end

MATLAB

y\_m = abs(y);
y\_s = sign(y) * abs(1.0);
function tmp = code(y_s, x, y_m, z)
	tmp = y_s * (0.5 * y_m);
end

Wolfram

y\_m = N[Abs[y], $MachinePrecision]
y\_s = N[With[{TMP1 = Abs[1.0], TMP2 = Sign[y]}, TMP1 * If[TMP2 == 0, 1, TMP2]], $MachinePrecision]
code[y$95$s_, x_, y$95$m_, z_] := N[(y$95$s * N[(0.5 * y$95$m), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 68.0%

   \[\frac{\left(x \cdot x + y \cdot y\right) - z \cdot z}{y \cdot 2} \]

2. Taylor expanded in y around inf

   \[\leadsto \color{blue}{\frac{1}{2} \cdot y} \]
3. Step-by-step derivation

   1. lower-*.f64 34.7

      \[\leadsto 0.5 \cdot \color{blue}{y} \]

4. Applied rewrites 34.7%

   \[\leadsto \color{blue}{0.5 \cdot y} \]
5. Add Preprocessing

Reproduce

herbie shell --seed 2025156 
(FPCore (x y z)
  :name "Diagrams.TwoD.Apollonian:initialConfig from diagrams-contrib-1.3.0.5, A"
  :precision binary64
  (/ (- (+ (* x x) (* y y)) (* z z)) (* y 2.0)))