Numeric.SpecFunctions:incompleteGamma from math-functions-0.1.5.2, B

Percentage Accurate: 99.4% → 99.4%

Time: 7.2s

Alternatives: 9

Speedup: 1.2×

Specification

Math

\[\begin{array}{l} \\ \left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (* (* 3.0 (sqrt x)) (- (+ y (/ 1.0 (* x 9.0))) 1.0)))

C

double code(double x, double y) {
	return (3.0 * sqrt(x)) * ((y + (1.0 / (x * 9.0))) - 1.0);
}

Fortran

real(8) function code(x, y)
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    code = (3.0d0 * sqrt(x)) * ((y + (1.0d0 / (x * 9.0d0))) - 1.0d0)
end function

Java

public static double code(double x, double y) {
	return (3.0 * Math.sqrt(x)) * ((y + (1.0 / (x * 9.0))) - 1.0);
}

Python

def code(x, y):
	return (3.0 * math.sqrt(x)) * ((y + (1.0 / (x * 9.0))) - 1.0)

Julia

function code(x, y)
	return Float64(Float64(3.0 * sqrt(x)) * Float64(Float64(y + Float64(1.0 / Float64(x * 9.0))) - 1.0))
end

MATLAB

function tmp = code(x, y)
	tmp = (3.0 * sqrt(x)) * ((y + (1.0 / (x * 9.0))) - 1.0);
end

Wolfram

code[x_, y_] := N[(N[(3.0 * N[Sqrt[x], $MachinePrecision]), $MachinePrecision] * N[(N[(y + N[(1.0 / N[(x * 9.0), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - 1.0), $MachinePrecision]), $MachinePrecision]

Initial Program: 99.4% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (* (* 3.0 (sqrt x)) (- (+ y (/ 1.0 (* x 9.0))) 1.0)))

C

double code(double x, double y) {
	return (3.0 * sqrt(x)) * ((y + (1.0 / (x * 9.0))) - 1.0);
}

Fortran

real(8) function code(x, y)
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    code = (3.0d0 * sqrt(x)) * ((y + (1.0d0 / (x * 9.0d0))) - 1.0d0)
end function

Java

public static double code(double x, double y) {
	return (3.0 * Math.sqrt(x)) * ((y + (1.0 / (x * 9.0))) - 1.0);
}

Python

def code(x, y):
	return (3.0 * math.sqrt(x)) * ((y + (1.0 / (x * 9.0))) - 1.0)

Julia

function code(x, y)
	return Float64(Float64(3.0 * sqrt(x)) * Float64(Float64(y + Float64(1.0 / Float64(x * 9.0))) - 1.0))
end

MATLAB

function tmp = code(x, y)
	tmp = (3.0 * sqrt(x)) * ((y + (1.0 / (x * 9.0))) - 1.0);
end

Wolfram

code[x_, y_] := N[(N[(3.0 * N[Sqrt[x], $MachinePrecision]), $MachinePrecision] * N[(N[(y + N[(1.0 / N[(x * 9.0), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] - 1.0), $MachinePrecision]), $MachinePrecision]

Alternative 1: 99.4% accurate, 1.2× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(1 - y, -3, \frac{0.3333333333333333}{x}\right) \cdot \sqrt{x} \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (* (fma (- 1.0 y) -3.0 (/ 0.3333333333333333 x)) (sqrt x)))

C

double code(double x, double y) {
	return fma((1.0 - y), -3.0, (0.3333333333333333 / x)) * sqrt(x);
}

Julia

function code(x, y)
	return Float64(fma(Float64(1.0 - y), -3.0, Float64(0.3333333333333333 / x)) * sqrt(x))
end

Wolfram

code[x_, y_] := N[(N[(N[(1.0 - y), $MachinePrecision] * -3.0 + N[(0.3333333333333333 / x), $MachinePrecision]), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 99.4%

   \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
2. Add Preprocessing

3. Taylor expanded in y around 0

   \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot y\right) + 3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]
4. Step-by-step derivation

   1. *-commutative N/A

      \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} + 3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \]

   2. associate-*l* N/A

      \[\leadsto \color{blue}{\sqrt{x} \cdot \left(y \cdot 3\right)} + 3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \]

   3. associate-*r* N/A

      \[\leadsto \sqrt{x} \cdot \left(y \cdot 3\right) + \color{blue}{\left(3 \cdot \sqrt{x}\right) \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)} \]

   4. *-commutative N/A

      \[\leadsto \sqrt{x} \cdot \left(y \cdot 3\right) + \color{blue}{\left(\sqrt{x} \cdot 3\right)} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \]

   5. associate-*l* N/A

      \[\leadsto \sqrt{x} \cdot \left(y \cdot 3\right) + \color{blue}{\sqrt{x} \cdot \left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]

   6. distribute-lft-out N/A

      \[\leadsto \color{blue}{\sqrt{x} \cdot \left(y \cdot 3 + 3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]

   7. lower-*.f64 N/A

      \[\leadsto \color{blue}{\sqrt{x} \cdot \left(y \cdot 3 + 3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]

   8. lower-sqrt.f64 N/A

      \[\leadsto \color{blue}{\sqrt{x}} \cdot \left(y \cdot 3 + 3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \]

   9. *-commutative N/A

      \[\leadsto \sqrt{x} \cdot \left(y \cdot 3 + \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \cdot 3}\right) \]

   10. distribute-rgt-in N/A

       \[\leadsto \sqrt{x} \cdot \color{blue}{\left(3 \cdot \left(y + \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)\right)} \]

   11. associate--l+ N/A

       \[\leadsto \sqrt{x} \cdot \left(3 \cdot \color{blue}{\left(\left(y + \frac{1}{9} \cdot \frac{1}{x}\right) - 1\right)}\right) \]

   12. +-commutative N/A

       \[\leadsto \sqrt{x} \cdot \left(3 \cdot \left(\color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + y\right)} - 1\right)\right) \]

   13. associate--l+ N/A

       \[\leadsto \sqrt{x} \cdot \left(3 \cdot \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + \left(y - 1\right)\right)}\right) \]

   14. +-commutative N/A

       \[\leadsto \sqrt{x} \cdot \left(3 \cdot \color{blue}{\left(\left(y - 1\right) + \frac{1}{9} \cdot \frac{1}{x}\right)}\right) \]

   15. distribute-rgt-in N/A

       \[\leadsto \sqrt{x} \cdot \color{blue}{\left(\left(y - 1\right) \cdot 3 + \left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3\right)} \]

5. Applied rewrites 99.5%

   \[\leadsto \color{blue}{\sqrt{x} \cdot \mathsf{fma}\left(1 - y, -3, \frac{0.3333333333333333}{x}\right)} \]

6. Final simplification 99.5%

   \[\leadsto \mathsf{fma}\left(1 - y, -3, \frac{0.3333333333333333}{x}\right) \cdot \sqrt{x} \]
7. Add Preprocessing

Alternative 2: 91.6% accurate, 0.3× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \sqrt{x} \cdot 3\\ t_1 := \left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot t\_0\\ \mathbf{if}\;t\_1 \leq -40000:\\ \;\;\;\;\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}\\ \mathbf{elif}\;t\_1 \leq 2 \cdot 10^{+149}:\\ \;\;\;\;\frac{\mathsf{fma}\left(-3, x, 0.3333333333333333\right)}{x} \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;\left(y - 1\right) \cdot t\_0\\ \end{array} \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (let* ((t_0 (* (sqrt x) 3.0)) (t_1 (* (- (+ (/ 1.0 (* 9.0 x)) y) 1.0) t_0)))
   (if (<= t_1 -40000.0)
     (* (fma 3.0 y -3.0) (sqrt x))
     (if (<= t_1 2e+149)
       (* (/ (fma -3.0 x 0.3333333333333333) x) (sqrt x))
       (* (- y 1.0) t_0)))))

C

double code(double x, double y) {
	double t_0 = sqrt(x) * 3.0;
	double t_1 = (((1.0 / (9.0 * x)) + y) - 1.0) * t_0;
	double tmp;
	if (t_1 <= -40000.0) {
		tmp = fma(3.0, y, -3.0) * sqrt(x);
	} else if (t_1 <= 2e+149) {
		tmp = (fma(-3.0, x, 0.3333333333333333) / x) * sqrt(x);
	} else {
		tmp = (y - 1.0) * t_0;
	}
	return tmp;
}

Julia

function code(x, y)
	t_0 = Float64(sqrt(x) * 3.0)
	t_1 = Float64(Float64(Float64(Float64(1.0 / Float64(9.0 * x)) + y) - 1.0) * t_0)
	tmp = 0.0
	if (t_1 <= -40000.0)
		tmp = Float64(fma(3.0, y, -3.0) * sqrt(x));
	elseif (t_1 <= 2e+149)
		tmp = Float64(Float64(fma(-3.0, x, 0.3333333333333333) / x) * sqrt(x));
	else
		tmp = Float64(Float64(y - 1.0) * t_0);
	end
	return tmp
end

Wolfram

code[x_, y_] := Block[{t$95$0 = N[(N[Sqrt[x], $MachinePrecision] * 3.0), $MachinePrecision]}, Block[{t$95$1 = N[(N[(N[(N[(1.0 / N[(9.0 * x), $MachinePrecision]), $MachinePrecision] + y), $MachinePrecision] - 1.0), $MachinePrecision] * t$95$0), $MachinePrecision]}, If[LessEqual[t$95$1, -40000.0], N[(N[(3.0 * y + -3.0), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], If[LessEqual[t$95$1, 2e+149], N[(N[(N[(-3.0 * x + 0.3333333333333333), $MachinePrecision] / x), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], N[(N[(y - 1.0), $MachinePrecision] * t$95$0), $MachinePrecision]]]]]

Derivation

1. Split input into 3 regimes

2. if (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64))) < -4e4

   1. Initial program 99.5%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in x around inf

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(y - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot \left(y - 1\right)\right) \cdot 3} \]

      2. associate-*l* N/A

         \[\leadsto \color{blue}{\sqrt{x} \cdot \left(\left(y - 1\right) \cdot 3\right)} \]

      3. *-commutative N/A

         \[\leadsto \color{blue}{\left(\left(y - 1\right) \cdot 3\right) \cdot \sqrt{x}} \]

      4. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\left(y - 1\right) \cdot 3\right) \cdot \sqrt{x}} \]

      5. *-commutative N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(y - 1\right)\right)} \cdot \sqrt{x} \]

      6. sub-neg N/A

         \[\leadsto \left(3 \cdot \color{blue}{\left(y + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

      7. metadata-eval N/A

         \[\leadsto \left(3 \cdot \left(y + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

      8. distribute-lft-in N/A

         \[\leadsto \color{blue}{\left(3 \cdot y + 3 \cdot -1\right)} \cdot \sqrt{x} \]

      9. metadata-eval N/A

         \[\leadsto \left(3 \cdot y + \color{blue}{-3}\right) \cdot \sqrt{x} \]

      10. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(3, y, -3\right)} \cdot \sqrt{x} \]

      11. lower-sqrt.f64 98.6

          \[\leadsto \mathsf{fma}\left(3, y, -3\right) \cdot \color{blue}{\sqrt{x}} \]

   5. Applied rewrites 98.6%

      \[\leadsto \color{blue}{\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}} \]

   if -4e4 < (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64))) < 2.0000000000000001e149

   1. Initial program 99.3%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around 0

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto 3 \cdot \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \cdot \sqrt{x}\right)} \]

      2. associate-*r* N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      4. sub-neg N/A

         \[\leadsto \left(3 \cdot \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

      5. metadata-eval N/A

         \[\leadsto \left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

      6. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -1 \cdot 3\right)} \cdot \sqrt{x} \]

      7. metadata-eval N/A

         \[\leadsto \left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + \color{blue}{-3}\right) \cdot \sqrt{x} \]

      8. lower-+.f64 N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -3\right)} \cdot \sqrt{x} \]

      9. associate-*r/ N/A

         \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 1}{x}} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      10. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{9}}}{x} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      11. associate-*l/ N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 3}{x}} + -3\right) \cdot \sqrt{x} \]

      12. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{3}}}{x} + -3\right) \cdot \sqrt{x} \]

      13. lower-/.f64 N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{3}}{x}} + -3\right) \cdot \sqrt{x} \]

      14. lower-sqrt.f64 86.6

          \[\leadsto \left(\frac{0.3333333333333333}{x} + -3\right) \cdot \color{blue}{\sqrt{x}} \]

   5. Applied rewrites 86.6%

      \[\leadsto \color{blue}{\left(\frac{0.3333333333333333}{x} + -3\right) \cdot \sqrt{x}} \]

   6. Taylor expanded in x around 0

      \[\leadsto \frac{\frac{1}{3} + -3 \cdot x}{x} \cdot \sqrt{\color{blue}{x}} \]
   7. Step-by-step derivation

   8. Applied rewrites 86.6%

      \[\leadsto \frac{\mathsf{fma}\left(-3, x, 0.3333333333333333\right)}{x} \cdot \sqrt{\color{blue}{x}} \]

   if 2.0000000000000001e149 < (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64)))

   1. Initial program 99.4%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in x around inf

      \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]
   4. Step-by-step derivation

      1. lower--.f64 93.0

         \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]

   5. Applied rewrites 93.0%

      \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]
3. Recombined 3 regimes into one program.

4. Final simplification 92.7%

   \[\leadsto \begin{array}{l} \mathbf{if}\;\left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot \left(\sqrt{x} \cdot 3\right) \leq -40000:\\ \;\;\;\;\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}\\ \mathbf{elif}\;\left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot \left(\sqrt{x} \cdot 3\right) \leq 2 \cdot 10^{+149}:\\ \;\;\;\;\frac{\mathsf{fma}\left(-3, x, 0.3333333333333333\right)}{x} \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;\left(y - 1\right) \cdot \left(\sqrt{x} \cdot 3\right)\\ \end{array} \]
5. Add Preprocessing

Alternative 3: 91.6% accurate, 0.3× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \sqrt{x} \cdot 3\\ t_1 := \left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot t\_0\\ \mathbf{if}\;t\_1 \leq -40000:\\ \;\;\;\;\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}\\ \mathbf{elif}\;t\_1 \leq 2 \cdot 10^{+149}:\\ \;\;\;\;\left(\frac{0.3333333333333333}{x} + -3\right) \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;\left(y - 1\right) \cdot t\_0\\ \end{array} \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (let* ((t_0 (* (sqrt x) 3.0)) (t_1 (* (- (+ (/ 1.0 (* 9.0 x)) y) 1.0) t_0)))
   (if (<= t_1 -40000.0)
     (* (fma 3.0 y -3.0) (sqrt x))
     (if (<= t_1 2e+149)
       (* (+ (/ 0.3333333333333333 x) -3.0) (sqrt x))
       (* (- y 1.0) t_0)))))

C

double code(double x, double y) {
	double t_0 = sqrt(x) * 3.0;
	double t_1 = (((1.0 / (9.0 * x)) + y) - 1.0) * t_0;
	double tmp;
	if (t_1 <= -40000.0) {
		tmp = fma(3.0, y, -3.0) * sqrt(x);
	} else if (t_1 <= 2e+149) {
		tmp = ((0.3333333333333333 / x) + -3.0) * sqrt(x);
	} else {
		tmp = (y - 1.0) * t_0;
	}
	return tmp;
}

Julia

function code(x, y)
	t_0 = Float64(sqrt(x) * 3.0)
	t_1 = Float64(Float64(Float64(Float64(1.0 / Float64(9.0 * x)) + y) - 1.0) * t_0)
	tmp = 0.0
	if (t_1 <= -40000.0)
		tmp = Float64(fma(3.0, y, -3.0) * sqrt(x));
	elseif (t_1 <= 2e+149)
		tmp = Float64(Float64(Float64(0.3333333333333333 / x) + -3.0) * sqrt(x));
	else
		tmp = Float64(Float64(y - 1.0) * t_0);
	end
	return tmp
end

Wolfram

code[x_, y_] := Block[{t$95$0 = N[(N[Sqrt[x], $MachinePrecision] * 3.0), $MachinePrecision]}, Block[{t$95$1 = N[(N[(N[(N[(1.0 / N[(9.0 * x), $MachinePrecision]), $MachinePrecision] + y), $MachinePrecision] - 1.0), $MachinePrecision] * t$95$0), $MachinePrecision]}, If[LessEqual[t$95$1, -40000.0], N[(N[(3.0 * y + -3.0), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], If[LessEqual[t$95$1, 2e+149], N[(N[(N[(0.3333333333333333 / x), $MachinePrecision] + -3.0), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], N[(N[(y - 1.0), $MachinePrecision] * t$95$0), $MachinePrecision]]]]]

Derivation

1. Split input into 3 regimes

2. if (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64))) < -4e4

   1. Initial program 99.5%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in x around inf

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(y - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot \left(y - 1\right)\right) \cdot 3} \]

      2. associate-*l* N/A

         \[\leadsto \color{blue}{\sqrt{x} \cdot \left(\left(y - 1\right) \cdot 3\right)} \]

      3. *-commutative N/A

         \[\leadsto \color{blue}{\left(\left(y - 1\right) \cdot 3\right) \cdot \sqrt{x}} \]

      4. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\left(y - 1\right) \cdot 3\right) \cdot \sqrt{x}} \]

      5. *-commutative N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(y - 1\right)\right)} \cdot \sqrt{x} \]

      6. sub-neg N/A

         \[\leadsto \left(3 \cdot \color{blue}{\left(y + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

      7. metadata-eval N/A

         \[\leadsto \left(3 \cdot \left(y + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

      8. distribute-lft-in N/A

         \[\leadsto \color{blue}{\left(3 \cdot y + 3 \cdot -1\right)} \cdot \sqrt{x} \]

      9. metadata-eval N/A

         \[\leadsto \left(3 \cdot y + \color{blue}{-3}\right) \cdot \sqrt{x} \]

      10. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(3, y, -3\right)} \cdot \sqrt{x} \]

      11. lower-sqrt.f64 98.6

          \[\leadsto \mathsf{fma}\left(3, y, -3\right) \cdot \color{blue}{\sqrt{x}} \]

   5. Applied rewrites 98.6%

      \[\leadsto \color{blue}{\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}} \]

   if -4e4 < (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64))) < 2.0000000000000001e149

   1. Initial program 99.3%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around 0

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto 3 \cdot \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \cdot \sqrt{x}\right)} \]

      2. associate-*r* N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      4. sub-neg N/A

         \[\leadsto \left(3 \cdot \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

      5. metadata-eval N/A

         \[\leadsto \left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

      6. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -1 \cdot 3\right)} \cdot \sqrt{x} \]

      7. metadata-eval N/A

         \[\leadsto \left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + \color{blue}{-3}\right) \cdot \sqrt{x} \]

      8. lower-+.f64 N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -3\right)} \cdot \sqrt{x} \]

      9. associate-*r/ N/A

         \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 1}{x}} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      10. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{9}}}{x} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      11. associate-*l/ N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 3}{x}} + -3\right) \cdot \sqrt{x} \]

      12. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{3}}}{x} + -3\right) \cdot \sqrt{x} \]

      13. lower-/.f64 N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{3}}{x}} + -3\right) \cdot \sqrt{x} \]

      14. lower-sqrt.f64 86.6

          \[\leadsto \left(\frac{0.3333333333333333}{x} + -3\right) \cdot \color{blue}{\sqrt{x}} \]

   5. Applied rewrites 86.6%

      \[\leadsto \color{blue}{\left(\frac{0.3333333333333333}{x} + -3\right) \cdot \sqrt{x}} \]

   if 2.0000000000000001e149 < (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64)))

   1. Initial program 99.4%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in x around inf

      \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]
   4. Step-by-step derivation

      1. lower--.f64 93.0

         \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]

   5. Applied rewrites 93.0%

      \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]
3. Recombined 3 regimes into one program.

4. Final simplification 92.7%

   \[\leadsto \begin{array}{l} \mathbf{if}\;\left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot \left(\sqrt{x} \cdot 3\right) \leq -40000:\\ \;\;\;\;\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}\\ \mathbf{elif}\;\left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot \left(\sqrt{x} \cdot 3\right) \leq 2 \cdot 10^{+149}:\\ \;\;\;\;\left(\frac{0.3333333333333333}{x} + -3\right) \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;\left(y - 1\right) \cdot \left(\sqrt{x} \cdot 3\right)\\ \end{array} \]
5. Add Preprocessing

Alternative 4: 90.7% accurate, 0.3× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \sqrt{x} \cdot 3\\ t_1 := \left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot t\_0\\ \mathbf{if}\;t\_1 \leq -5:\\ \;\;\;\;\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}\\ \mathbf{elif}\;t\_1 \leq 2 \cdot 10^{+149}:\\ \;\;\;\;\frac{0.3333333333333333}{x} \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;\left(y - 1\right) \cdot t\_0\\ \end{array} \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (let* ((t_0 (* (sqrt x) 3.0)) (t_1 (* (- (+ (/ 1.0 (* 9.0 x)) y) 1.0) t_0)))
   (if (<= t_1 -5.0)
     (* (fma 3.0 y -3.0) (sqrt x))
     (if (<= t_1 2e+149)
       (* (/ 0.3333333333333333 x) (sqrt x))
       (* (- y 1.0) t_0)))))

C

double code(double x, double y) {
	double t_0 = sqrt(x) * 3.0;
	double t_1 = (((1.0 / (9.0 * x)) + y) - 1.0) * t_0;
	double tmp;
	if (t_1 <= -5.0) {
		tmp = fma(3.0, y, -3.0) * sqrt(x);
	} else if (t_1 <= 2e+149) {
		tmp = (0.3333333333333333 / x) * sqrt(x);
	} else {
		tmp = (y - 1.0) * t_0;
	}
	return tmp;
}

Julia

function code(x, y)
	t_0 = Float64(sqrt(x) * 3.0)
	t_1 = Float64(Float64(Float64(Float64(1.0 / Float64(9.0 * x)) + y) - 1.0) * t_0)
	tmp = 0.0
	if (t_1 <= -5.0)
		tmp = Float64(fma(3.0, y, -3.0) * sqrt(x));
	elseif (t_1 <= 2e+149)
		tmp = Float64(Float64(0.3333333333333333 / x) * sqrt(x));
	else
		tmp = Float64(Float64(y - 1.0) * t_0);
	end
	return tmp
end

Wolfram

code[x_, y_] := Block[{t$95$0 = N[(N[Sqrt[x], $MachinePrecision] * 3.0), $MachinePrecision]}, Block[{t$95$1 = N[(N[(N[(N[(1.0 / N[(9.0 * x), $MachinePrecision]), $MachinePrecision] + y), $MachinePrecision] - 1.0), $MachinePrecision] * t$95$0), $MachinePrecision]}, If[LessEqual[t$95$1, -5.0], N[(N[(3.0 * y + -3.0), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], If[LessEqual[t$95$1, 2e+149], N[(N[(0.3333333333333333 / x), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], N[(N[(y - 1.0), $MachinePrecision] * t$95$0), $MachinePrecision]]]]]

Derivation

1. Split input into 3 regimes

2. if (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64))) < -5

   1. Initial program 99.5%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in x around inf

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(y - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot \left(y - 1\right)\right) \cdot 3} \]

      2. associate-*l* N/A

         \[\leadsto \color{blue}{\sqrt{x} \cdot \left(\left(y - 1\right) \cdot 3\right)} \]

      3. *-commutative N/A

         \[\leadsto \color{blue}{\left(\left(y - 1\right) \cdot 3\right) \cdot \sqrt{x}} \]

      4. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\left(y - 1\right) \cdot 3\right) \cdot \sqrt{x}} \]

      5. *-commutative N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(y - 1\right)\right)} \cdot \sqrt{x} \]

      6. sub-neg N/A

         \[\leadsto \left(3 \cdot \color{blue}{\left(y + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

      7. metadata-eval N/A

         \[\leadsto \left(3 \cdot \left(y + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

      8. distribute-lft-in N/A

         \[\leadsto \color{blue}{\left(3 \cdot y + 3 \cdot -1\right)} \cdot \sqrt{x} \]

      9. metadata-eval N/A

         \[\leadsto \left(3 \cdot y + \color{blue}{-3}\right) \cdot \sqrt{x} \]

      10. lower-fma.f64 N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(3, y, -3\right)} \cdot \sqrt{x} \]

      11. lower-sqrt.f64 98.0

          \[\leadsto \mathsf{fma}\left(3, y, -3\right) \cdot \color{blue}{\sqrt{x}} \]

   5. Applied rewrites 98.0%

      \[\leadsto \color{blue}{\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}} \]

   if -5 < (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64))) < 2.0000000000000001e149

   1. Initial program 99.3%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around 0

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot y\right) + 3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} + 3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \]

      2. associate-*l* N/A

         \[\leadsto \color{blue}{\sqrt{x} \cdot \left(y \cdot 3\right)} + 3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \]

      3. associate-*r* N/A

         \[\leadsto \sqrt{x} \cdot \left(y \cdot 3\right) + \color{blue}{\left(3 \cdot \sqrt{x}\right) \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)} \]

      4. *-commutative N/A

         \[\leadsto \sqrt{x} \cdot \left(y \cdot 3\right) + \color{blue}{\left(\sqrt{x} \cdot 3\right)} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \]

      5. associate-*l* N/A

         \[\leadsto \sqrt{x} \cdot \left(y \cdot 3\right) + \color{blue}{\sqrt{x} \cdot \left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]

      6. distribute-lft-out N/A

         \[\leadsto \color{blue}{\sqrt{x} \cdot \left(y \cdot 3 + 3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]

      7. lower-*.f64 N/A

         \[\leadsto \color{blue}{\sqrt{x} \cdot \left(y \cdot 3 + 3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]

      8. lower-sqrt.f64 N/A

         \[\leadsto \color{blue}{\sqrt{x}} \cdot \left(y \cdot 3 + 3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \]

      9. *-commutative N/A

         \[\leadsto \sqrt{x} \cdot \left(y \cdot 3 + \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \cdot 3}\right) \]

      10. distribute-rgt-in N/A

          \[\leadsto \sqrt{x} \cdot \color{blue}{\left(3 \cdot \left(y + \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)\right)} \]

      11. associate--l+ N/A

          \[\leadsto \sqrt{x} \cdot \left(3 \cdot \color{blue}{\left(\left(y + \frac{1}{9} \cdot \frac{1}{x}\right) - 1\right)}\right) \]

      12. +-commutative N/A

          \[\leadsto \sqrt{x} \cdot \left(3 \cdot \left(\color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + y\right)} - 1\right)\right) \]

      13. associate--l+ N/A

          \[\leadsto \sqrt{x} \cdot \left(3 \cdot \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + \left(y - 1\right)\right)}\right) \]

      14. +-commutative N/A

          \[\leadsto \sqrt{x} \cdot \left(3 \cdot \color{blue}{\left(\left(y - 1\right) + \frac{1}{9} \cdot \frac{1}{x}\right)}\right) \]

      15. distribute-rgt-in N/A

          \[\leadsto \sqrt{x} \cdot \color{blue}{\left(\left(y - 1\right) \cdot 3 + \left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3\right)} \]

   5. Applied rewrites 99.4%

      \[\leadsto \color{blue}{\sqrt{x} \cdot \mathsf{fma}\left(1 - y, -3, \frac{0.3333333333333333}{x}\right)} \]

   6. Taylor expanded in x around 0

      \[\leadsto \sqrt{x} \cdot \frac{\frac{1}{3}}{\color{blue}{x}} \]
   7. Step-by-step derivation

   8. Applied rewrites 86.5%

      \[\leadsto \sqrt{x} \cdot \frac{0.3333333333333333}{\color{blue}{x}} \]

   if 2.0000000000000001e149 < (*.f64 (*.f64 #s(literal 3 binary64) (sqrt.f64 x)) (-.f64 (+.f64 y (/.f64 #s(literal 1 binary64) (*.f64 x #s(literal 9 binary64)))) #s(literal 1 binary64)))

   1. Initial program 99.4%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in x around inf

      \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]
   4. Step-by-step derivation

      1. lower--.f64 93.0

         \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]

   5. Applied rewrites 93.0%

      \[\leadsto \left(3 \cdot \sqrt{x}\right) \cdot \color{blue}{\left(y - 1\right)} \]
3. Recombined 3 regimes into one program.

4. Final simplification 92.4%

   \[\leadsto \begin{array}{l} \mathbf{if}\;\left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot \left(\sqrt{x} \cdot 3\right) \leq -5:\\ \;\;\;\;\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}\\ \mathbf{elif}\;\left(\left(\frac{1}{9 \cdot x} + y\right) - 1\right) \cdot \left(\sqrt{x} \cdot 3\right) \leq 2 \cdot 10^{+149}:\\ \;\;\;\;\frac{0.3333333333333333}{x} \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;\left(y - 1\right) \cdot \left(\sqrt{x} \cdot 3\right)\\ \end{array} \]
5. Add Preprocessing

Alternative 5: 61.0% accurate, 1.3× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(\sqrt{x} \cdot y\right) \cdot 3\\ \mathbf{if}\;y \leq -1:\\ \;\;\;\;t\_0\\ \mathbf{elif}\;y \leq 7.8 \cdot 10^{-31}:\\ \;\;\;\;-3 \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;t\_0\\ \end{array} \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (let* ((t_0 (* (* (sqrt x) y) 3.0)))
   (if (<= y -1.0) t_0 (if (<= y 7.8e-31) (* -3.0 (sqrt x)) t_0))))

C

double code(double x, double y) {
	double t_0 = (sqrt(x) * y) * 3.0;
	double tmp;
	if (y <= -1.0) {
		tmp = t_0;
	} else if (y <= 7.8e-31) {
		tmp = -3.0 * sqrt(x);
	} else {
		tmp = t_0;
	}
	return tmp;
}

Fortran

real(8) function code(x, y)
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8) :: t_0
    real(8) :: tmp
    t_0 = (sqrt(x) * y) * 3.0d0
    if (y <= (-1.0d0)) then
        tmp = t_0
    else if (y <= 7.8d-31) then
        tmp = (-3.0d0) * sqrt(x)
    else
        tmp = t_0
    end if
    code = tmp
end function

Java

public static double code(double x, double y) {
	double t_0 = (Math.sqrt(x) * y) * 3.0;
	double tmp;
	if (y <= -1.0) {
		tmp = t_0;
	} else if (y <= 7.8e-31) {
		tmp = -3.0 * Math.sqrt(x);
	} else {
		tmp = t_0;
	}
	return tmp;
}

Python

def code(x, y):
	t_0 = (math.sqrt(x) * y) * 3.0
	tmp = 0
	if y <= -1.0:
		tmp = t_0
	elif y <= 7.8e-31:
		tmp = -3.0 * math.sqrt(x)
	else:
		tmp = t_0
	return tmp

Julia

function code(x, y)
	t_0 = Float64(Float64(sqrt(x) * y) * 3.0)
	tmp = 0.0
	if (y <= -1.0)
		tmp = t_0;
	elseif (y <= 7.8e-31)
		tmp = Float64(-3.0 * sqrt(x));
	else
		tmp = t_0;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y)
	t_0 = (sqrt(x) * y) * 3.0;
	tmp = 0.0;
	if (y <= -1.0)
		tmp = t_0;
	elseif (y <= 7.8e-31)
		tmp = -3.0 * sqrt(x);
	else
		tmp = t_0;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_] := Block[{t$95$0 = N[(N[(N[Sqrt[x], $MachinePrecision] * y), $MachinePrecision] * 3.0), $MachinePrecision]}, If[LessEqual[y, -1.0], t$95$0, If[LessEqual[y, 7.8e-31], N[(-3.0 * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], t$95$0]]]

Derivation

1. Split input into 2 regimes

2. if y < -1 or 7.8000000000000003e-31 < y

   1. Initial program 99.5%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around inf

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

      1. *-commutative N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} \]

      2. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right)} \cdot 3 \]

      4. lower-sqrt.f64 69.3

         \[\leadsto \left(\color{blue}{\sqrt{x}} \cdot y\right) \cdot 3 \]

   5. Applied rewrites 69.3%

      \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} \]

   if -1 < y < 7.8000000000000003e-31

   1. Initial program 99.4%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around 0

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto 3 \cdot \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \cdot \sqrt{x}\right)} \]

      2. associate-*r* N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      4. sub-neg N/A

         \[\leadsto \left(3 \cdot \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

      5. metadata-eval N/A

         \[\leadsto \left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

      6. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -1 \cdot 3\right)} \cdot \sqrt{x} \]

      7. metadata-eval N/A

         \[\leadsto \left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + \color{blue}{-3}\right) \cdot \sqrt{x} \]

      8. lower-+.f64 N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -3\right)} \cdot \sqrt{x} \]

      9. associate-*r/ N/A

         \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 1}{x}} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      10. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{9}}}{x} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      11. associate-*l/ N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 3}{x}} + -3\right) \cdot \sqrt{x} \]

      12. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{3}}}{x} + -3\right) \cdot \sqrt{x} \]

      13. lower-/.f64 N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{3}}{x}} + -3\right) \cdot \sqrt{x} \]

      14. lower-sqrt.f64 98.6

          \[\leadsto \left(\frac{0.3333333333333333}{x} + -3\right) \cdot \color{blue}{\sqrt{x}} \]

   5. Applied rewrites 98.6%

      \[\leadsto \color{blue}{\left(\frac{0.3333333333333333}{x} + -3\right) \cdot \sqrt{x}} \]

   6. Taylor expanded in x around inf

      \[\leadsto -3 \cdot \sqrt{\color{blue}{x}} \]
   7. Step-by-step derivation

   8. Applied rewrites 48.8%

      \[\leadsto -3 \cdot \sqrt{\color{blue}{x}} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 6: 61.0% accurate, 1.3× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(\sqrt{x} \cdot 3\right) \cdot y\\ \mathbf{if}\;y \leq -1:\\ \;\;\;\;t\_0\\ \mathbf{elif}\;y \leq 7.8 \cdot 10^{-31}:\\ \;\;\;\;-3 \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;t\_0\\ \end{array} \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (let* ((t_0 (* (* (sqrt x) 3.0) y)))
   (if (<= y -1.0) t_0 (if (<= y 7.8e-31) (* -3.0 (sqrt x)) t_0))))

C

double code(double x, double y) {
	double t_0 = (sqrt(x) * 3.0) * y;
	double tmp;
	if (y <= -1.0) {
		tmp = t_0;
	} else if (y <= 7.8e-31) {
		tmp = -3.0 * sqrt(x);
	} else {
		tmp = t_0;
	}
	return tmp;
}

Fortran

real(8) function code(x, y)
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8) :: t_0
    real(8) :: tmp
    t_0 = (sqrt(x) * 3.0d0) * y
    if (y <= (-1.0d0)) then
        tmp = t_0
    else if (y <= 7.8d-31) then
        tmp = (-3.0d0) * sqrt(x)
    else
        tmp = t_0
    end if
    code = tmp
end function

Java

public static double code(double x, double y) {
	double t_0 = (Math.sqrt(x) * 3.0) * y;
	double tmp;
	if (y <= -1.0) {
		tmp = t_0;
	} else if (y <= 7.8e-31) {
		tmp = -3.0 * Math.sqrt(x);
	} else {
		tmp = t_0;
	}
	return tmp;
}

Python

def code(x, y):
	t_0 = (math.sqrt(x) * 3.0) * y
	tmp = 0
	if y <= -1.0:
		tmp = t_0
	elif y <= 7.8e-31:
		tmp = -3.0 * math.sqrt(x)
	else:
		tmp = t_0
	return tmp

Julia

function code(x, y)
	t_0 = Float64(Float64(sqrt(x) * 3.0) * y)
	tmp = 0.0
	if (y <= -1.0)
		tmp = t_0;
	elseif (y <= 7.8e-31)
		tmp = Float64(-3.0 * sqrt(x));
	else
		tmp = t_0;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y)
	t_0 = (sqrt(x) * 3.0) * y;
	tmp = 0.0;
	if (y <= -1.0)
		tmp = t_0;
	elseif (y <= 7.8e-31)
		tmp = -3.0 * sqrt(x);
	else
		tmp = t_0;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_] := Block[{t$95$0 = N[(N[(N[Sqrt[x], $MachinePrecision] * 3.0), $MachinePrecision] * y), $MachinePrecision]}, If[LessEqual[y, -1.0], t$95$0, If[LessEqual[y, 7.8e-31], N[(-3.0 * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], t$95$0]]]

Derivation

1. Split input into 2 regimes

2. if y < -1 or 7.8000000000000003e-31 < y

   1. Initial program 99.5%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around inf

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

      1. *-commutative N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} \]

      2. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right)} \cdot 3 \]

      4. lower-sqrt.f64 69.3

         \[\leadsto \left(\color{blue}{\sqrt{x}} \cdot y\right) \cdot 3 \]

   5. Applied rewrites 69.3%

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

   7. Applied rewrites 69.3%

      \[\leadsto \left(\sqrt{x} \cdot 3\right) \cdot \color{blue}{y} \]

   if -1 < y < 7.8000000000000003e-31

   1. Initial program 99.4%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around 0

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto 3 \cdot \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \cdot \sqrt{x}\right)} \]

      2. associate-*r* N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      4. sub-neg N/A

         \[\leadsto \left(3 \cdot \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

      5. metadata-eval N/A

         \[\leadsto \left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

      6. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -1 \cdot 3\right)} \cdot \sqrt{x} \]

      7. metadata-eval N/A

         \[\leadsto \left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + \color{blue}{-3}\right) \cdot \sqrt{x} \]

      8. lower-+.f64 N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -3\right)} \cdot \sqrt{x} \]

      9. associate-*r/ N/A

         \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 1}{x}} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      10. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{9}}}{x} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      11. associate-*l/ N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 3}{x}} + -3\right) \cdot \sqrt{x} \]

      12. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{3}}}{x} + -3\right) \cdot \sqrt{x} \]

      13. lower-/.f64 N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{3}}{x}} + -3\right) \cdot \sqrt{x} \]

      14. lower-sqrt.f64 98.6

          \[\leadsto \left(\frac{0.3333333333333333}{x} + -3\right) \cdot \color{blue}{\sqrt{x}} \]

   5. Applied rewrites 98.6%

      \[\leadsto \color{blue}{\left(\frac{0.3333333333333333}{x} + -3\right) \cdot \sqrt{x}} \]

   6. Taylor expanded in x around inf

      \[\leadsto -3 \cdot \sqrt{\color{blue}{x}} \]
   7. Step-by-step derivation

   8. Applied rewrites 48.8%

      \[\leadsto -3 \cdot \sqrt{\color{blue}{x}} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 7: 61.0% accurate, 1.3× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(y \cdot 3\right) \cdot \sqrt{x}\\ \mathbf{if}\;y \leq -1:\\ \;\;\;\;t\_0\\ \mathbf{elif}\;y \leq 7.8 \cdot 10^{-31}:\\ \;\;\;\;-3 \cdot \sqrt{x}\\ \mathbf{else}:\\ \;\;\;\;t\_0\\ \end{array} \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (let* ((t_0 (* (* y 3.0) (sqrt x))))
   (if (<= y -1.0) t_0 (if (<= y 7.8e-31) (* -3.0 (sqrt x)) t_0))))

C

double code(double x, double y) {
	double t_0 = (y * 3.0) * sqrt(x);
	double tmp;
	if (y <= -1.0) {
		tmp = t_0;
	} else if (y <= 7.8e-31) {
		tmp = -3.0 * sqrt(x);
	} else {
		tmp = t_0;
	}
	return tmp;
}

Fortran

real(8) function code(x, y)
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    real(8) :: t_0
    real(8) :: tmp
    t_0 = (y * 3.0d0) * sqrt(x)
    if (y <= (-1.0d0)) then
        tmp = t_0
    else if (y <= 7.8d-31) then
        tmp = (-3.0d0) * sqrt(x)
    else
        tmp = t_0
    end if
    code = tmp
end function

Java

public static double code(double x, double y) {
	double t_0 = (y * 3.0) * Math.sqrt(x);
	double tmp;
	if (y <= -1.0) {
		tmp = t_0;
	} else if (y <= 7.8e-31) {
		tmp = -3.0 * Math.sqrt(x);
	} else {
		tmp = t_0;
	}
	return tmp;
}

Python

def code(x, y):
	t_0 = (y * 3.0) * math.sqrt(x)
	tmp = 0
	if y <= -1.0:
		tmp = t_0
	elif y <= 7.8e-31:
		tmp = -3.0 * math.sqrt(x)
	else:
		tmp = t_0
	return tmp

Julia

function code(x, y)
	t_0 = Float64(Float64(y * 3.0) * sqrt(x))
	tmp = 0.0
	if (y <= -1.0)
		tmp = t_0;
	elseif (y <= 7.8e-31)
		tmp = Float64(-3.0 * sqrt(x));
	else
		tmp = t_0;
	end
	return tmp
end

MATLAB

function tmp_2 = code(x, y)
	t_0 = (y * 3.0) * sqrt(x);
	tmp = 0.0;
	if (y <= -1.0)
		tmp = t_0;
	elseif (y <= 7.8e-31)
		tmp = -3.0 * sqrt(x);
	else
		tmp = t_0;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_, y_] := Block[{t$95$0 = N[(N[(y * 3.0), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision]}, If[LessEqual[y, -1.0], t$95$0, If[LessEqual[y, 7.8e-31], N[(-3.0 * N[Sqrt[x], $MachinePrecision]), $MachinePrecision], t$95$0]]]

Derivation

1. Split input into 2 regimes

2. if y < -1 or 7.8000000000000003e-31 < y

   1. Initial program 99.5%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around inf

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

      1. *-commutative N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} \]

      2. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right) \cdot 3} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(\sqrt{x} \cdot y\right)} \cdot 3 \]

      4. lower-sqrt.f64 69.3

         \[\leadsto \left(\color{blue}{\sqrt{x}} \cdot y\right) \cdot 3 \]

   5. Applied rewrites 69.3%

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

   7. Applied rewrites 69.2%

      \[\leadsto \left(y \cdot 3\right) \cdot \color{blue}{\sqrt{x}} \]

   if -1 < y < 7.8000000000000003e-31

   1. Initial program 99.4%

      \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
   2. Add Preprocessing

   3. Taylor expanded in y around 0

      \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]
   4. Step-by-step derivation

      1. *-commutative N/A

         \[\leadsto 3 \cdot \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \cdot \sqrt{x}\right)} \]

      2. associate-*r* N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

      4. sub-neg N/A

         \[\leadsto \left(3 \cdot \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

      5. metadata-eval N/A

         \[\leadsto \left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

      6. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -1 \cdot 3\right)} \cdot \sqrt{x} \]

      7. metadata-eval N/A

         \[\leadsto \left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + \color{blue}{-3}\right) \cdot \sqrt{x} \]

      8. lower-+.f64 N/A

         \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -3\right)} \cdot \sqrt{x} \]

      9. associate-*r/ N/A

         \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 1}{x}} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      10. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{9}}}{x} \cdot 3 + -3\right) \cdot \sqrt{x} \]

      11. associate-*l/ N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 3}{x}} + -3\right) \cdot \sqrt{x} \]

      12. metadata-eval N/A

          \[\leadsto \left(\frac{\color{blue}{\frac{1}{3}}}{x} + -3\right) \cdot \sqrt{x} \]

      13. lower-/.f64 N/A

          \[\leadsto \left(\color{blue}{\frac{\frac{1}{3}}{x}} + -3\right) \cdot \sqrt{x} \]

      14. lower-sqrt.f64 98.6

          \[\leadsto \left(\frac{0.3333333333333333}{x} + -3\right) \cdot \color{blue}{\sqrt{x}} \]

   5. Applied rewrites 98.6%

      \[\leadsto \color{blue}{\left(\frac{0.3333333333333333}{x} + -3\right) \cdot \sqrt{x}} \]

   6. Taylor expanded in x around inf

      \[\leadsto -3 \cdot \sqrt{\color{blue}{x}} \]
   7. Step-by-step derivation

   8. Applied rewrites 48.8%

      \[\leadsto -3 \cdot \sqrt{\color{blue}{x}} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 8: 63.0% accurate, 2.0× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x} \end{array} \]

FPCore

(FPCore (x y) :precision binary64 (* (fma 3.0 y -3.0) (sqrt x)))

C

double code(double x, double y) {
	return fma(3.0, y, -3.0) * sqrt(x);
}

Julia

function code(x, y)
	return Float64(fma(3.0, y, -3.0) * sqrt(x))
end

Wolfram

code[x_, y_] := N[(N[(3.0 * y + -3.0), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 99.4%

   \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
2. Add Preprocessing

3. Taylor expanded in x around inf

   \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(y - 1\right)\right)} \]
4. Step-by-step derivation

   1. *-commutative N/A

      \[\leadsto \color{blue}{\left(\sqrt{x} \cdot \left(y - 1\right)\right) \cdot 3} \]

   2. associate-*l* N/A

      \[\leadsto \color{blue}{\sqrt{x} \cdot \left(\left(y - 1\right) \cdot 3\right)} \]

   3. *-commutative N/A

      \[\leadsto \color{blue}{\left(\left(y - 1\right) \cdot 3\right) \cdot \sqrt{x}} \]

   4. lower-*.f64 N/A

      \[\leadsto \color{blue}{\left(\left(y - 1\right) \cdot 3\right) \cdot \sqrt{x}} \]

   5. *-commutative N/A

      \[\leadsto \color{blue}{\left(3 \cdot \left(y - 1\right)\right)} \cdot \sqrt{x} \]

   6. sub-neg N/A

      \[\leadsto \left(3 \cdot \color{blue}{\left(y + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

   7. metadata-eval N/A

      \[\leadsto \left(3 \cdot \left(y + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

   8. distribute-lft-in N/A

      \[\leadsto \color{blue}{\left(3 \cdot y + 3 \cdot -1\right)} \cdot \sqrt{x} \]

   9. metadata-eval N/A

      \[\leadsto \left(3 \cdot y + \color{blue}{-3}\right) \cdot \sqrt{x} \]

   10. lower-fma.f64 N/A

       \[\leadsto \color{blue}{\mathsf{fma}\left(3, y, -3\right)} \cdot \sqrt{x} \]

   11. lower-sqrt.f64 60.9

       \[\leadsto \mathsf{fma}\left(3, y, -3\right) \cdot \color{blue}{\sqrt{x}} \]

5. Applied rewrites 60.9%

   \[\leadsto \color{blue}{\mathsf{fma}\left(3, y, -3\right) \cdot \sqrt{x}} \]
6. Add Preprocessing

Alternative 9: 25.5% accurate, 2.7× speedup

Math

\[\begin{array}{l} \\ -3 \cdot \sqrt{x} \end{array} \]

FPCore

(FPCore (x y) :precision binary64 (* -3.0 (sqrt x)))

C

double code(double x, double y) {
	return -3.0 * sqrt(x);
}

Fortran

real(8) function code(x, y)
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    code = (-3.0d0) * sqrt(x)
end function

Java

public static double code(double x, double y) {
	return -3.0 * Math.sqrt(x);
}

Python

def code(x, y):
	return -3.0 * math.sqrt(x)

Julia

function code(x, y)
	return Float64(-3.0 * sqrt(x))
end

MATLAB

function tmp = code(x, y)
	tmp = -3.0 * sqrt(x);
end

Wolfram

code[x_, y_] := N[(-3.0 * N[Sqrt[x], $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 99.4%

   \[\left(3 \cdot \sqrt{x}\right) \cdot \left(\left(y + \frac{1}{x \cdot 9}\right) - 1\right) \]
2. Add Preprocessing

3. Taylor expanded in y around 0

   \[\leadsto \color{blue}{3 \cdot \left(\sqrt{x} \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right)} \]
4. Step-by-step derivation

   1. *-commutative N/A

      \[\leadsto 3 \cdot \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x} - 1\right) \cdot \sqrt{x}\right)} \]

   2. associate-*r* N/A

      \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

   3. lower-*.f64 N/A

      \[\leadsto \color{blue}{\left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} - 1\right)\right) \cdot \sqrt{x}} \]

   4. sub-neg N/A

      \[\leadsto \left(3 \cdot \color{blue}{\left(\frac{1}{9} \cdot \frac{1}{x} + \left(\mathsf{neg}\left(1\right)\right)\right)}\right) \cdot \sqrt{x} \]

   5. metadata-eval N/A

      \[\leadsto \left(3 \cdot \left(\frac{1}{9} \cdot \frac{1}{x} + \color{blue}{-1}\right)\right) \cdot \sqrt{x} \]

   6. distribute-rgt-in N/A

      \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -1 \cdot 3\right)} \cdot \sqrt{x} \]

   7. metadata-eval N/A

      \[\leadsto \left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + \color{blue}{-3}\right) \cdot \sqrt{x} \]

   8. lower-+.f64 N/A

      \[\leadsto \color{blue}{\left(\left(\frac{1}{9} \cdot \frac{1}{x}\right) \cdot 3 + -3\right)} \cdot \sqrt{x} \]

   9. associate-*r/ N/A

      \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 1}{x}} \cdot 3 + -3\right) \cdot \sqrt{x} \]

   10. metadata-eval N/A

       \[\leadsto \left(\frac{\color{blue}{\frac{1}{9}}}{x} \cdot 3 + -3\right) \cdot \sqrt{x} \]

   11. associate-*l/ N/A

       \[\leadsto \left(\color{blue}{\frac{\frac{1}{9} \cdot 3}{x}} + -3\right) \cdot \sqrt{x} \]

   12. metadata-eval N/A

       \[\leadsto \left(\frac{\color{blue}{\frac{1}{3}}}{x} + -3\right) \cdot \sqrt{x} \]

   13. lower-/.f64 N/A

       \[\leadsto \left(\color{blue}{\frac{\frac{1}{3}}{x}} + -3\right) \cdot \sqrt{x} \]

   14. lower-sqrt.f64 62.6

       \[\leadsto \left(\frac{0.3333333333333333}{x} + -3\right) \cdot \color{blue}{\sqrt{x}} \]

5. Applied rewrites 62.6%

   \[\leadsto \color{blue}{\left(\frac{0.3333333333333333}{x} + -3\right) \cdot \sqrt{x}} \]

6. Taylor expanded in x around inf

   \[\leadsto -3 \cdot \sqrt{\color{blue}{x}} \]
7. Step-by-step derivation

8. Applied rewrites 24.8%

   \[\leadsto -3 \cdot \sqrt{\color{blue}{x}} \]
9. Add Preprocessing

Developer Target 1: 99.4% accurate, 0.7× speedup

Math

\[\begin{array}{l} \\ 3 \cdot \left(y \cdot \sqrt{x} + \left(\frac{1}{x \cdot 9} - 1\right) \cdot \sqrt{x}\right) \end{array} \]

FPCore

(FPCore (x y)
 :precision binary64
 (* 3.0 (+ (* y (sqrt x)) (* (- (/ 1.0 (* x 9.0)) 1.0) (sqrt x)))))

C

double code(double x, double y) {
	return 3.0 * ((y * sqrt(x)) + (((1.0 / (x * 9.0)) - 1.0) * sqrt(x)));
}

Fortran

real(8) function code(x, y)
    real(8), intent (in) :: x
    real(8), intent (in) :: y
    code = 3.0d0 * ((y * sqrt(x)) + (((1.0d0 / (x * 9.0d0)) - 1.0d0) * sqrt(x)))
end function

Java

public static double code(double x, double y) {
	return 3.0 * ((y * Math.sqrt(x)) + (((1.0 / (x * 9.0)) - 1.0) * Math.sqrt(x)));
}

Python

def code(x, y):
	return 3.0 * ((y * math.sqrt(x)) + (((1.0 / (x * 9.0)) - 1.0) * math.sqrt(x)))

Julia

function code(x, y)
	return Float64(3.0 * Float64(Float64(y * sqrt(x)) + Float64(Float64(Float64(1.0 / Float64(x * 9.0)) - 1.0) * sqrt(x))))
end

MATLAB

function tmp = code(x, y)
	tmp = 3.0 * ((y * sqrt(x)) + (((1.0 / (x * 9.0)) - 1.0) * sqrt(x)));
end

Wolfram

code[x_, y_] := N[(3.0 * N[(N[(y * N[Sqrt[x], $MachinePrecision]), $MachinePrecision] + N[(N[(N[(1.0 / N[(x * 9.0), $MachinePrecision]), $MachinePrecision] - 1.0), $MachinePrecision] * N[Sqrt[x], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Reproduce

herbie shell --seed 2024294 
(FPCore (x y)
  :name "Numeric.SpecFunctions:incompleteGamma from math-functions-0.1.5.2, B"
  :precision binary64

  :alt
  (! :herbie-platform default (* 3 (+ (* y (sqrt x)) (* (- (/ 1 (* x 9)) 1) (sqrt x)))))

  (* (* 3.0 (sqrt x)) (- (+ y (/ 1.0 (* x 9.0))) 1.0)))