Octave 3.8, oct_fill_randg

Percentage Accurate: 99.7% → 99.8%

Time: 9.3s

Alternatives: 10

Speedup: 2.4×

Specification

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := a - \frac{1}{3}\\ t\_0 \cdot \left(1 + \frac{1}{\sqrt{9 \cdot t\_0}} \cdot rand\right) \end{array} \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (let* ((t_0 (- a (/ 1.0 3.0))))
   (* t_0 (+ 1.0 (* (/ 1.0 (sqrt (* 9.0 t_0))) rand)))))

C

double code(double a, double rand) {
	double t_0 = a - (1.0 / 3.0);
	return t_0 * (1.0 + ((1.0 / sqrt((9.0 * t_0))) * rand));
}

Fortran

real(8) function code(a, rand)
    real(8), intent (in) :: a
    real(8), intent (in) :: rand
    real(8) :: t_0
    t_0 = a - (1.0d0 / 3.0d0)
    code = t_0 * (1.0d0 + ((1.0d0 / sqrt((9.0d0 * t_0))) * rand))
end function

Java

public static double code(double a, double rand) {
	double t_0 = a - (1.0 / 3.0);
	return t_0 * (1.0 + ((1.0 / Math.sqrt((9.0 * t_0))) * rand));
}

Python

def code(a, rand):
	t_0 = a - (1.0 / 3.0)
	return t_0 * (1.0 + ((1.0 / math.sqrt((9.0 * t_0))) * rand))

Julia

function code(a, rand)
	t_0 = Float64(a - Float64(1.0 / 3.0))
	return Float64(t_0 * Float64(1.0 + Float64(Float64(1.0 / sqrt(Float64(9.0 * t_0))) * rand)))
end

MATLAB

function tmp = code(a, rand)
	t_0 = a - (1.0 / 3.0);
	tmp = t_0 * (1.0 + ((1.0 / sqrt((9.0 * t_0))) * rand));
end

Wolfram

code[a_, rand_] := Block[{t$95$0 = N[(a - N[(1.0 / 3.0), $MachinePrecision]), $MachinePrecision]}, N[(t$95$0 * N[(1.0 + N[(N[(1.0 / N[Sqrt[N[(9.0 * t$95$0), $MachinePrecision]], $MachinePrecision]), $MachinePrecision] * rand), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Initial Program: 99.7% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := a - \frac{1}{3}\\ t\_0 \cdot \left(1 + \frac{1}{\sqrt{9 \cdot t\_0}} \cdot rand\right) \end{array} \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (let* ((t_0 (- a (/ 1.0 3.0))))
   (* t_0 (+ 1.0 (* (/ 1.0 (sqrt (* 9.0 t_0))) rand)))))

C

double code(double a, double rand) {
	double t_0 = a - (1.0 / 3.0);
	return t_0 * (1.0 + ((1.0 / sqrt((9.0 * t_0))) * rand));
}

Fortran

real(8) function code(a, rand)
    real(8), intent (in) :: a
    real(8), intent (in) :: rand
    real(8) :: t_0
    t_0 = a - (1.0d0 / 3.0d0)
    code = t_0 * (1.0d0 + ((1.0d0 / sqrt((9.0d0 * t_0))) * rand))
end function

Java

public static double code(double a, double rand) {
	double t_0 = a - (1.0 / 3.0);
	return t_0 * (1.0 + ((1.0 / Math.sqrt((9.0 * t_0))) * rand));
}

Python

def code(a, rand):
	t_0 = a - (1.0 / 3.0)
	return t_0 * (1.0 + ((1.0 / math.sqrt((9.0 * t_0))) * rand))

Julia

function code(a, rand)
	t_0 = Float64(a - Float64(1.0 / 3.0))
	return Float64(t_0 * Float64(1.0 + Float64(Float64(1.0 / sqrt(Float64(9.0 * t_0))) * rand)))
end

MATLAB

function tmp = code(a, rand)
	t_0 = a - (1.0 / 3.0);
	tmp = t_0 * (1.0 + ((1.0 / sqrt((9.0 * t_0))) * rand));
end

Wolfram

code[a_, rand_] := Block[{t$95$0 = N[(a - N[(1.0 / 3.0), $MachinePrecision]), $MachinePrecision]}, N[(t$95$0 * N[(1.0 + N[(N[(1.0 / N[Sqrt[N[(9.0 * t$95$0), $MachinePrecision]], $MachinePrecision]), $MachinePrecision] * rand), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]

Alternative 1: 99.8% accurate, 1.7× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(\frac{rand}{\sqrt{\mathsf{fma}\left(a, 9, -3\right)}}, a - 0.3333333333333333, -0.3333333333333333\right) + a \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (+
  (fma
   (/ rand (sqrt (fma a 9.0 -3.0)))
   (- a 0.3333333333333333)
   -0.3333333333333333)
  a))

C

double code(double a, double rand) {
	return fma((rand / sqrt(fma(a, 9.0, -3.0))), (a - 0.3333333333333333), -0.3333333333333333) + a;
}

Julia

function code(a, rand)
	return Float64(fma(Float64(rand / sqrt(fma(a, 9.0, -3.0))), Float64(a - 0.3333333333333333), -0.3333333333333333) + a)
end

Wolfram

code[a_, rand_] := N[(N[(N[(rand / N[Sqrt[N[(a * 9.0 + -3.0), $MachinePrecision]], $MachinePrecision]), $MachinePrecision] * N[(a - 0.3333333333333333), $MachinePrecision] + -0.3333333333333333), $MachinePrecision] + a), $MachinePrecision]

Derivation

1. Initial program 99.8%

   \[\left(a - \frac{1}{3}\right) \cdot \left(1 + \frac{1}{\sqrt{9 \cdot \left(a - \frac{1}{3}\right)}} \cdot rand\right) \]
2. Add Preprocessing
3. Step-by-step derivation

   1. lift-*.f64 N/A

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

   2. lift-+.f64 N/A

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

   3. +-commutative N/A

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

   4. distribute-lft-in N/A

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

   5. *-rgt-identity N/A

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

   6. lift--.f64 N/A

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

   7. sub-neg N/A

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

   8. +-commutative N/A

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

   9. associate-+r+ N/A

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

   10. lower-+.f64 N/A

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

4. Applied rewrites 99.9%

   \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{rand}{\sqrt{\mathsf{fma}\left(a, 9, -3\right)}}, a - 0.3333333333333333, -0.3333333333333333\right) + a} \]
5. Add Preprocessing

Alternative 2: 90.8% accurate, 2.1× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(\sqrt{a} \cdot 0.3333333333333333\right) \cdot rand\\ \mathbf{if}\;rand \leq -3.2 \cdot 10^{+110}:\\ \;\;\;\;t\_0\\ \mathbf{elif}\;rand \leq 3.4 \cdot 10^{+38}:\\ \;\;\;\;\left(1 - \frac{0.3333333333333333}{a}\right) \cdot a\\ \mathbf{else}:\\ \;\;\;\;t\_0\\ \end{array} \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (let* ((t_0 (* (* (sqrt a) 0.3333333333333333) rand)))
   (if (<= rand -3.2e+110)
     t_0
     (if (<= rand 3.4e+38) (* (- 1.0 (/ 0.3333333333333333 a)) a) t_0))))

C

double code(double a, double rand) {
	double t_0 = (sqrt(a) * 0.3333333333333333) * rand;
	double tmp;
	if (rand <= -3.2e+110) {
		tmp = t_0;
	} else if (rand <= 3.4e+38) {
		tmp = (1.0 - (0.3333333333333333 / a)) * a;
	} else {
		tmp = t_0;
	}
	return tmp;
}

Fortran

real(8) function code(a, rand)
    real(8), intent (in) :: a
    real(8), intent (in) :: rand
    real(8) :: t_0
    real(8) :: tmp
    t_0 = (sqrt(a) * 0.3333333333333333d0) * rand
    if (rand <= (-3.2d+110)) then
        tmp = t_0
    else if (rand <= 3.4d+38) then
        tmp = (1.0d0 - (0.3333333333333333d0 / a)) * a
    else
        tmp = t_0
    end if
    code = tmp
end function

Java

public static double code(double a, double rand) {
	double t_0 = (Math.sqrt(a) * 0.3333333333333333) * rand;
	double tmp;
	if (rand <= -3.2e+110) {
		tmp = t_0;
	} else if (rand <= 3.4e+38) {
		tmp = (1.0 - (0.3333333333333333 / a)) * a;
	} else {
		tmp = t_0;
	}
	return tmp;
}

Python

def code(a, rand):
	t_0 = (math.sqrt(a) * 0.3333333333333333) * rand
	tmp = 0
	if rand <= -3.2e+110:
		tmp = t_0
	elif rand <= 3.4e+38:
		tmp = (1.0 - (0.3333333333333333 / a)) * a
	else:
		tmp = t_0
	return tmp

Julia

function code(a, rand)
	t_0 = Float64(Float64(sqrt(a) * 0.3333333333333333) * rand)
	tmp = 0.0
	if (rand <= -3.2e+110)
		tmp = t_0;
	elseif (rand <= 3.4e+38)
		tmp = Float64(Float64(1.0 - Float64(0.3333333333333333 / a)) * a);
	else
		tmp = t_0;
	end
	return tmp
end

MATLAB

function tmp_2 = code(a, rand)
	t_0 = (sqrt(a) * 0.3333333333333333) * rand;
	tmp = 0.0;
	if (rand <= -3.2e+110)
		tmp = t_0;
	elseif (rand <= 3.4e+38)
		tmp = (1.0 - (0.3333333333333333 / a)) * a;
	else
		tmp = t_0;
	end
	tmp_2 = tmp;
end

Wolfram

code[a_, rand_] := Block[{t$95$0 = N[(N[(N[Sqrt[a], $MachinePrecision] * 0.3333333333333333), $MachinePrecision] * rand), $MachinePrecision]}, If[LessEqual[rand, -3.2e+110], t$95$0, If[LessEqual[rand, 3.4e+38], N[(N[(1.0 - N[(0.3333333333333333 / a), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision], t$95$0]]]

Derivation

1. Split input into 2 regimes

2. if rand < -3.19999999999999994e110 or 3.39999999999999996e38 < rand

   1. Initial program 99.7%

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

   3. Taylor expanded in rand around inf

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

      1. *-commutative N/A

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

      2. lower-*.f64 N/A

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

      3. associate--l+ N/A

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

      4. *-commutative N/A

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

      5. associate-*r/ N/A

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

      6. metadata-eval N/A

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

      7. div-sub N/A

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

      8. lower-fma.f64 N/A

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

      9. lower-sqrt.f64 N/A

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

      10. lower--.f64 N/A

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

      11. lower-/.f64 N/A

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

      12. lower--.f64 99.6

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

   5. Applied rewrites 99.6%

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

   6. Taylor expanded in rand around inf

      \[\leadsto \left(\frac{1}{3} \cdot \sqrt{a - \frac{1}{3}}\right) \cdot rand \]
   7. Step-by-step derivation

   8. Applied rewrites 90.1%

      \[\leadsto \left(\sqrt{a - 0.3333333333333333} \cdot 0.3333333333333333\right) \cdot rand \]

   9. Taylor expanded in a around -inf

      \[\leadsto \left(\frac{-1}{3} \cdot \left(\sqrt{a} \cdot {\left(\sqrt{-1}\right)}^{2}\right)\right) \cdot rand \]
   10. Step-by-step derivation

   11. Applied rewrites 88.8%

       \[\leadsto \left(\sqrt{a} \cdot 0.3333333333333333\right) \cdot rand \]

   if -3.19999999999999994e110 < rand < 3.39999999999999996e38

   1. Initial program 100.0%

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

   3. Taylor expanded in rand around 0

      \[\leadsto \color{blue}{a - \frac{1}{3}} \]
   4. Step-by-step derivation

      1. lower--.f64 93.7

         \[\leadsto \color{blue}{a - 0.3333333333333333} \]

   5. Applied rewrites 93.7%

      \[\leadsto \color{blue}{a - 0.3333333333333333} \]

   6. Taylor expanded in a around inf

      \[\leadsto a \cdot \color{blue}{\left(1 - \frac{1}{3} \cdot \frac{1}{a}\right)} \]
   7. Step-by-step derivation

   8. Applied rewrites 93.7%

      \[\leadsto \left(1 - \frac{0.3333333333333333}{a}\right) \cdot \color{blue}{a} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 3: 99.8% accurate, 2.4× speedup

Math

\[\begin{array}{l} \\ a - \mathsf{fma}\left(\sqrt{a - 0.3333333333333333} \cdot -0.3333333333333333, rand, 0.3333333333333333\right) \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (-
  a
  (fma
   (* (sqrt (- a 0.3333333333333333)) -0.3333333333333333)
   rand
   0.3333333333333333)))

C

double code(double a, double rand) {
	return a - fma((sqrt((a - 0.3333333333333333)) * -0.3333333333333333), rand, 0.3333333333333333);
}

Julia

function code(a, rand)
	return Float64(a - fma(Float64(sqrt(Float64(a - 0.3333333333333333)) * -0.3333333333333333), rand, 0.3333333333333333))
end

Wolfram

code[a_, rand_] := N[(a - N[(N[(N[Sqrt[N[(a - 0.3333333333333333), $MachinePrecision]], $MachinePrecision] * -0.3333333333333333), $MachinePrecision] * rand + 0.3333333333333333), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 99.8%

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

3. Taylor expanded in rand around 0

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

   1. +-commutative N/A

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

   2. associate--l+ N/A

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

   3. associate-*r* N/A

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

   4. lower-fma.f64 N/A

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

   5. lower-*.f64 N/A

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

   6. lower-sqrt.f64 N/A

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

   7. lower--.f64 N/A

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

   8. lower--.f64 99.8

      \[\leadsto \mathsf{fma}\left(0.3333333333333333 \cdot rand, \sqrt{a - 0.3333333333333333}, \color{blue}{a - 0.3333333333333333}\right) \]

5. Applied rewrites 99.8%

   \[\leadsto \color{blue}{\mathsf{fma}\left(0.3333333333333333 \cdot rand, \sqrt{a - 0.3333333333333333}, a - 0.3333333333333333\right)} \]
6. Step-by-step derivation

7. Applied rewrites 99.8%

   \[\leadsto a - \color{blue}{\left(0.3333333333333333 - \left(\sqrt{a - 0.3333333333333333} \cdot 0.3333333333333333\right) \cdot rand\right)} \]
8. Step-by-step derivation

9. Applied rewrites 99.8%

   \[\leadsto a - \color{blue}{\mathsf{fma}\left(-0.3333333333333333 \cdot \sqrt{a - 0.3333333333333333}, rand, 0.3333333333333333\right)} \]

10. Final simplification 99.8%

    \[\leadsto a - \mathsf{fma}\left(\sqrt{a - 0.3333333333333333} \cdot -0.3333333333333333, rand, 0.3333333333333333\right) \]
11. Add Preprocessing

Alternative 4: 67.3% accurate, 2.6× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;rand \leq 7.6 \cdot 10^{+135}:\\ \;\;\;\;\left(1 - \frac{0.3333333333333333}{a}\right) \cdot a\\ \mathbf{else}:\\ \;\;\;\;\frac{a \cdot a}{0.3333333333333333}\\ \end{array} \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (if (<= rand 7.6e+135)
   (* (- 1.0 (/ 0.3333333333333333 a)) a)
   (/ (* a a) 0.3333333333333333)))

C

double code(double a, double rand) {
	double tmp;
	if (rand <= 7.6e+135) {
		tmp = (1.0 - (0.3333333333333333 / a)) * a;
	} else {
		tmp = (a * a) / 0.3333333333333333;
	}
	return tmp;
}

Fortran

real(8) function code(a, rand)
    real(8), intent (in) :: a
    real(8), intent (in) :: rand
    real(8) :: tmp
    if (rand <= 7.6d+135) then
        tmp = (1.0d0 - (0.3333333333333333d0 / a)) * a
    else
        tmp = (a * a) / 0.3333333333333333d0
    end if
    code = tmp
end function

Java

public static double code(double a, double rand) {
	double tmp;
	if (rand <= 7.6e+135) {
		tmp = (1.0 - (0.3333333333333333 / a)) * a;
	} else {
		tmp = (a * a) / 0.3333333333333333;
	}
	return tmp;
}

Python

def code(a, rand):
	tmp = 0
	if rand <= 7.6e+135:
		tmp = (1.0 - (0.3333333333333333 / a)) * a
	else:
		tmp = (a * a) / 0.3333333333333333
	return tmp

Julia

function code(a, rand)
	tmp = 0.0
	if (rand <= 7.6e+135)
		tmp = Float64(Float64(1.0 - Float64(0.3333333333333333 / a)) * a);
	else
		tmp = Float64(Float64(a * a) / 0.3333333333333333);
	end
	return tmp
end

MATLAB

function tmp_2 = code(a, rand)
	tmp = 0.0;
	if (rand <= 7.6e+135)
		tmp = (1.0 - (0.3333333333333333 / a)) * a;
	else
		tmp = (a * a) / 0.3333333333333333;
	end
	tmp_2 = tmp;
end

Wolfram

code[a_, rand_] := If[LessEqual[rand, 7.6e+135], N[(N[(1.0 - N[(0.3333333333333333 / a), $MachinePrecision]), $MachinePrecision] * a), $MachinePrecision], N[(N[(a * a), $MachinePrecision] / 0.3333333333333333), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if rand < 7.6000000000000003e135

   1. Initial program 99.9%

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

   3. Taylor expanded in rand around 0

      \[\leadsto \color{blue}{a - \frac{1}{3}} \]
   4. Step-by-step derivation

      1. lower--.f64 71.7

         \[\leadsto \color{blue}{a - 0.3333333333333333} \]

   5. Applied rewrites 71.7%

      \[\leadsto \color{blue}{a - 0.3333333333333333} \]

   6. Taylor expanded in a around inf

      \[\leadsto a \cdot \color{blue}{\left(1 - \frac{1}{3} \cdot \frac{1}{a}\right)} \]
   7. Step-by-step derivation

   8. Applied rewrites 71.7%

      \[\leadsto \left(1 - \frac{0.3333333333333333}{a}\right) \cdot \color{blue}{a} \]

   if 7.6000000000000003e135 < rand

   1. Initial program 99.7%

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

   3. Taylor expanded in rand around 0

      \[\leadsto \color{blue}{a - \frac{1}{3}} \]
   4. Step-by-step derivation

      1. lower--.f64 5.7

         \[\leadsto \color{blue}{a - 0.3333333333333333} \]

   5. Applied rewrites 5.7%

      \[\leadsto \color{blue}{a - 0.3333333333333333} \]
   6. Step-by-step derivation

   7. Applied rewrites 46.1%

      \[\leadsto \frac{\mathsf{fma}\left(a, a, -0.1111111111111111\right)}{\color{blue}{a - -0.3333333333333333}} \]

   8. Taylor expanded in a around 0

      \[\leadsto \frac{\mathsf{fma}\left(a, a, \frac{-1}{9}\right)}{\frac{1}{3}} \]
   9. Step-by-step derivation

   10. Applied rewrites 47.3%

       \[\leadsto \frac{\mathsf{fma}\left(a, a, -0.1111111111111111\right)}{0.3333333333333333} \]

   11. Taylor expanded in a around inf

       \[\leadsto \frac{{a}^{2}}{\frac{1}{3}} \]
   12. Step-by-step derivation

   13. Applied rewrites 47.3%

       \[\leadsto \frac{a \cdot a}{0.3333333333333333} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 5: 98.8% accurate, 2.7× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(0.3333333333333333 \cdot rand, \sqrt{a}, a - 0.3333333333333333\right) \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (fma (* 0.3333333333333333 rand) (sqrt a) (- a 0.3333333333333333)))

C

double code(double a, double rand) {
	return fma((0.3333333333333333 * rand), sqrt(a), (a - 0.3333333333333333));
}

Julia

function code(a, rand)
	return fma(Float64(0.3333333333333333 * rand), sqrt(a), Float64(a - 0.3333333333333333))
end

Wolfram

code[a_, rand_] := N[(N[(0.3333333333333333 * rand), $MachinePrecision] * N[Sqrt[a], $MachinePrecision] + N[(a - 0.3333333333333333), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 99.8%

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

3. Taylor expanded in rand around 0

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

   1. +-commutative N/A

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

   2. associate--l+ N/A

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

   3. associate-*r* N/A

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

   4. lower-fma.f64 N/A

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

   5. lower-*.f64 N/A

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

   6. lower-sqrt.f64 N/A

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

   7. lower--.f64 N/A

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

   8. lower--.f64 99.8

      \[\leadsto \mathsf{fma}\left(0.3333333333333333 \cdot rand, \sqrt{a - 0.3333333333333333}, \color{blue}{a - 0.3333333333333333}\right) \]

5. Applied rewrites 99.8%

   \[\leadsto \color{blue}{\mathsf{fma}\left(0.3333333333333333 \cdot rand, \sqrt{a - 0.3333333333333333}, a - 0.3333333333333333\right)} \]

6. Taylor expanded in a around inf

   \[\leadsto \mathsf{fma}\left(\frac{1}{3} \cdot rand, \sqrt{a}, a - \frac{1}{3}\right) \]
7. Step-by-step derivation

8. Applied rewrites 98.9%

   \[\leadsto \mathsf{fma}\left(0.3333333333333333 \cdot rand, \sqrt{a}, a - 0.3333333333333333\right) \]
9. Add Preprocessing

Alternative 6: 97.8% accurate, 2.8× speedup

Math

\[\begin{array}{l} \\ \left(0.3333333333333333 \cdot rand\right) \cdot \sqrt{a} + a \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (+ (* (* 0.3333333333333333 rand) (sqrt a)) a))

C

double code(double a, double rand) {
	return ((0.3333333333333333 * rand) * sqrt(a)) + a;
}

Fortran

real(8) function code(a, rand)
    real(8), intent (in) :: a
    real(8), intent (in) :: rand
    code = ((0.3333333333333333d0 * rand) * sqrt(a)) + a
end function

Java

public static double code(double a, double rand) {
	return ((0.3333333333333333 * rand) * Math.sqrt(a)) + a;
}

Python

def code(a, rand):
	return ((0.3333333333333333 * rand) * math.sqrt(a)) + a

Julia

function code(a, rand)
	return Float64(Float64(Float64(0.3333333333333333 * rand) * sqrt(a)) + a)
end

MATLAB

function tmp = code(a, rand)
	tmp = ((0.3333333333333333 * rand) * sqrt(a)) + a;
end

Wolfram

code[a_, rand_] := N[(N[(N[(0.3333333333333333 * rand), $MachinePrecision] * N[Sqrt[a], $MachinePrecision]), $MachinePrecision] + a), $MachinePrecision]

Derivation

1. Initial program 99.8%

   \[\left(a - \frac{1}{3}\right) \cdot \left(1 + \frac{1}{\sqrt{9 \cdot \left(a - \frac{1}{3}\right)}} \cdot rand\right) \]
2. Add Preprocessing
3. Step-by-step derivation

   1. lift-*.f64 N/A

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

   2. lift-+.f64 N/A

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

   3. +-commutative N/A

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

   4. distribute-lft-in N/A

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

   5. *-rgt-identity N/A

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

   6. lift--.f64 N/A

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

   7. sub-neg N/A

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

   8. +-commutative N/A

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

   9. associate-+r+ N/A

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

   10. lower-+.f64 N/A

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

4. Applied rewrites 99.9%

   \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{rand}{\sqrt{\mathsf{fma}\left(a, 9, -3\right)}}, a - 0.3333333333333333, -0.3333333333333333\right) + a} \]

5. Taylor expanded in a around inf

   \[\leadsto \color{blue}{\frac{1}{3} \cdot \left(\sqrt{a} \cdot rand\right)} + a \]
6. Step-by-step derivation

   1. *-commutative N/A

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

   2. lower-*.f64 N/A

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

   3. lower-*.f64 N/A

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

   4. lower-sqrt.f64 97.7

      \[\leadsto \left(\color{blue}{\sqrt{a}} \cdot rand\right) \cdot 0.3333333333333333 + a \]

7. Applied rewrites 97.7%

   \[\leadsto \color{blue}{\left(\sqrt{a} \cdot rand\right) \cdot 0.3333333333333333} + a \]
8. Step-by-step derivation

9. Applied rewrites 98.0%

   \[\leadsto \left(rand \cdot 0.3333333333333333\right) \cdot \color{blue}{\sqrt{a}} + a \]

10. Final simplification 98.0%

    \[\leadsto \left(0.3333333333333333 \cdot rand\right) \cdot \sqrt{a} + a \]
11. Add Preprocessing

Alternative 7: 67.3% accurate, 3.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;rand \leq 7.6 \cdot 10^{+135}:\\ \;\;\;\;a - 0.3333333333333333\\ \mathbf{else}:\\ \;\;\;\;\frac{a \cdot a}{0.3333333333333333}\\ \end{array} \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (if (<= rand 7.6e+135)
   (- a 0.3333333333333333)
   (/ (* a a) 0.3333333333333333)))

C

double code(double a, double rand) {
	double tmp;
	if (rand <= 7.6e+135) {
		tmp = a - 0.3333333333333333;
	} else {
		tmp = (a * a) / 0.3333333333333333;
	}
	return tmp;
}

Fortran

real(8) function code(a, rand)
    real(8), intent (in) :: a
    real(8), intent (in) :: rand
    real(8) :: tmp
    if (rand <= 7.6d+135) then
        tmp = a - 0.3333333333333333d0
    else
        tmp = (a * a) / 0.3333333333333333d0
    end if
    code = tmp
end function

Java

public static double code(double a, double rand) {
	double tmp;
	if (rand <= 7.6e+135) {
		tmp = a - 0.3333333333333333;
	} else {
		tmp = (a * a) / 0.3333333333333333;
	}
	return tmp;
}

Python

def code(a, rand):
	tmp = 0
	if rand <= 7.6e+135:
		tmp = a - 0.3333333333333333
	else:
		tmp = (a * a) / 0.3333333333333333
	return tmp

Julia

function code(a, rand)
	tmp = 0.0
	if (rand <= 7.6e+135)
		tmp = Float64(a - 0.3333333333333333);
	else
		tmp = Float64(Float64(a * a) / 0.3333333333333333);
	end
	return tmp
end

MATLAB

function tmp_2 = code(a, rand)
	tmp = 0.0;
	if (rand <= 7.6e+135)
		tmp = a - 0.3333333333333333;
	else
		tmp = (a * a) / 0.3333333333333333;
	end
	tmp_2 = tmp;
end

Wolfram

code[a_, rand_] := If[LessEqual[rand, 7.6e+135], N[(a - 0.3333333333333333), $MachinePrecision], N[(N[(a * a), $MachinePrecision] / 0.3333333333333333), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if rand < 7.6000000000000003e135

   1. Initial program 99.9%

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

   3. Taylor expanded in rand around 0

      \[\leadsto \color{blue}{a - \frac{1}{3}} \]
   4. Step-by-step derivation

      1. lower--.f64 71.7

         \[\leadsto \color{blue}{a - 0.3333333333333333} \]

   5. Applied rewrites 71.7%

      \[\leadsto \color{blue}{a - 0.3333333333333333} \]

   if 7.6000000000000003e135 < rand

   1. Initial program 99.7%

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

   3. Taylor expanded in rand around 0

      \[\leadsto \color{blue}{a - \frac{1}{3}} \]
   4. Step-by-step derivation

      1. lower--.f64 5.7

         \[\leadsto \color{blue}{a - 0.3333333333333333} \]

   5. Applied rewrites 5.7%

      \[\leadsto \color{blue}{a - 0.3333333333333333} \]
   6. Step-by-step derivation

   7. Applied rewrites 46.1%

      \[\leadsto \frac{\mathsf{fma}\left(a, a, -0.1111111111111111\right)}{\color{blue}{a - -0.3333333333333333}} \]

   8. Taylor expanded in a around 0

      \[\leadsto \frac{\mathsf{fma}\left(a, a, \frac{-1}{9}\right)}{\frac{1}{3}} \]
   9. Step-by-step derivation

   10. Applied rewrites 47.3%

       \[\leadsto \frac{\mathsf{fma}\left(a, a, -0.1111111111111111\right)}{0.3333333333333333} \]

   11. Taylor expanded in a around inf

       \[\leadsto \frac{{a}^{2}}{\frac{1}{3}} \]
   12. Step-by-step derivation

   13. Applied rewrites 47.3%

       \[\leadsto \frac{a \cdot a}{0.3333333333333333} \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 8: 97.7% accurate, 3.1× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(\sqrt{a} \cdot rand, 0.3333333333333333, a\right) \end{array} \]

FPCore

(FPCore (a rand)
 :precision binary64
 (fma (* (sqrt a) rand) 0.3333333333333333 a))

C

double code(double a, double rand) {
	return fma((sqrt(a) * rand), 0.3333333333333333, a);
}

Julia

function code(a, rand)
	return fma(Float64(sqrt(a) * rand), 0.3333333333333333, a)
end

Wolfram

code[a_, rand_] := N[(N[(N[Sqrt[a], $MachinePrecision] * rand), $MachinePrecision] * 0.3333333333333333 + a), $MachinePrecision]

Derivation

1. Initial program 99.8%

   \[\left(a - \frac{1}{3}\right) \cdot \left(1 + \frac{1}{\sqrt{9 \cdot \left(a - \frac{1}{3}\right)}} \cdot rand\right) \]
2. Add Preprocessing
3. Step-by-step derivation

   1. lift-*.f64 N/A

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

   2. lift-+.f64 N/A

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

   3. +-commutative N/A

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

   4. distribute-lft-in N/A

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

   5. *-rgt-identity N/A

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

   6. lift--.f64 N/A

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

   7. sub-neg N/A

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

   8. +-commutative N/A

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

   9. associate-+r+ N/A

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

   10. lower-+.f64 N/A

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

4. Applied rewrites 99.9%

   \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{rand}{\sqrt{\mathsf{fma}\left(a, 9, -3\right)}}, a - 0.3333333333333333, -0.3333333333333333\right) + a} \]

5. Taylor expanded in a around inf

   \[\leadsto \color{blue}{a \cdot \left(1 + \frac{1}{3} \cdot \left(\sqrt{\frac{1}{a}} \cdot rand\right)\right)} \]
6. Step-by-step derivation

   1. lft-mult-inverse N/A

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

   2. associate-*r* N/A

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

   3. distribute-rgt-in N/A

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

   4. +-commutative N/A

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

   5. *-commutative N/A

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

   6. lower-*.f64 N/A

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

   7. distribute-lft-in N/A

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

   8. associate-*r* N/A

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

   9. *-commutative N/A

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

   10. rgt-mult-inverse N/A

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

   11. lower-fma.f64 N/A

       \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{1}{3} \cdot rand, \sqrt{\frac{1}{a}}, 1\right)} \cdot a \]

   12. lower-*.f64 N/A

       \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{1}{3} \cdot rand}, \sqrt{\frac{1}{a}}, 1\right) \cdot a \]

   13. lower-sqrt.f64 N/A

       \[\leadsto \mathsf{fma}\left(\frac{1}{3} \cdot rand, \color{blue}{\sqrt{\frac{1}{a}}}, 1\right) \cdot a \]

   14. lower-/.f64 98.0

       \[\leadsto \mathsf{fma}\left(0.3333333333333333 \cdot rand, \sqrt{\color{blue}{\frac{1}{a}}}, 1\right) \cdot a \]

7. Applied rewrites 98.0%

   \[\leadsto \color{blue}{\mathsf{fma}\left(0.3333333333333333 \cdot rand, \sqrt{\frac{1}{a}}, 1\right) \cdot a} \]

8. Taylor expanded in a around 0

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

10. Applied rewrites 97.7%

    \[\leadsto \mathsf{fma}\left(\sqrt{a} \cdot rand, \color{blue}{0.3333333333333333}, a\right) \]
11. Add Preprocessing

Alternative 9: 62.7% accurate, 17.0× speedup

Math

\[\begin{array}{l} \\ a - 0.3333333333333333 \end{array} \]

FPCore

(FPCore (a rand) :precision binary64 (- a 0.3333333333333333))

C

double code(double a, double rand) {
	return a - 0.3333333333333333;
}

Fortran

real(8) function code(a, rand)
    real(8), intent (in) :: a
    real(8), intent (in) :: rand
    code = a - 0.3333333333333333d0
end function

Java

public static double code(double a, double rand) {
	return a - 0.3333333333333333;
}

Python

def code(a, rand):
	return a - 0.3333333333333333

Julia

function code(a, rand)
	return Float64(a - 0.3333333333333333)
end

MATLAB

function tmp = code(a, rand)
	tmp = a - 0.3333333333333333;
end

Wolfram

code[a_, rand_] := N[(a - 0.3333333333333333), $MachinePrecision]

Derivation

1. Initial program 99.8%

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

3. Taylor expanded in rand around 0

   \[\leadsto \color{blue}{a - \frac{1}{3}} \]
4. Step-by-step derivation

   1. lower--.f64 61.4

      \[\leadsto \color{blue}{a - 0.3333333333333333} \]

5. Applied rewrites 61.4%

   \[\leadsto \color{blue}{a - 0.3333333333333333} \]
6. Add Preprocessing

Alternative 10: 1.5% accurate, 68.0× speedup

Math

\[\begin{array}{l} \\ -0.3333333333333333 \end{array} \]

FPCore

(FPCore (a rand) :precision binary64 -0.3333333333333333)

C

double code(double a, double rand) {
	return -0.3333333333333333;
}

Fortran

real(8) function code(a, rand)
    real(8), intent (in) :: a
    real(8), intent (in) :: rand
    code = -0.3333333333333333d0
end function

Java

public static double code(double a, double rand) {
	return -0.3333333333333333;
}

Python

def code(a, rand):
	return -0.3333333333333333

Julia

function code(a, rand)
	return -0.3333333333333333
end

MATLAB

function tmp = code(a, rand)
	tmp = -0.3333333333333333;
end

Wolfram

code[a_, rand_] := -0.3333333333333333

Derivation

1. Initial program 99.8%

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

3. Taylor expanded in rand around 0

   \[\leadsto \color{blue}{a - \frac{1}{3}} \]
4. Step-by-step derivation

   1. lower--.f64 61.4

      \[\leadsto \color{blue}{a - 0.3333333333333333} \]

5. Applied rewrites 61.4%

   \[\leadsto \color{blue}{a - 0.3333333333333333} \]

6. Taylor expanded in a around 0

   \[\leadsto \frac{-1}{3} \]
7. Step-by-step derivation

8. Applied rewrites 1.5%

   \[\leadsto -0.3333333333333333 \]
9. Add Preprocessing

Reproduce

herbie shell --seed 2024308 
(FPCore (a rand)
  :name "Octave 3.8, oct_fill_randg"
  :precision binary64
  (* (- a (/ 1.0 3.0)) (+ 1.0 (* (/ 1.0 (sqrt (* 9.0 (- a (/ 1.0 3.0))))) rand))))