normal distribution

Percentage Accurate: 99.4% → 99.4%

Time: 3.9s

Alternatives: 7

Speedup: 1.3×

Specification

\[\left(0 \leq u1 \land u1 \leq 1\right) \land \left(0 \leq u2 \land u2 \leq 1\right)\]

Math

\[\begin{array}{l} \\ \left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \end{array} \]

FPCore

(FPCore (u1 u2)
 :precision binary64
 (+
  (* (* (/ 1.0 6.0) (pow (* -2.0 (log u1)) 0.5)) (cos (* (* 2.0 PI) u2)))
  0.5))

C

double code(double u1, double u2) {
	return (((1.0 / 6.0) * pow((-2.0 * log(u1)), 0.5)) * cos(((2.0 * ((double) M_PI)) * u2))) + 0.5;
}

Java

public static double code(double u1, double u2) {
	return (((1.0 / 6.0) * Math.pow((-2.0 * Math.log(u1)), 0.5)) * Math.cos(((2.0 * Math.PI) * u2))) + 0.5;
}

Python

def code(u1, u2):
	return (((1.0 / 6.0) * math.pow((-2.0 * math.log(u1)), 0.5)) * math.cos(((2.0 * math.pi) * u2))) + 0.5

Julia

function code(u1, u2)
	return Float64(Float64(Float64(Float64(1.0 / 6.0) * (Float64(-2.0 * log(u1)) ^ 0.5)) * cos(Float64(Float64(2.0 * pi) * u2))) + 0.5)
end

MATLAB

function tmp = code(u1, u2)
	tmp = (((1.0 / 6.0) * ((-2.0 * log(u1)) ^ 0.5)) * cos(((2.0 * pi) * u2))) + 0.5;
end

Wolfram

code[u1_, u2_] := N[(N[(N[(N[(1.0 / 6.0), $MachinePrecision] * N[Power[N[(-2.0 * N[Log[u1], $MachinePrecision]), $MachinePrecision], 0.5], $MachinePrecision]), $MachinePrecision] * N[Cos[N[(N[(2.0 * Pi), $MachinePrecision] * u2), $MachinePrecision]], $MachinePrecision]), $MachinePrecision] + 0.5), $MachinePrecision]

Initial Program: 99.4% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \end{array} \]

FPCore

(FPCore (u1 u2)
 :precision binary64
 (+
  (* (* (/ 1.0 6.0) (pow (* -2.0 (log u1)) 0.5)) (cos (* (* 2.0 PI) u2)))
  0.5))

C

double code(double u1, double u2) {
	return (((1.0 / 6.0) * pow((-2.0 * log(u1)), 0.5)) * cos(((2.0 * ((double) M_PI)) * u2))) + 0.5;
}

Java

public static double code(double u1, double u2) {
	return (((1.0 / 6.0) * Math.pow((-2.0 * Math.log(u1)), 0.5)) * Math.cos(((2.0 * Math.PI) * u2))) + 0.5;
}

Python

def code(u1, u2):
	return (((1.0 / 6.0) * math.pow((-2.0 * math.log(u1)), 0.5)) * math.cos(((2.0 * math.pi) * u2))) + 0.5

Julia

function code(u1, u2)
	return Float64(Float64(Float64(Float64(1.0 / 6.0) * (Float64(-2.0 * log(u1)) ^ 0.5)) * cos(Float64(Float64(2.0 * pi) * u2))) + 0.5)
end

MATLAB

function tmp = code(u1, u2)
	tmp = (((1.0 / 6.0) * ((-2.0 * log(u1)) ^ 0.5)) * cos(((2.0 * pi) * u2))) + 0.5;
end

Wolfram

code[u1_, u2_] := N[(N[(N[(N[(1.0 / 6.0), $MachinePrecision] * N[Power[N[(-2.0 * N[Log[u1], $MachinePrecision]), $MachinePrecision], 0.5], $MachinePrecision]), $MachinePrecision] * N[Cos[N[(N[(2.0 * Pi), $MachinePrecision] * u2), $MachinePrecision]], $MachinePrecision]), $MachinePrecision] + 0.5), $MachinePrecision]

Alternative 1: 99.4% accurate, 1.3× speedup

Math

\[\begin{array}{l} \\ 0.16666666666666666 \cdot \left(\cos \left(2 \cdot \left(u2 \cdot \pi\right)\right) \cdot \sqrt{-2 \cdot \log u1}\right) + 0.5 \end{array} \]

FPCore

(FPCore (u1 u2)
 :precision binary64
 (+
  (* 0.16666666666666666 (* (cos (* 2.0 (* u2 PI))) (sqrt (* -2.0 (log u1)))))
  0.5))

C

double code(double u1, double u2) {
	return (0.16666666666666666 * (cos((2.0 * (u2 * ((double) M_PI)))) * sqrt((-2.0 * log(u1))))) + 0.5;
}

Java

public static double code(double u1, double u2) {
	return (0.16666666666666666 * (Math.cos((2.0 * (u2 * Math.PI))) * Math.sqrt((-2.0 * Math.log(u1))))) + 0.5;
}

Python

def code(u1, u2):
	return (0.16666666666666666 * (math.cos((2.0 * (u2 * math.pi))) * math.sqrt((-2.0 * math.log(u1))))) + 0.5

Julia

function code(u1, u2)
	return Float64(Float64(0.16666666666666666 * Float64(cos(Float64(2.0 * Float64(u2 * pi))) * sqrt(Float64(-2.0 * log(u1))))) + 0.5)
end

MATLAB

function tmp = code(u1, u2)
	tmp = (0.16666666666666666 * (cos((2.0 * (u2 * pi))) * sqrt((-2.0 * log(u1))))) + 0.5;
end

Wolfram

code[u1_, u2_] := N[(N[(0.16666666666666666 * N[(N[Cos[N[(2.0 * N[(u2 * Pi), $MachinePrecision]), $MachinePrecision]], $MachinePrecision] * N[Sqrt[N[(-2.0 * N[Log[u1], $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision] + 0.5), $MachinePrecision]

Derivation

1. Initial program 99.4%

   \[\left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \]

2. Taylor expanded in u1 around 0

   \[\leadsto \color{blue}{\frac{1}{6} \cdot \left(\cos \left(2 \cdot \left(u2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{-2 \cdot \log u1}\right)} + \frac{1}{2} \]

4. Applied rewrites 99.4%

   \[\leadsto \color{blue}{0.16666666666666666 \cdot \left(\cos \left(2 \cdot \left(u2 \cdot \pi\right)\right) \cdot \sqrt{-2 \cdot \log u1}\right)} + 0.5 \]
5. Add Preprocessing

Alternative 2: 98.1% accurate, 1.8× speedup

Math

\[\begin{array}{l} \\ \frac{\sqrt{-2 \cdot \log u1}}{6} \cdot \left(1 + -2 \cdot {u2}^{2}\right) + 0.5 \end{array} \]

FPCore

(FPCore (u1 u2)
 :precision binary64
 (+ (* (/ (sqrt (* -2.0 (log u1))) 6.0) (+ 1.0 (* -2.0 (pow u2 2.0)))) 0.5))

C

double code(double u1, double u2) {
	return ((sqrt((-2.0 * log(u1))) / 6.0) * (1.0 + (-2.0 * pow(u2, 2.0)))) + 0.5;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

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

real(8) function code(u1, u2)
use fmin_fmax_functions
    real(8), intent (in) :: u1
    real(8), intent (in) :: u2
    code = ((sqrt(((-2.0d0) * log(u1))) / 6.0d0) * (1.0d0 + ((-2.0d0) * (u2 ** 2.0d0)))) + 0.5d0
end function

Java

public static double code(double u1, double u2) {
	return ((Math.sqrt((-2.0 * Math.log(u1))) / 6.0) * (1.0 + (-2.0 * Math.pow(u2, 2.0)))) + 0.5;
}

Python

def code(u1, u2):
	return ((math.sqrt((-2.0 * math.log(u1))) / 6.0) * (1.0 + (-2.0 * math.pow(u2, 2.0)))) + 0.5

Julia

function code(u1, u2)
	return Float64(Float64(Float64(sqrt(Float64(-2.0 * log(u1))) / 6.0) * Float64(1.0 + Float64(-2.0 * (u2 ^ 2.0)))) + 0.5)
end

MATLAB

function tmp = code(u1, u2)
	tmp = ((sqrt((-2.0 * log(u1))) / 6.0) * (1.0 + (-2.0 * (u2 ^ 2.0)))) + 0.5;
end

Wolfram

code[u1_, u2_] := N[(N[(N[(N[Sqrt[N[(-2.0 * N[Log[u1], $MachinePrecision]), $MachinePrecision]], $MachinePrecision] / 6.0), $MachinePrecision] * N[(1.0 + N[(-2.0 * N[Power[u2, 2.0], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision] + 0.5), $MachinePrecision]

Derivation

1. Initial program 99.4%

   \[\left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \]

3. Applied rewrites 98.1%

   \[\leadsto \color{blue}{\frac{\sqrt{-2 \cdot \log u1}}{6} \cdot \cos \left(u2 + u2\right)} + 0.5 \]

4. Taylor expanded in u2 around 0

   \[\leadsto \frac{\sqrt{-2 \cdot \log u1}}{6} \cdot \color{blue}{\left(1 + -2 \cdot {u2}^{2}\right)} + \frac{1}{2} \]

6. Applied rewrites 98.1%

   \[\leadsto \frac{\sqrt{-2 \cdot \log u1}}{6} \cdot \color{blue}{\left(1 + -2 \cdot {u2}^{2}\right)} + 0.5 \]
7. Add Preprocessing

Alternative 3: 98.1% accurate, 3.7× speedup

Math

\[\begin{array}{l} \\ \frac{\sqrt{-2 \cdot \log u1}}{6} \cdot 1 + 0.5 \end{array} \]

FPCore

(FPCore (u1 u2)
 :precision binary64
 (+ (* (/ (sqrt (* -2.0 (log u1))) 6.0) 1.0) 0.5))

C

double code(double u1, double u2) {
	return ((sqrt((-2.0 * log(u1))) / 6.0) * 1.0) + 0.5;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

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

real(8) function code(u1, u2)
use fmin_fmax_functions
    real(8), intent (in) :: u1
    real(8), intent (in) :: u2
    code = ((sqrt(((-2.0d0) * log(u1))) / 6.0d0) * 1.0d0) + 0.5d0
end function

Java

public static double code(double u1, double u2) {
	return ((Math.sqrt((-2.0 * Math.log(u1))) / 6.0) * 1.0) + 0.5;
}

Python

def code(u1, u2):
	return ((math.sqrt((-2.0 * math.log(u1))) / 6.0) * 1.0) + 0.5

Julia

function code(u1, u2)
	return Float64(Float64(Float64(sqrt(Float64(-2.0 * log(u1))) / 6.0) * 1.0) + 0.5)
end

MATLAB

function tmp = code(u1, u2)
	tmp = ((sqrt((-2.0 * log(u1))) / 6.0) * 1.0) + 0.5;
end

Wolfram

code[u1_, u2_] := N[(N[(N[(N[Sqrt[N[(-2.0 * N[Log[u1], $MachinePrecision]), $MachinePrecision]], $MachinePrecision] / 6.0), $MachinePrecision] * 1.0), $MachinePrecision] + 0.5), $MachinePrecision]

Derivation

1. Initial program 99.4%

   \[\left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \]

3. Applied rewrites 98.1%

   \[\leadsto \color{blue}{\frac{\sqrt{-2 \cdot \log u1}}{6} \cdot \cos \left(u2 + u2\right)} + 0.5 \]

4. Taylor expanded in u2 around 0

   \[\leadsto \frac{\sqrt{-2 \cdot \log u1}}{6} \cdot \color{blue}{1} + \frac{1}{2} \]

6. Applied rewrites 98.1%

   \[\leadsto \frac{\sqrt{-2 \cdot \log u1}}{6} \cdot \color{blue}{1} + 0.5 \]
7. Add Preprocessing

Alternative 4: 98.0% accurate, 4.4× speedup

Math

\[\begin{array}{l} \\ 0.5 + 0.16666666666666666 \cdot \sqrt{-2 \cdot \log u1} \end{array} \]

FPCore

(FPCore (u1 u2)
 :precision binary64
 (+ 0.5 (* 0.16666666666666666 (sqrt (* -2.0 (log u1))))))

C

double code(double u1, double u2) {
	return 0.5 + (0.16666666666666666 * sqrt((-2.0 * log(u1))));
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

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

real(8) function code(u1, u2)
use fmin_fmax_functions
    real(8), intent (in) :: u1
    real(8), intent (in) :: u2
    code = 0.5d0 + (0.16666666666666666d0 * sqrt(((-2.0d0) * log(u1))))
end function

Java

public static double code(double u1, double u2) {
	return 0.5 + (0.16666666666666666 * Math.sqrt((-2.0 * Math.log(u1))));
}

Python

def code(u1, u2):
	return 0.5 + (0.16666666666666666 * math.sqrt((-2.0 * math.log(u1))))

Julia

function code(u1, u2)
	return Float64(0.5 + Float64(0.16666666666666666 * sqrt(Float64(-2.0 * log(u1)))))
end

MATLAB

function tmp = code(u1, u2)
	tmp = 0.5 + (0.16666666666666666 * sqrt((-2.0 * log(u1))));
end

Wolfram

code[u1_, u2_] := N[(0.5 + N[(0.16666666666666666 * N[Sqrt[N[(-2.0 * N[Log[u1], $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 99.4%

   \[\left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \]

2. Taylor expanded in u2 around 0

   \[\leadsto \color{blue}{\frac{1}{2} + \frac{1}{6} \cdot \sqrt{-2 \cdot \log u1}} \]

4. Applied rewrites 98.0%

   \[\leadsto \color{blue}{0.5 + 0.16666666666666666 \cdot \sqrt{-2 \cdot \log u1}} \]
5. Add Preprocessing

Alternative 5: 20.1% accurate, 75.3× speedup

Math

\[\begin{array}{l} \\ \pi \end{array} \]

FPCore

(FPCore (u1 u2) :precision binary64 PI)

C

double code(double u1, double u2) {
	return (double) M_PI;
}

Java

public static double code(double u1, double u2) {
	return Math.PI;
}

Python

def code(u1, u2):
	return math.pi

Julia

function code(u1, u2)
	return pi
end

MATLAB

function tmp = code(u1, u2)
	tmp = pi;
end

Wolfram

code[u1_, u2_] := Pi

Derivation

1. Initial program 99.4%

   \[\left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \]

3. Applied rewrites 98.1%

   \[\leadsto \color{blue}{\mathsf{fma}\left(\sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6}, \sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6} - 0.5, 0.5 \cdot 0.5\right) \cdot \frac{\mathsf{fma}\left(\frac{\sqrt{-2 \cdot \log u1}}{6}, \cos \left(u2 + u2\right), 0.5\right)}{\mathsf{fma}\left(\sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6}, \sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6} - 0.5, 0.5 \cdot 0.5\right)}} \]

5. Applied rewrites 20.1%

   \[\leadsto \color{blue}{\pi} \]
6. Add Preprocessing

Alternative 6: 18.9% accurate, 75.3× speedup

Math

\[\begin{array}{l} \\ 2 \end{array} \]

FPCore

(FPCore (u1 u2) :precision binary64 2.0)

C

double code(double u1, double u2) {
	return 2.0;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

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

real(8) function code(u1, u2)
use fmin_fmax_functions
    real(8), intent (in) :: u1
    real(8), intent (in) :: u2
    code = 2.0d0
end function

Java

public static double code(double u1, double u2) {
	return 2.0;
}

Python

def code(u1, u2):
	return 2.0

Julia

function code(u1, u2)
	return 2.0
end

MATLAB

function tmp = code(u1, u2)
	tmp = 2.0;
end

Wolfram

code[u1_, u2_] := 2.0

Derivation

1. Initial program 99.4%

   \[\left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \]

3. Applied rewrites 98.1%

   \[\leadsto \color{blue}{\mathsf{fma}\left(\sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6}, \sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6} - 0.5, 0.5 \cdot 0.5\right) \cdot \frac{\mathsf{fma}\left(\frac{\sqrt{-2 \cdot \log u1}}{6}, \cos \left(u2 + u2\right), 0.5\right)}{\mathsf{fma}\left(\sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6}, \sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6} - 0.5, 0.5 \cdot 0.5\right)}} \]

5. Applied rewrites 18.9%

   \[\leadsto \color{blue}{2} \]
6. Add Preprocessing

Alternative 7: 17.2% accurate, 75.3× speedup

Math

\[\begin{array}{l} \\ 1 \end{array} \]

FPCore

(FPCore (u1 u2) :precision binary64 1.0)

C

double code(double u1, double u2) {
	return 1.0;
}

Fortran

module fmin_fmax_functions
    implicit none
    private
    public fmax
    public fmin

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

real(8) function code(u1, u2)
use fmin_fmax_functions
    real(8), intent (in) :: u1
    real(8), intent (in) :: u2
    code = 1.0d0
end function

Java

public static double code(double u1, double u2) {
	return 1.0;
}

Python

def code(u1, u2):
	return 1.0

Julia

function code(u1, u2)
	return 1.0
end

MATLAB

function tmp = code(u1, u2)
	tmp = 1.0;
end

Wolfram

code[u1_, u2_] := 1.0

Derivation

1. Initial program 99.4%

   \[\left(\frac{1}{6} \cdot {\left(-2 \cdot \log u1\right)}^{0.5}\right) \cdot \cos \left(\left(2 \cdot \pi\right) \cdot u2\right) + 0.5 \]

3. Applied rewrites 98.1%

   \[\leadsto \color{blue}{\mathsf{fma}\left(\sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6}, \sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6} - 0.5, 0.5 \cdot 0.5\right) \cdot \frac{\mathsf{fma}\left(\frac{\sqrt{-2 \cdot \log u1}}{6}, \cos \left(u2 + u2\right), 0.5\right)}{\mathsf{fma}\left(\sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6}, \sqrt{-2 \cdot \log u1} \cdot \frac{\cos \left(u2 + u2\right)}{6} - 0.5, 0.5 \cdot 0.5\right)}} \]

5. Applied rewrites 17.2%

   \[\leadsto \color{blue}{1} \]
6. Add Preprocessing

Reproduce

herbie shell --seed 2025159 
(FPCore (u1 u2)
  :name "normal distribution"
  :precision binary64
  :pre (and (and (<= 0.0 u1) (<= u1 1.0)) (and (<= 0.0 u2) (<= u2 1.0)))
  (+ (* (* (/ 1.0 6.0) (pow (* -2.0 (log u1)) 0.5)) (cos (* (* 2.0 PI) u2))) 0.5))