a parameter of renormalized beta distribution

Percentage Accurate: 99.8% → 99.7%

Time: 6.4s

Alternatives: 9

Speedup: N/A×

• 42.0% of points produce a very large (infinite) output. You may want to add a precondition.

Specification

\[\left(0 < m \land 0 < v\right) \land v < 0.25\]

Math

\[\begin{array}{l} \\ \left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \end{array} \]

FPCore

(FPCore (m v) :precision binary64 (* (- (/ (* m (- 1.0 m)) v) 1.0) m))

C

double code(double m, double v) {
	return (((m * (1.0 - m)) / v) - 1.0) * m;
}

Fortran

real(8) function code(m, v)
    real(8), intent (in) :: m
    real(8), intent (in) :: v
    code = (((m * (1.0d0 - m)) / v) - 1.0d0) * m
end function

Java

public static double code(double m, double v) {
	return (((m * (1.0 - m)) / v) - 1.0) * m;
}

Python

def code(m, v):
	return (((m * (1.0 - m)) / v) - 1.0) * m

Julia

function code(m, v)
	return Float64(Float64(Float64(Float64(m * Float64(1.0 - m)) / v) - 1.0) * m)
end

MATLAB

function tmp = code(m, v)
	tmp = (((m * (1.0 - m)) / v) - 1.0) * m;
end

Wolfram

code[m_, v_] := N[(N[(N[(N[(m * N[(1.0 - m), $MachinePrecision]), $MachinePrecision] / v), $MachinePrecision] - 1.0), $MachinePrecision] * m), $MachinePrecision]

Initial Program: 99.8% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \end{array} \]

FPCore

(FPCore (m v) :precision binary64 (* (- (/ (* m (- 1.0 m)) v) 1.0) m))

C

double code(double m, double v) {
	return (((m * (1.0 - m)) / v) - 1.0) * m;
}

Fortran

real(8) function code(m, v)
    real(8), intent (in) :: m
    real(8), intent (in) :: v
    code = (((m * (1.0d0 - m)) / v) - 1.0d0) * m
end function

Java

public static double code(double m, double v) {
	return (((m * (1.0 - m)) / v) - 1.0) * m;
}

Python

def code(m, v):
	return (((m * (1.0 - m)) / v) - 1.0) * m

Julia

function code(m, v)
	return Float64(Float64(Float64(Float64(m * Float64(1.0 - m)) / v) - 1.0) * m)
end

MATLAB

function tmp = code(m, v)
	tmp = (((m * (1.0 - m)) / v) - 1.0) * m;
end

Wolfram

code[m_, v_] := N[(N[(N[(N[(m * N[(1.0 - m), $MachinePrecision]), $MachinePrecision] / v), $MachinePrecision] - 1.0), $MachinePrecision] * m), $MachinePrecision]

Alternative 1: 99.7% accurate, 0.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;m \leq 2.2 \cdot 10^{-17}:\\ \;\;\;\;\mathsf{fma}\left(\frac{m}{v}, m, -m\right)\\ \mathbf{else}:\\ \;\;\;\;\left(m \cdot m\right) \cdot \frac{1 - m}{v}\\ \end{array} \end{array} \]

FPCore

(FPCore (m v)
 :precision binary64
 (if (<= m 2.2e-17) (fma (/ m v) m (- m)) (* (* m m) (/ (- 1.0 m) v))))

C

double code(double m, double v) {
	double tmp;
	if (m <= 2.2e-17) {
		tmp = fma((m / v), m, -m);
	} else {
		tmp = (m * m) * ((1.0 - m) / v);
	}
	return tmp;
}

Julia

function code(m, v)
	tmp = 0.0
	if (m <= 2.2e-17)
		tmp = fma(Float64(m / v), m, Float64(-m));
	else
		tmp = Float64(Float64(m * m) * Float64(Float64(1.0 - m) / v));
	end
	return tmp
end

Wolfram

code[m_, v_] := If[LessEqual[m, 2.2e-17], N[(N[(m / v), $MachinePrecision] * m + (-m)), $MachinePrecision], N[(N[(m * m), $MachinePrecision] * N[(N[(1.0 - m), $MachinePrecision] / v), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if m < 2.2e-17

   1. Initial program 99.7%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in m around 0

      \[\leadsto \color{blue}{m \cdot \left(\frac{m}{v} - 1\right)} \]
   4. Step-by-step derivation

      1. sub-neg N/A

         \[\leadsto m \cdot \color{blue}{\left(\frac{m}{v} + \left(\mathsf{neg}\left(1\right)\right)\right)} \]

      2. metadata-eval N/A

         \[\leadsto m \cdot \left(\frac{m}{v} + \color{blue}{-1}\right) \]

      3. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\frac{m}{v} \cdot m + -1 \cdot m} \]

      4. lower-fma.f64 N/A

         \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -1 \cdot m\right)} \]

      5. lower-/.f64 N/A

         \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{m}{v}}, m, -1 \cdot m\right) \]

      6. mul-1-neg N/A

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{\mathsf{neg}\left(m\right)}\right) \]

      7. lower-neg.f64 99.8

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{-m}\right) \]

   5. Applied rewrites 99.8%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -m\right)} \]

   if 2.2e-17 < m

   1. Initial program 99.8%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing
   3. Step-by-step derivation

      1. lift-*.f64 N/A

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

      2. *-commutative N/A

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

      3. lift--.f64 N/A

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

      4. flip3-- N/A

         \[\leadsto m \cdot \color{blue}{\frac{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}} \]

      5. clear-num N/A

         \[\leadsto m \cdot \color{blue}{\frac{1}{\frac{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}}} \]

      6. un-div-inv N/A

         \[\leadsto \color{blue}{\frac{m}{\frac{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}}} \]

      7. lower-/.f64 N/A

         \[\leadsto \color{blue}{\frac{m}{\frac{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}}} \]

      8. clear-num N/A

         \[\leadsto \frac{m}{\color{blue}{\frac{1}{\frac{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}}}} \]

      9. flip3-- N/A

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

      10. lift--.f64 N/A

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

   4. Applied rewrites 99.9%

      \[\leadsto \color{blue}{\frac{m}{\frac{1}{\mathsf{fma}\left(\frac{m}{v}, 1 - m, -1\right)}}} \]

   5. Taylor expanded in v around 0

      \[\leadsto \color{blue}{\frac{{m}^{2} \cdot \left(1 - m\right)}{v}} \]
   6. Step-by-step derivation

      1. associate-/l* N/A

         \[\leadsto \color{blue}{{m}^{2} \cdot \frac{1 - m}{v}} \]

      2. *-commutative N/A

         \[\leadsto \color{blue}{\frac{1 - m}{v} \cdot {m}^{2}} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\frac{1 - m}{v} \cdot {m}^{2}} \]

      4. lower-/.f64 N/A

         \[\leadsto \color{blue}{\frac{1 - m}{v}} \cdot {m}^{2} \]

      5. lower--.f64 N/A

         \[\leadsto \frac{\color{blue}{1 - m}}{v} \cdot {m}^{2} \]

      6. unpow2 N/A

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

      7. lower-*.f64 99.9

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

   7. Applied rewrites 99.9%

      \[\leadsto \color{blue}{\frac{1 - m}{v} \cdot \left(m \cdot m\right)} \]
3. Recombined 2 regimes into one program.

4. Final simplification 99.8%

   \[\leadsto \begin{array}{l} \mathbf{if}\;m \leq 2.2 \cdot 10^{-17}:\\ \;\;\;\;\mathsf{fma}\left(\frac{m}{v}, m, -m\right)\\ \mathbf{else}:\\ \;\;\;\;\left(m \cdot m\right) \cdot \frac{1 - m}{v}\\ \end{array} \]
5. Add Preprocessing

Alternative 2: 51.4% accurate, 0.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;\left(\frac{\left(1 - m\right) \cdot m}{v} - 1\right) \cdot m \leq -5 \cdot 10^{+53}:\\ \;\;\;\;-m\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(\frac{m}{v}, m, -m\right)\\ \end{array} \end{array} \]

FPCore

(FPCore (m v)
 :precision binary64
 (if (<= (* (- (/ (* (- 1.0 m) m) v) 1.0) m) -5e+53)
   (- m)
   (fma (/ m v) m (- m))))

C

double code(double m, double v) {
	double tmp;
	if ((((((1.0 - m) * m) / v) - 1.0) * m) <= -5e+53) {
		tmp = -m;
	} else {
		tmp = fma((m / v), m, -m);
	}
	return tmp;
}

Julia

function code(m, v)
	tmp = 0.0
	if (Float64(Float64(Float64(Float64(Float64(1.0 - m) * m) / v) - 1.0) * m) <= -5e+53)
		tmp = Float64(-m);
	else
		tmp = fma(Float64(m / v), m, Float64(-m));
	end
	return tmp
end

Wolfram

code[m_, v_] := If[LessEqual[N[(N[(N[(N[(N[(1.0 - m), $MachinePrecision] * m), $MachinePrecision] / v), $MachinePrecision] - 1.0), $MachinePrecision] * m), $MachinePrecision], -5e+53], (-m), N[(N[(m / v), $MachinePrecision] * m + (-m)), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if (*.f64 (-.f64 (/.f64 (*.f64 m (-.f64 #s(literal 1 binary64) m)) v) #s(literal 1 binary64)) m) < -5.0000000000000004e53

   1. Initial program 99.9%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in v around inf

      \[\leadsto \color{blue}{-1 \cdot m} \]
   4. Step-by-step derivation

      1. mul-1-neg N/A

         \[\leadsto \color{blue}{\mathsf{neg}\left(m\right)} \]

      2. lower-neg.f64 5.4

         \[\leadsto \color{blue}{-m} \]

   5. Applied rewrites 5.4%

      \[\leadsto \color{blue}{-m} \]

   if -5.0000000000000004e53 < (*.f64 (-.f64 (/.f64 (*.f64 m (-.f64 #s(literal 1 binary64) m)) v) #s(literal 1 binary64)) m)

   1. Initial program 99.7%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in m around 0

      \[\leadsto \color{blue}{m \cdot \left(\frac{m}{v} - 1\right)} \]
   4. Step-by-step derivation

      1. sub-neg N/A

         \[\leadsto m \cdot \color{blue}{\left(\frac{m}{v} + \left(\mathsf{neg}\left(1\right)\right)\right)} \]

      2. metadata-eval N/A

         \[\leadsto m \cdot \left(\frac{m}{v} + \color{blue}{-1}\right) \]

      3. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\frac{m}{v} \cdot m + -1 \cdot m} \]

      4. lower-fma.f64 N/A

         \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -1 \cdot m\right)} \]

      5. lower-/.f64 N/A

         \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{m}{v}}, m, -1 \cdot m\right) \]

      6. mul-1-neg N/A

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{\mathsf{neg}\left(m\right)}\right) \]

      7. lower-neg.f64 97.3

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{-m}\right) \]

   5. Applied rewrites 97.3%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -m\right)} \]
3. Recombined 2 regimes into one program.

4. Final simplification 53.5%

   \[\leadsto \begin{array}{l} \mathbf{if}\;\left(\frac{\left(1 - m\right) \cdot m}{v} - 1\right) \cdot m \leq -5 \cdot 10^{+53}:\\ \;\;\;\;-m\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(\frac{m}{v}, m, -m\right)\\ \end{array} \]
5. Add Preprocessing

Alternative 3: 49.3% accurate, 0.6× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;\left(\frac{\left(1 - m\right) \cdot m}{v} - 1\right) \cdot m \leq -1 \cdot 10^{-308}:\\ \;\;\;\;-m\\ \mathbf{else}:\\ \;\;\;\;\frac{m}{v} \cdot m\\ \end{array} \end{array} \]

FPCore

(FPCore (m v)
 :precision binary64
 (if (<= (* (- (/ (* (- 1.0 m) m) v) 1.0) m) -1e-308) (- m) (* (/ m v) m)))

C

double code(double m, double v) {
	double tmp;
	if ((((((1.0 - m) * m) / v) - 1.0) * m) <= -1e-308) {
		tmp = -m;
	} else {
		tmp = (m / v) * m;
	}
	return tmp;
}

Fortran

real(8) function code(m, v)
    real(8), intent (in) :: m
    real(8), intent (in) :: v
    real(8) :: tmp
    if ((((((1.0d0 - m) * m) / v) - 1.0d0) * m) <= (-1d-308)) then
        tmp = -m
    else
        tmp = (m / v) * m
    end if
    code = tmp
end function

Java

public static double code(double m, double v) {
	double tmp;
	if ((((((1.0 - m) * m) / v) - 1.0) * m) <= -1e-308) {
		tmp = -m;
	} else {
		tmp = (m / v) * m;
	}
	return tmp;
}

Python

def code(m, v):
	tmp = 0
	if (((((1.0 - m) * m) / v) - 1.0) * m) <= -1e-308:
		tmp = -m
	else:
		tmp = (m / v) * m
	return tmp

Julia

function code(m, v)
	tmp = 0.0
	if (Float64(Float64(Float64(Float64(Float64(1.0 - m) * m) / v) - 1.0) * m) <= -1e-308)
		tmp = Float64(-m);
	else
		tmp = Float64(Float64(m / v) * m);
	end
	return tmp
end

MATLAB

function tmp_2 = code(m, v)
	tmp = 0.0;
	if ((((((1.0 - m) * m) / v) - 1.0) * m) <= -1e-308)
		tmp = -m;
	else
		tmp = (m / v) * m;
	end
	tmp_2 = tmp;
end

Wolfram

code[m_, v_] := If[LessEqual[N[(N[(N[(N[(N[(1.0 - m), $MachinePrecision] * m), $MachinePrecision] / v), $MachinePrecision] - 1.0), $MachinePrecision] * m), $MachinePrecision], -1e-308], (-m), N[(N[(m / v), $MachinePrecision] * m), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if (*.f64 (-.f64 (/.f64 (*.f64 m (-.f64 #s(literal 1 binary64) m)) v) #s(literal 1 binary64)) m) < -9.9999999999999991e-309

   1. Initial program 99.9%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in v around inf

      \[\leadsto \color{blue}{-1 \cdot m} \]
   4. Step-by-step derivation

      1. mul-1-neg N/A

         \[\leadsto \color{blue}{\mathsf{neg}\left(m\right)} \]

      2. lower-neg.f64 35.7

         \[\leadsto \color{blue}{-m} \]

   5. Applied rewrites 35.7%

      \[\leadsto \color{blue}{-m} \]

   if -9.9999999999999991e-309 < (*.f64 (-.f64 (/.f64 (*.f64 m (-.f64 #s(literal 1 binary64) m)) v) #s(literal 1 binary64)) m)

   1. Initial program 99.5%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in v around 0

      \[\leadsto \color{blue}{\frac{m \cdot \left(1 - m\right)}{v}} \cdot m \]
   4. Step-by-step derivation

      1. lower-/.f64 N/A

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

      2. *-commutative N/A

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

      3. lower-*.f64 N/A

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

      4. lower--.f64 96.5

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

   5. Applied rewrites 96.5%

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

   6. Taylor expanded in m around 0

      \[\leadsto \frac{m}{\color{blue}{v}} \cdot m \]
   7. Step-by-step derivation

   8. Applied rewrites 92.2%

      \[\leadsto \frac{m}{\color{blue}{v}} \cdot m \]
3. Recombined 2 regimes into one program.

4. Final simplification 52.0%

   \[\leadsto \begin{array}{l} \mathbf{if}\;\left(\frac{\left(1 - m\right) \cdot m}{v} - 1\right) \cdot m \leq -1 \cdot 10^{-308}:\\ \;\;\;\;-m\\ \mathbf{else}:\\ \;\;\;\;\frac{m}{v} \cdot m\\ \end{array} \]
5. Add Preprocessing

Alternative 4: 97.8% accurate, 0.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;m \leq 1:\\ \;\;\;\;\mathsf{fma}\left(\frac{m}{v}, m, -m\right)\\ \mathbf{else}:\\ \;\;\;\;\left(\frac{-m}{v} \cdot m\right) \cdot m\\ \end{array} \end{array} \]

FPCore

(FPCore (m v)
 :precision binary64
 (if (<= m 1.0) (fma (/ m v) m (- m)) (* (* (/ (- m) v) m) m)))

C

double code(double m, double v) {
	double tmp;
	if (m <= 1.0) {
		tmp = fma((m / v), m, -m);
	} else {
		tmp = ((-m / v) * m) * m;
	}
	return tmp;
}

Julia

function code(m, v)
	tmp = 0.0
	if (m <= 1.0)
		tmp = fma(Float64(m / v), m, Float64(-m));
	else
		tmp = Float64(Float64(Float64(Float64(-m) / v) * m) * m);
	end
	return tmp
end

Wolfram

code[m_, v_] := If[LessEqual[m, 1.0], N[(N[(m / v), $MachinePrecision] * m + (-m)), $MachinePrecision], N[(N[(N[((-m) / v), $MachinePrecision] * m), $MachinePrecision] * m), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if m < 1

   1. Initial program 99.7%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in m around 0

      \[\leadsto \color{blue}{m \cdot \left(\frac{m}{v} - 1\right)} \]
   4. Step-by-step derivation

      1. sub-neg N/A

         \[\leadsto m \cdot \color{blue}{\left(\frac{m}{v} + \left(\mathsf{neg}\left(1\right)\right)\right)} \]

      2. metadata-eval N/A

         \[\leadsto m \cdot \left(\frac{m}{v} + \color{blue}{-1}\right) \]

      3. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\frac{m}{v} \cdot m + -1 \cdot m} \]

      4. lower-fma.f64 N/A

         \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -1 \cdot m\right)} \]

      5. lower-/.f64 N/A

         \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{m}{v}}, m, -1 \cdot m\right) \]

      6. mul-1-neg N/A

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{\mathsf{neg}\left(m\right)}\right) \]

      7. lower-neg.f64 97.3

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{-m}\right) \]

   5. Applied rewrites 97.3%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -m\right)} \]

   if 1 < m

   1. Initial program 99.9%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in m around inf

      \[\leadsto \color{blue}{\left(-1 \cdot \frac{{m}^{2}}{v}\right)} \cdot m \]
   4. Step-by-step derivation

      1. unpow2 N/A

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

      2. associate-/l* N/A

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

      3. associate-*r* N/A

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

      4. associate-*r/ N/A

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

      5. associate-*l/ N/A

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

      6. associate-*r/ N/A

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

      7. lower-*.f64 N/A

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

      8. associate-*r/ N/A

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

      9. lower-/.f64 N/A

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

      10. mul-1-neg N/A

          \[\leadsto \left(\frac{\color{blue}{\mathsf{neg}\left(m\right)}}{v} \cdot m\right) \cdot m \]

      11. lower-neg.f64 97.1

          \[\leadsto \left(\frac{\color{blue}{-m}}{v} \cdot m\right) \cdot m \]

   5. Applied rewrites 97.1%

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

Alternative 5: 97.8% accurate, 0.9× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;m \leq 1:\\ \;\;\;\;\mathsf{fma}\left(\frac{m}{v}, m, -m\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{-m}{v} \cdot \left(m \cdot m\right)\\ \end{array} \end{array} \]

FPCore

(FPCore (m v)
 :precision binary64
 (if (<= m 1.0) (fma (/ m v) m (- m)) (* (/ (- m) v) (* m m))))

C

double code(double m, double v) {
	double tmp;
	if (m <= 1.0) {
		tmp = fma((m / v), m, -m);
	} else {
		tmp = (-m / v) * (m * m);
	}
	return tmp;
}

Julia

function code(m, v)
	tmp = 0.0
	if (m <= 1.0)
		tmp = fma(Float64(m / v), m, Float64(-m));
	else
		tmp = Float64(Float64(Float64(-m) / v) * Float64(m * m));
	end
	return tmp
end

Wolfram

code[m_, v_] := If[LessEqual[m, 1.0], N[(N[(m / v), $MachinePrecision] * m + (-m)), $MachinePrecision], N[(N[((-m) / v), $MachinePrecision] * N[(m * m), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if m < 1

   1. Initial program 99.7%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in m around 0

      \[\leadsto \color{blue}{m \cdot \left(\frac{m}{v} - 1\right)} \]
   4. Step-by-step derivation

      1. sub-neg N/A

         \[\leadsto m \cdot \color{blue}{\left(\frac{m}{v} + \left(\mathsf{neg}\left(1\right)\right)\right)} \]

      2. metadata-eval N/A

         \[\leadsto m \cdot \left(\frac{m}{v} + \color{blue}{-1}\right) \]

      3. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\frac{m}{v} \cdot m + -1 \cdot m} \]

      4. lower-fma.f64 N/A

         \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -1 \cdot m\right)} \]

      5. lower-/.f64 N/A

         \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{m}{v}}, m, -1 \cdot m\right) \]

      6. mul-1-neg N/A

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{\mathsf{neg}\left(m\right)}\right) \]

      7. lower-neg.f64 97.3

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{-m}\right) \]

   5. Applied rewrites 97.3%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -m\right)} \]

   if 1 < m

   1. Initial program 99.9%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing
   3. Step-by-step derivation

      1. lift-*.f64 N/A

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

      2. *-commutative N/A

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

      3. lift--.f64 N/A

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

      4. flip3-- N/A

         \[\leadsto m \cdot \color{blue}{\frac{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}} \]

      5. clear-num N/A

         \[\leadsto m \cdot \color{blue}{\frac{1}{\frac{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}}} \]

      6. un-div-inv N/A

         \[\leadsto \color{blue}{\frac{m}{\frac{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}}} \]

      7. lower-/.f64 N/A

         \[\leadsto \color{blue}{\frac{m}{\frac{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}}} \]

      8. clear-num N/A

         \[\leadsto \frac{m}{\color{blue}{\frac{1}{\frac{{\left(\frac{m \cdot \left(1 - m\right)}{v}\right)}^{3} - {1}^{3}}{\frac{m \cdot \left(1 - m\right)}{v} \cdot \frac{m \cdot \left(1 - m\right)}{v} + \left(1 \cdot 1 + \frac{m \cdot \left(1 - m\right)}{v} \cdot 1\right)}}}} \]

      9. flip3-- N/A

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

      10. lift--.f64 N/A

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

   4. Applied rewrites 99.9%

      \[\leadsto \color{blue}{\frac{m}{\frac{1}{\mathsf{fma}\left(\frac{m}{v}, 1 - m, -1\right)}}} \]

   5. Taylor expanded in v around 0

      \[\leadsto \color{blue}{\frac{{m}^{2} \cdot \left(1 - m\right)}{v}} \]
   6. Step-by-step derivation

      1. associate-/l* N/A

         \[\leadsto \color{blue}{{m}^{2} \cdot \frac{1 - m}{v}} \]

      2. *-commutative N/A

         \[\leadsto \color{blue}{\frac{1 - m}{v} \cdot {m}^{2}} \]

      3. lower-*.f64 N/A

         \[\leadsto \color{blue}{\frac{1 - m}{v} \cdot {m}^{2}} \]

      4. lower-/.f64 N/A

         \[\leadsto \color{blue}{\frac{1 - m}{v}} \cdot {m}^{2} \]

      5. lower--.f64 N/A

         \[\leadsto \frac{\color{blue}{1 - m}}{v} \cdot {m}^{2} \]

      6. unpow2 N/A

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

      7. lower-*.f64 99.9

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

   7. Applied rewrites 99.9%

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

   8. Taylor expanded in m around inf

      \[\leadsto \frac{-1 \cdot m}{v} \cdot \left(m \cdot m\right) \]
   9. Step-by-step derivation

   10. Applied rewrites 97.1%

       \[\leadsto \frac{-m}{v} \cdot \left(m \cdot m\right) \]
3. Recombined 2 regimes into one program.
4. Add Preprocessing

Alternative 6: 99.8% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(1 - m, \frac{m}{v} \cdot m, -m\right) \end{array} \]

FPCore

(FPCore (m v) :precision binary64 (fma (- 1.0 m) (* (/ m v) m) (- m)))

C

double code(double m, double v) {
	return fma((1.0 - m), ((m / v) * m), -m);
}

Julia

function code(m, v)
	return fma(Float64(1.0 - m), Float64(Float64(m / v) * m), Float64(-m))
end

Wolfram

code[m_, v_] := N[(N[(1.0 - m), $MachinePrecision] * N[(N[(m / v), $MachinePrecision] * m), $MachinePrecision] + (-m)), $MachinePrecision]

Derivation

1. Initial program 99.8%

   \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
2. Add Preprocessing
3. Step-by-step derivation

   1. lift-*.f64 N/A

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

   2. *-commutative N/A

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

   3. lift--.f64 N/A

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

   4. sub-neg N/A

      \[\leadsto m \cdot \color{blue}{\left(\frac{m \cdot \left(1 - m\right)}{v} + \left(\mathsf{neg}\left(1\right)\right)\right)} \]

   5. distribute-rgt-in N/A

      \[\leadsto \color{blue}{\frac{m \cdot \left(1 - m\right)}{v} \cdot m + \left(\mathsf{neg}\left(1\right)\right) \cdot m} \]

   6. metadata-eval N/A

      \[\leadsto \frac{m \cdot \left(1 - m\right)}{v} \cdot m + \color{blue}{-1} \cdot m \]

   7. neg-mul-1 N/A

      \[\leadsto \frac{m \cdot \left(1 - m\right)}{v} \cdot m + \color{blue}{\left(\mathsf{neg}\left(m\right)\right)} \]

   8. lift-/.f64 N/A

      \[\leadsto \color{blue}{\frac{m \cdot \left(1 - m\right)}{v}} \cdot m + \left(\mathsf{neg}\left(m\right)\right) \]

   9. lift-*.f64 N/A

      \[\leadsto \frac{\color{blue}{m \cdot \left(1 - m\right)}}{v} \cdot m + \left(\mathsf{neg}\left(m\right)\right) \]

   10. *-commutative N/A

       \[\leadsto \frac{\color{blue}{\left(1 - m\right) \cdot m}}{v} \cdot m + \left(\mathsf{neg}\left(m\right)\right) \]

   11. associate-/l* N/A

       \[\leadsto \color{blue}{\left(\left(1 - m\right) \cdot \frac{m}{v}\right)} \cdot m + \left(\mathsf{neg}\left(m\right)\right) \]

   12. associate-*l* N/A

       \[\leadsto \color{blue}{\left(1 - m\right) \cdot \left(\frac{m}{v} \cdot m\right)} + \left(\mathsf{neg}\left(m\right)\right) \]

   13. lower-fma.f64 N/A

       \[\leadsto \color{blue}{\mathsf{fma}\left(1 - m, \frac{m}{v} \cdot m, \mathsf{neg}\left(m\right)\right)} \]

   14. lower-*.f64 N/A

       \[\leadsto \mathsf{fma}\left(1 - m, \color{blue}{\frac{m}{v} \cdot m}, \mathsf{neg}\left(m\right)\right) \]

   15. lower-/.f64 N/A

       \[\leadsto \mathsf{fma}\left(1 - m, \color{blue}{\frac{m}{v}} \cdot m, \mathsf{neg}\left(m\right)\right) \]

   16. lower-neg.f64 99.8

       \[\leadsto \mathsf{fma}\left(1 - m, \frac{m}{v} \cdot m, \color{blue}{-m}\right) \]

4. Applied rewrites 99.8%

   \[\leadsto \color{blue}{\mathsf{fma}\left(1 - m, \frac{m}{v} \cdot m, -m\right)} \]
5. Add Preprocessing

Alternative 7: 74.7% accurate, 1.1× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;m \leq 1:\\ \;\;\;\;\mathsf{fma}\left(\frac{m}{v}, m, -m\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{\left(-m\right) \cdot m}{m}\\ \end{array} \end{array} \]

FPCore

(FPCore (m v)
 :precision binary64
 (if (<= m 1.0) (fma (/ m v) m (- m)) (/ (* (- m) m) m)))

C

double code(double m, double v) {
	double tmp;
	if (m <= 1.0) {
		tmp = fma((m / v), m, -m);
	} else {
		tmp = (-m * m) / m;
	}
	return tmp;
}

Julia

function code(m, v)
	tmp = 0.0
	if (m <= 1.0)
		tmp = fma(Float64(m / v), m, Float64(-m));
	else
		tmp = Float64(Float64(Float64(-m) * m) / m);
	end
	return tmp
end

Wolfram

code[m_, v_] := If[LessEqual[m, 1.0], N[(N[(m / v), $MachinePrecision] * m + (-m)), $MachinePrecision], N[(N[((-m) * m), $MachinePrecision] / m), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if m < 1

   1. Initial program 99.7%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in m around 0

      \[\leadsto \color{blue}{m \cdot \left(\frac{m}{v} - 1\right)} \]
   4. Step-by-step derivation

      1. sub-neg N/A

         \[\leadsto m \cdot \color{blue}{\left(\frac{m}{v} + \left(\mathsf{neg}\left(1\right)\right)\right)} \]

      2. metadata-eval N/A

         \[\leadsto m \cdot \left(\frac{m}{v} + \color{blue}{-1}\right) \]

      3. distribute-rgt-in N/A

         \[\leadsto \color{blue}{\frac{m}{v} \cdot m + -1 \cdot m} \]

      4. lower-fma.f64 N/A

         \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -1 \cdot m\right)} \]

      5. lower-/.f64 N/A

         \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{m}{v}}, m, -1 \cdot m\right) \]

      6. mul-1-neg N/A

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{\mathsf{neg}\left(m\right)}\right) \]

      7. lower-neg.f64 97.3

         \[\leadsto \mathsf{fma}\left(\frac{m}{v}, m, \color{blue}{-m}\right) \]

   5. Applied rewrites 97.3%

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, m, -m\right)} \]

   if 1 < m

   1. Initial program 99.9%

      \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
   2. Add Preprocessing

   3. Taylor expanded in v around inf

      \[\leadsto \color{blue}{-1 \cdot m} \]
   4. Step-by-step derivation

      1. mul-1-neg N/A

         \[\leadsto \color{blue}{\mathsf{neg}\left(m\right)} \]

      2. lower-neg.f64 5.4

         \[\leadsto \color{blue}{-m} \]

   5. Applied rewrites 5.4%

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

   7. Applied rewrites 41.9%

      \[\leadsto \frac{0 - m \cdot m}{\color{blue}{0 + m}} \]
3. Recombined 2 regimes into one program.

4. Final simplification 70.9%

   \[\leadsto \begin{array}{l} \mathbf{if}\;m \leq 1:\\ \;\;\;\;\mathsf{fma}\left(\frac{m}{v}, m, -m\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{\left(-m\right) \cdot m}{m}\\ \end{array} \]
5. Add Preprocessing

Alternative 8: 99.8% accurate, 1.1× speedup

Math

\[\begin{array}{l} \\ \mathsf{fma}\left(\frac{m}{v}, 1 - m, -1\right) \cdot m \end{array} \]

FPCore

(FPCore (m v) :precision binary64 (* (fma (/ m v) (- 1.0 m) -1.0) m))

C

double code(double m, double v) {
	return fma((m / v), (1.0 - m), -1.0) * m;
}

Julia

function code(m, v)
	return Float64(fma(Float64(m / v), Float64(1.0 - m), -1.0) * m)
end

Wolfram

code[m_, v_] := N[(N[(N[(m / v), $MachinePrecision] * N[(1.0 - m), $MachinePrecision] + -1.0), $MachinePrecision] * m), $MachinePrecision]

Derivation

1. Initial program 99.8%

   \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
2. Add Preprocessing
3. Step-by-step derivation

   1. lift--.f64 N/A

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

   2. sub-neg N/A

      \[\leadsto \color{blue}{\left(\frac{m \cdot \left(1 - m\right)}{v} + \left(\mathsf{neg}\left(1\right)\right)\right)} \cdot m \]

   3. lift-/.f64 N/A

      \[\leadsto \left(\color{blue}{\frac{m \cdot \left(1 - m\right)}{v}} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot m \]

   4. lift-*.f64 N/A

      \[\leadsto \left(\frac{\color{blue}{m \cdot \left(1 - m\right)}}{v} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot m \]

   5. *-commutative N/A

      \[\leadsto \left(\frac{\color{blue}{\left(1 - m\right) \cdot m}}{v} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot m \]

   6. associate-/l* N/A

      \[\leadsto \left(\color{blue}{\left(1 - m\right) \cdot \frac{m}{v}} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot m \]

   7. *-commutative N/A

      \[\leadsto \left(\color{blue}{\frac{m}{v} \cdot \left(1 - m\right)} + \left(\mathsf{neg}\left(1\right)\right)\right) \cdot m \]

   8. lower-fma.f64 N/A

      \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, 1 - m, \mathsf{neg}\left(1\right)\right)} \cdot m \]

   9. lower-/.f64 N/A

      \[\leadsto \mathsf{fma}\left(\color{blue}{\frac{m}{v}}, 1 - m, \mathsf{neg}\left(1\right)\right) \cdot m \]

   10. metadata-eval 99.8

       \[\leadsto \mathsf{fma}\left(\frac{m}{v}, 1 - m, \color{blue}{-1}\right) \cdot m \]

4. Applied rewrites 99.8%

   \[\leadsto \color{blue}{\mathsf{fma}\left(\frac{m}{v}, 1 - m, -1\right)} \cdot m \]
5. Add Preprocessing

Alternative 9: 27.9% accurate, 9.3× speedup

Math

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

FPCore

(FPCore (m v) :precision binary64 (- m))

C

double code(double m, double v) {
	return -m;
}

Fortran

real(8) function code(m, v)
    real(8), intent (in) :: m
    real(8), intent (in) :: v
    code = -m
end function

Java

public static double code(double m, double v) {
	return -m;
}

Python

def code(m, v):
	return -m

Julia

function code(m, v)
	return Float64(-m)
end

MATLAB

function tmp = code(m, v)
	tmp = -m;
end

Wolfram

code[m_, v_] := (-m)

Derivation

1. Initial program 99.8%

   \[\left(\frac{m \cdot \left(1 - m\right)}{v} - 1\right) \cdot m \]
2. Add Preprocessing

3. Taylor expanded in v around inf

   \[\leadsto \color{blue}{-1 \cdot m} \]
4. Step-by-step derivation

   1. mul-1-neg N/A

      \[\leadsto \color{blue}{\mathsf{neg}\left(m\right)} \]

   2. lower-neg.f64 26.0

      \[\leadsto \color{blue}{-m} \]

5. Applied rewrites 26.0%

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

Reproduce

herbie shell --seed 2024240 
(FPCore (m v)
  :name "a parameter of renormalized beta distribution"
  :precision binary64
  :pre (and (and (< 0.0 m) (< 0.0 v)) (< v 0.25))
  (* (- (/ (* m (- 1.0 m)) v) 1.0) m))