Quotient of sum of exps

Percentage Accurate: 98.8% → 99.0%

Time: 7.8s

Alternatives: 8

Speedup: 1.0×

• could not determine a ground truth

  1. a = -2.00000000000000007e154
  2. b = -2.00000000000000007e154

Specification

Math

\[\begin{array}{l} \\ \frac{e^{a}}{e^{a} + e^{b}} \end{array} \]

FPCore

(FPCore (a b) :precision binary64 (/ (exp a) (+ (exp a) (exp b))))

C

double code(double a, double b) {
	return exp(a) / (exp(a) + exp(b));
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = exp(a) / (exp(a) + exp(b))
end function

Java

public static double code(double a, double b) {
	return Math.exp(a) / (Math.exp(a) + Math.exp(b));
}

Python

def code(a, b):
	return math.exp(a) / (math.exp(a) + math.exp(b))

Julia

function code(a, b)
	return Float64(exp(a) / Float64(exp(a) + exp(b)))
end

MATLAB

function tmp = code(a, b)
	tmp = exp(a) / (exp(a) + exp(b));
end

Wolfram

code[a_, b_] := N[(N[Exp[a], $MachinePrecision] / N[(N[Exp[a], $MachinePrecision] + N[Exp[b], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Initial Program: 98.8% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \frac{e^{a}}{e^{a} + e^{b}} \end{array} \]

FPCore

(FPCore (a b) :precision binary64 (/ (exp a) (+ (exp a) (exp b))))

C

double code(double a, double b) {
	return exp(a) / (exp(a) + exp(b));
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = exp(a) / (exp(a) + exp(b))
end function

Java

public static double code(double a, double b) {
	return Math.exp(a) / (Math.exp(a) + Math.exp(b));
}

Python

def code(a, b):
	return math.exp(a) / (math.exp(a) + math.exp(b))

Julia

function code(a, b)
	return Float64(exp(a) / Float64(exp(a) + exp(b)))
end

MATLAB

function tmp = code(a, b)
	tmp = exp(a) / (exp(a) + exp(b));
end

Wolfram

code[a_, b_] := N[(N[Exp[a], $MachinePrecision] / N[(N[Exp[a], $MachinePrecision] + N[Exp[b], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Alternative 1: 99.0% accurate, 0.7× speedup

Math

\[\begin{array}{l} \\ e^{\left(0.3333333333333333 \cdot \left(a - \log \left(e^{a} + e^{b}\right)\right)\right) \cdot 3} \end{array} \]

FPCore

(FPCore (a b)
 :precision binary64
 (exp (* (* 0.3333333333333333 (- a (log (+ (exp a) (exp b))))) 3.0)))

C

double code(double a, double b) {
	return exp(((0.3333333333333333 * (a - log((exp(a) + exp(b))))) * 3.0));
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = exp(((0.3333333333333333d0 * (a - log((exp(a) + exp(b))))) * 3.0d0))
end function

Java

public static double code(double a, double b) {
	return Math.exp(((0.3333333333333333 * (a - Math.log((Math.exp(a) + Math.exp(b))))) * 3.0));
}

Python

def code(a, b):
	return math.exp(((0.3333333333333333 * (a - math.log((math.exp(a) + math.exp(b))))) * 3.0))

Julia

function code(a, b)
	return exp(Float64(Float64(0.3333333333333333 * Float64(a - log(Float64(exp(a) + exp(b))))) * 3.0))
end

MATLAB

function tmp = code(a, b)
	tmp = exp(((0.3333333333333333 * (a - log((exp(a) + exp(b))))) * 3.0));
end

Wolfram

code[a_, b_] := N[Exp[N[(N[(0.3333333333333333 * N[(a - N[Log[N[(N[Exp[a], $MachinePrecision] + N[Exp[b], $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision] * 3.0), $MachinePrecision]], $MachinePrecision]

Derivation

1. Initial program 98.4%

   \[\frac{e^{a}}{e^{a} + e^{b}} \]
2. Add Preprocessing
3. Step-by-step derivation

   1. add-cube-cbrt 97.8%

      \[\leadsto \color{blue}{\left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}} \cdot \sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right) \cdot \sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}} \]

   2. pow3 97.8%

      \[\leadsto \color{blue}{{\left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right)}^{3}} \]

   3. pow-to-exp 97.8%

      \[\leadsto \color{blue}{e^{\log \left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right) \cdot 3}} \]

   4. pow1/3 97.9%

      \[\leadsto e^{\log \color{blue}{\left({\left(\frac{e^{a}}{e^{a} + e^{b}}\right)}^{0.3333333333333333}\right)} \cdot 3} \]

   5. log-pow 98.4%

      \[\leadsto e^{\color{blue}{\left(0.3333333333333333 \cdot \log \left(\frac{e^{a}}{e^{a} + e^{b}}\right)\right)} \cdot 3} \]

   6. log-div 98.4%

      \[\leadsto e^{\left(0.3333333333333333 \cdot \color{blue}{\left(\log \left(e^{a}\right) - \log \left(e^{a} + e^{b}\right)\right)}\right) \cdot 3} \]

   7. add-log-exp 98.8%

      \[\leadsto e^{\left(0.3333333333333333 \cdot \left(\color{blue}{a} - \log \left(e^{a} + e^{b}\right)\right)\right) \cdot 3} \]

4. Applied egg-rr 98.8%

   \[\leadsto \color{blue}{e^{\left(0.3333333333333333 \cdot \left(a - \log \left(e^{a} + e^{b}\right)\right)\right) \cdot 3}} \]

5. Final simplification 98.8%

   \[\leadsto e^{\left(0.3333333333333333 \cdot \left(a - \log \left(e^{a} + e^{b}\right)\right)\right) \cdot 3} \]
6. Add Preprocessing

Alternative 2: 98.8% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \frac{e^{a}}{e^{a} + e^{b}} \end{array} \]

FPCore

(FPCore (a b) :precision binary64 (/ (exp a) (+ (exp a) (exp b))))

C

double code(double a, double b) {
	return exp(a) / (exp(a) + exp(b));
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = exp(a) / (exp(a) + exp(b))
end function

Java

public static double code(double a, double b) {
	return Math.exp(a) / (Math.exp(a) + Math.exp(b));
}

Python

def code(a, b):
	return math.exp(a) / (math.exp(a) + math.exp(b))

Julia

function code(a, b)
	return Float64(exp(a) / Float64(exp(a) + exp(b)))
end

MATLAB

function tmp = code(a, b)
	tmp = exp(a) / (exp(a) + exp(b));
end

Wolfram

code[a_, b_] := N[(N[Exp[a], $MachinePrecision] / N[(N[Exp[a], $MachinePrecision] + N[Exp[b], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Derivation

1. Initial program 98.4%

   \[\frac{e^{a}}{e^{a} + e^{b}} \]
2. Add Preprocessing

3. Final simplification 98.4%

   \[\leadsto \frac{e^{a}}{e^{a} + e^{b}} \]
4. Add Preprocessing

Alternative 3: 66.4% accurate, 1.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;e^{a} \leq 0.05:\\ \;\;\;\;e^{a}\\ \mathbf{else}:\\ \;\;\;\;0.5 + a \cdot 0.25\\ \end{array} \end{array} \]

FPCore

(FPCore (a b)
 :precision binary64
 (if (<= (exp a) 0.05) (exp a) (+ 0.5 (* a 0.25))))

C

double code(double a, double b) {
	double tmp;
	if (exp(a) <= 0.05) {
		tmp = exp(a);
	} else {
		tmp = 0.5 + (a * 0.25);
	}
	return tmp;
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8) :: tmp
    if (exp(a) <= 0.05d0) then
        tmp = exp(a)
    else
        tmp = 0.5d0 + (a * 0.25d0)
    end if
    code = tmp
end function

Java

public static double code(double a, double b) {
	double tmp;
	if (Math.exp(a) <= 0.05) {
		tmp = Math.exp(a);
	} else {
		tmp = 0.5 + (a * 0.25);
	}
	return tmp;
}

Python

def code(a, b):
	tmp = 0
	if math.exp(a) <= 0.05:
		tmp = math.exp(a)
	else:
		tmp = 0.5 + (a * 0.25)
	return tmp

Julia

function code(a, b)
	tmp = 0.0
	if (exp(a) <= 0.05)
		tmp = exp(a);
	else
		tmp = Float64(0.5 + Float64(a * 0.25));
	end
	return tmp
end

MATLAB

function tmp_2 = code(a, b)
	tmp = 0.0;
	if (exp(a) <= 0.05)
		tmp = exp(a);
	else
		tmp = 0.5 + (a * 0.25);
	end
	tmp_2 = tmp;
end

Wolfram

code[a_, b_] := If[LessEqual[N[Exp[a], $MachinePrecision], 0.05], N[Exp[a], $MachinePrecision], N[(0.5 + N[(a * 0.25), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if (exp.f64 a) < 0.050000000000000003

   1. Initial program 98.6%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing
   3. Step-by-step derivation

      1. add-cube-cbrt 98.5%

         \[\leadsto \color{blue}{\left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}} \cdot \sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right) \cdot \sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}} \]

      2. pow3 98.5%

         \[\leadsto \color{blue}{{\left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right)}^{3}} \]

      3. pow-to-exp 98.5%

         \[\leadsto \color{blue}{e^{\log \left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right) \cdot 3}} \]

      4. pow1/3 98.5%

         \[\leadsto e^{\log \color{blue}{\left({\left(\frac{e^{a}}{e^{a} + e^{b}}\right)}^{0.3333333333333333}\right)} \cdot 3} \]

      5. log-pow 98.5%

         \[\leadsto e^{\color{blue}{\left(0.3333333333333333 \cdot \log \left(\frac{e^{a}}{e^{a} + e^{b}}\right)\right)} \cdot 3} \]

      6. log-div 98.5%

         \[\leadsto e^{\left(0.3333333333333333 \cdot \color{blue}{\left(\log \left(e^{a}\right) - \log \left(e^{a} + e^{b}\right)\right)}\right) \cdot 3} \]

      7. add-log-exp 98.6%

         \[\leadsto e^{\left(0.3333333333333333 \cdot \left(\color{blue}{a} - \log \left(e^{a} + e^{b}\right)\right)\right) \cdot 3} \]

   4. Applied egg-rr 98.6%

      \[\leadsto \color{blue}{e^{\left(0.3333333333333333 \cdot \left(a - \log \left(e^{a} + e^{b}\right)\right)\right) \cdot 3}} \]

   5. Taylor expanded in a around inf 96.2%

      \[\leadsto e^{\color{blue}{a}} \]

   if 0.050000000000000003 < (exp.f64 a)

   1. Initial program 98.4%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in b around 0 57.5%

      \[\leadsto \color{blue}{\frac{e^{a}}{1 + e^{a}}} \]

   4. Taylor expanded in a around 0 57.1%

      \[\leadsto \color{blue}{0.5 + 0.25 \cdot a} \]
   5. Step-by-step derivation

      1. *-commutative 57.1%

         \[\leadsto 0.5 + \color{blue}{a \cdot 0.25} \]

   6. Simplified 57.1%

      \[\leadsto \color{blue}{0.5 + a \cdot 0.25} \]
3. Recombined 2 regimes into one program.

4. Final simplification 67.8%

   \[\leadsto \begin{array}{l} \mathbf{if}\;e^{a} \leq 0.05:\\ \;\;\;\;e^{a}\\ \mathbf{else}:\\ \;\;\;\;0.5 + a \cdot 0.25\\ \end{array} \]
5. Add Preprocessing

Alternative 4: 98.3% accurate, 2.8× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;a \leq -720000000000:\\ \;\;\;\;e^{a}\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{e^{b} + 1}\\ \end{array} \end{array} \]

FPCore

(FPCore (a b)
 :precision binary64
 (if (<= a -720000000000.0) (exp a) (/ 1.0 (+ (exp b) 1.0))))

C

double code(double a, double b) {
	double tmp;
	if (a <= -720000000000.0) {
		tmp = exp(a);
	} else {
		tmp = 1.0 / (exp(b) + 1.0);
	}
	return tmp;
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8) :: tmp
    if (a <= (-720000000000.0d0)) then
        tmp = exp(a)
    else
        tmp = 1.0d0 / (exp(b) + 1.0d0)
    end if
    code = tmp
end function

Java

public static double code(double a, double b) {
	double tmp;
	if (a <= -720000000000.0) {
		tmp = Math.exp(a);
	} else {
		tmp = 1.0 / (Math.exp(b) + 1.0);
	}
	return tmp;
}

Python

def code(a, b):
	tmp = 0
	if a <= -720000000000.0:
		tmp = math.exp(a)
	else:
		tmp = 1.0 / (math.exp(b) + 1.0)
	return tmp

Julia

function code(a, b)
	tmp = 0.0
	if (a <= -720000000000.0)
		tmp = exp(a);
	else
		tmp = Float64(1.0 / Float64(exp(b) + 1.0));
	end
	return tmp
end

MATLAB

function tmp_2 = code(a, b)
	tmp = 0.0;
	if (a <= -720000000000.0)
		tmp = exp(a);
	else
		tmp = 1.0 / (exp(b) + 1.0);
	end
	tmp_2 = tmp;
end

Wolfram

code[a_, b_] := If[LessEqual[a, -720000000000.0], N[Exp[a], $MachinePrecision], N[(1.0 / N[(N[Exp[b], $MachinePrecision] + 1.0), $MachinePrecision]), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if a < -7.2e11

   1. Initial program 100.0%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing
   3. Step-by-step derivation

      1. add-cube-cbrt 100.0%

         \[\leadsto \color{blue}{\left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}} \cdot \sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right) \cdot \sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}} \]

      2. pow3 100.0%

         \[\leadsto \color{blue}{{\left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right)}^{3}} \]

      3. pow-to-exp 100.0%

         \[\leadsto \color{blue}{e^{\log \left(\sqrt[3]{\frac{e^{a}}{e^{a} + e^{b}}}\right) \cdot 3}} \]

      4. pow1/3 100.0%

         \[\leadsto e^{\log \color{blue}{\left({\left(\frac{e^{a}}{e^{a} + e^{b}}\right)}^{0.3333333333333333}\right)} \cdot 3} \]

      5. log-pow 100.0%

         \[\leadsto e^{\color{blue}{\left(0.3333333333333333 \cdot \log \left(\frac{e^{a}}{e^{a} + e^{b}}\right)\right)} \cdot 3} \]

      6. log-div 100.0%

         \[\leadsto e^{\left(0.3333333333333333 \cdot \color{blue}{\left(\log \left(e^{a}\right) - \log \left(e^{a} + e^{b}\right)\right)}\right) \cdot 3} \]

      7. add-log-exp 100.0%

         \[\leadsto e^{\left(0.3333333333333333 \cdot \left(\color{blue}{a} - \log \left(e^{a} + e^{b}\right)\right)\right) \cdot 3} \]

   4. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{e^{\left(0.3333333333333333 \cdot \left(a - \log \left(e^{a} + e^{b}\right)\right)\right) \cdot 3}} \]

   5. Taylor expanded in a around inf 100.0%

      \[\leadsto e^{\color{blue}{a}} \]

   if -7.2e11 < a

   1. Initial program 97.9%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in a around 0 97.0%

      \[\leadsto \color{blue}{\frac{1}{1 + e^{b}}} \]
3. Recombined 2 regimes into one program.

4. Final simplification 97.7%

   \[\leadsto \begin{array}{l} \mathbf{if}\;a \leq -720000000000:\\ \;\;\;\;e^{a}\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{e^{b} + 1}\\ \end{array} \]
5. Add Preprocessing

Alternative 5: 52.0% accurate, 14.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;b \leq 260:\\ \;\;\;\;0.5 + a \cdot 0.25\\ \mathbf{elif}\;b \leq 1.35 \cdot 10^{+154}:\\ \;\;\;\;\frac{a}{b}\\ \mathbf{else}:\\ \;\;\;\;\frac{a \cdot \left(-b\right) - b}{b \cdot \left(-b\right)}\\ \end{array} \end{array} \]

FPCore

(FPCore (a b)
 :precision binary64
 (if (<= b 260.0)
   (+ 0.5 (* a 0.25))
   (if (<= b 1.35e+154) (/ a b) (/ (- (* a (- b)) b) (* b (- b))))))

C

double code(double a, double b) {
	double tmp;
	if (b <= 260.0) {
		tmp = 0.5 + (a * 0.25);
	} else if (b <= 1.35e+154) {
		tmp = a / b;
	} else {
		tmp = ((a * -b) - b) / (b * -b);
	}
	return tmp;
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8) :: tmp
    if (b <= 260.0d0) then
        tmp = 0.5d0 + (a * 0.25d0)
    else if (b <= 1.35d+154) then
        tmp = a / b
    else
        tmp = ((a * -b) - b) / (b * -b)
    end if
    code = tmp
end function

Java

public static double code(double a, double b) {
	double tmp;
	if (b <= 260.0) {
		tmp = 0.5 + (a * 0.25);
	} else if (b <= 1.35e+154) {
		tmp = a / b;
	} else {
		tmp = ((a * -b) - b) / (b * -b);
	}
	return tmp;
}

Python

def code(a, b):
	tmp = 0
	if b <= 260.0:
		tmp = 0.5 + (a * 0.25)
	elif b <= 1.35e+154:
		tmp = a / b
	else:
		tmp = ((a * -b) - b) / (b * -b)
	return tmp

Julia

function code(a, b)
	tmp = 0.0
	if (b <= 260.0)
		tmp = Float64(0.5 + Float64(a * 0.25));
	elseif (b <= 1.35e+154)
		tmp = Float64(a / b);
	else
		tmp = Float64(Float64(Float64(a * Float64(-b)) - b) / Float64(b * Float64(-b)));
	end
	return tmp
end

MATLAB

function tmp_2 = code(a, b)
	tmp = 0.0;
	if (b <= 260.0)
		tmp = 0.5 + (a * 0.25);
	elseif (b <= 1.35e+154)
		tmp = a / b;
	else
		tmp = ((a * -b) - b) / (b * -b);
	end
	tmp_2 = tmp;
end

Wolfram

code[a_, b_] := If[LessEqual[b, 260.0], N[(0.5 + N[(a * 0.25), $MachinePrecision]), $MachinePrecision], If[LessEqual[b, 1.35e+154], N[(a / b), $MachinePrecision], N[(N[(N[(a * (-b)), $MachinePrecision] - b), $MachinePrecision] / N[(b * (-b)), $MachinePrecision]), $MachinePrecision]]]

Derivation

1. Split input into 3 regimes

2. if b < 260

   1. Initial program 98.4%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in b around 0 79.1%

      \[\leadsto \color{blue}{\frac{e^{a}}{1 + e^{a}}} \]

   4. Taylor expanded in a around 0 54.9%

      \[\leadsto \color{blue}{0.5 + 0.25 \cdot a} \]
   5. Step-by-step derivation

      1. *-commutative 54.9%

         \[\leadsto 0.5 + \color{blue}{a \cdot 0.25} \]

   6. Simplified 54.9%

      \[\leadsto \color{blue}{0.5 + a \cdot 0.25} \]

   if 260 < b < 1.35000000000000003e154

   1. Initial program 100.0%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in b around 0 36.8%

      \[\leadsto \frac{e^{a}}{\color{blue}{1 + \left(b + e^{a}\right)}} \]
   4. Step-by-step derivation

      1. associate-+r+ 36.8%

         \[\leadsto \frac{e^{a}}{\color{blue}{\left(1 + b\right) + e^{a}}} \]

      2. +-commutative 36.8%

         \[\leadsto \frac{e^{a}}{\color{blue}{e^{a} + \left(1 + b\right)}} \]

   5. Simplified 36.8%

      \[\leadsto \frac{e^{a}}{\color{blue}{e^{a} + \left(1 + b\right)}} \]

   6. Taylor expanded in b around inf 36.8%

      \[\leadsto \color{blue}{\frac{e^{a}}{b}} \]

   7. Taylor expanded in a around 0 3.4%

      \[\leadsto \color{blue}{\frac{1}{b} + \frac{a}{b}} \]

   8. Taylor expanded in a around inf 18.7%

      \[\leadsto \color{blue}{\frac{a}{b}} \]

   if 1.35000000000000003e154 < b

   1. Initial program 96.4%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in b around 0 36.8%

      \[\leadsto \frac{e^{a}}{\color{blue}{1 + \left(b + e^{a}\right)}} \]
   4. Step-by-step derivation

      1. associate-+r+ 36.8%

         \[\leadsto \frac{e^{a}}{\color{blue}{\left(1 + b\right) + e^{a}}} \]

      2. +-commutative 36.8%

         \[\leadsto \frac{e^{a}}{\color{blue}{e^{a} + \left(1 + b\right)}} \]

   5. Simplified 36.8%

      \[\leadsto \frac{e^{a}}{\color{blue}{e^{a} + \left(1 + b\right)}} \]

   6. Taylor expanded in b around inf 36.9%

      \[\leadsto \color{blue}{\frac{e^{a}}{b}} \]

   7. Taylor expanded in a around 0 6.1%

      \[\leadsto \color{blue}{\frac{1}{b} + \frac{a}{b}} \]
   8. Step-by-step derivation

      1. frac-2neg 6.1%

         \[\leadsto \color{blue}{\frac{-1}{-b}} + \frac{a}{b} \]

      2. metadata-eval 6.1%

         \[\leadsto \frac{\color{blue}{-1}}{-b} + \frac{a}{b} \]

      3. frac-add 75.0%

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

      4. neg-mul-1 75.0%

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

   9. Applied egg-rr 75.0%

      \[\leadsto \color{blue}{\frac{\left(-b\right) + \left(-b\right) \cdot a}{\left(-b\right) \cdot b}} \]
3. Recombined 3 regimes into one program.

4. Final simplification 52.2%

   \[\leadsto \begin{array}{l} \mathbf{if}\;b \leq 260:\\ \;\;\;\;0.5 + a \cdot 0.25\\ \mathbf{elif}\;b \leq 1.35 \cdot 10^{+154}:\\ \;\;\;\;\frac{a}{b}\\ \mathbf{else}:\\ \;\;\;\;\frac{a \cdot \left(-b\right) - b}{b \cdot \left(-b\right)}\\ \end{array} \]
5. Add Preprocessing

Alternative 6: 48.8% accurate, 30.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;b \leq 320:\\ \;\;\;\;0.5 + a \cdot 0.25\\ \mathbf{else}:\\ \;\;\;\;\frac{a}{b}\\ \end{array} \end{array} \]

FPCore

(FPCore (a b) :precision binary64 (if (<= b 320.0) (+ 0.5 (* a 0.25)) (/ a b)))

C

double code(double a, double b) {
	double tmp;
	if (b <= 320.0) {
		tmp = 0.5 + (a * 0.25);
	} else {
		tmp = a / b;
	}
	return tmp;
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8) :: tmp
    if (b <= 320.0d0) then
        tmp = 0.5d0 + (a * 0.25d0)
    else
        tmp = a / b
    end if
    code = tmp
end function

Java

public static double code(double a, double b) {
	double tmp;
	if (b <= 320.0) {
		tmp = 0.5 + (a * 0.25);
	} else {
		tmp = a / b;
	}
	return tmp;
}

Python

def code(a, b):
	tmp = 0
	if b <= 320.0:
		tmp = 0.5 + (a * 0.25)
	else:
		tmp = a / b
	return tmp

Julia

function code(a, b)
	tmp = 0.0
	if (b <= 320.0)
		tmp = Float64(0.5 + Float64(a * 0.25));
	else
		tmp = Float64(a / b);
	end
	return tmp
end

MATLAB

function tmp_2 = code(a, b)
	tmp = 0.0;
	if (b <= 320.0)
		tmp = 0.5 + (a * 0.25);
	else
		tmp = a / b;
	end
	tmp_2 = tmp;
end

Wolfram

code[a_, b_] := If[LessEqual[b, 320.0], N[(0.5 + N[(a * 0.25), $MachinePrecision]), $MachinePrecision], N[(a / b), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if b < 320

   1. Initial program 98.4%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in b around 0 79.1%

      \[\leadsto \color{blue}{\frac{e^{a}}{1 + e^{a}}} \]

   4. Taylor expanded in a around 0 54.9%

      \[\leadsto \color{blue}{0.5 + 0.25 \cdot a} \]
   5. Step-by-step derivation

      1. *-commutative 54.9%

         \[\leadsto 0.5 + \color{blue}{a \cdot 0.25} \]

   6. Simplified 54.9%

      \[\leadsto \color{blue}{0.5 + a \cdot 0.25} \]

   if 320 < b

   1. Initial program 98.4%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in b around 0 36.8%

      \[\leadsto \frac{e^{a}}{\color{blue}{1 + \left(b + e^{a}\right)}} \]
   4. Step-by-step derivation

      1. associate-+r+ 36.8%

         \[\leadsto \frac{e^{a}}{\color{blue}{\left(1 + b\right) + e^{a}}} \]

      2. +-commutative 36.8%

         \[\leadsto \frac{e^{a}}{\color{blue}{e^{a} + \left(1 + b\right)}} \]

   5. Simplified 36.8%

      \[\leadsto \frac{e^{a}}{\color{blue}{e^{a} + \left(1 + b\right)}} \]

   6. Taylor expanded in b around inf 36.8%

      \[\leadsto \color{blue}{\frac{e^{a}}{b}} \]

   7. Taylor expanded in a around 0 4.6%

      \[\leadsto \color{blue}{\frac{1}{b} + \frac{a}{b}} \]

   8. Taylor expanded in a around inf 30.5%

      \[\leadsto \color{blue}{\frac{a}{b}} \]
3. Recombined 2 regimes into one program.

4. Final simplification 48.9%

   \[\leadsto \begin{array}{l} \mathbf{if}\;b \leq 320:\\ \;\;\;\;0.5 + a \cdot 0.25\\ \mathbf{else}:\\ \;\;\;\;\frac{a}{b}\\ \end{array} \]
5. Add Preprocessing

Alternative 7: 48.6% accurate, 38.1× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;b \leq 1.35 \cdot 10^{+28}:\\ \;\;\;\;0.5\\ \mathbf{else}:\\ \;\;\;\;\frac{a}{b}\\ \end{array} \end{array} \]

FPCore

(FPCore (a b) :precision binary64 (if (<= b 1.35e+28) 0.5 (/ a b)))

C

double code(double a, double b) {
	double tmp;
	if (b <= 1.35e+28) {
		tmp = 0.5;
	} else {
		tmp = a / b;
	}
	return tmp;
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    real(8) :: tmp
    if (b <= 1.35d+28) then
        tmp = 0.5d0
    else
        tmp = a / b
    end if
    code = tmp
end function

Java

public static double code(double a, double b) {
	double tmp;
	if (b <= 1.35e+28) {
		tmp = 0.5;
	} else {
		tmp = a / b;
	}
	return tmp;
}

Python

def code(a, b):
	tmp = 0
	if b <= 1.35e+28:
		tmp = 0.5
	else:
		tmp = a / b
	return tmp

Julia

function code(a, b)
	tmp = 0.0
	if (b <= 1.35e+28)
		tmp = 0.5;
	else
		tmp = Float64(a / b);
	end
	return tmp
end

MATLAB

function tmp_2 = code(a, b)
	tmp = 0.0;
	if (b <= 1.35e+28)
		tmp = 0.5;
	else
		tmp = a / b;
	end
	tmp_2 = tmp;
end

Wolfram

code[a_, b_] := If[LessEqual[b, 1.35e+28], 0.5, N[(a / b), $MachinePrecision]]

Derivation

1. Split input into 2 regimes

2. if b < 1.3500000000000001e28

   1. Initial program 98.4%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in a around 0 74.2%

      \[\leadsto \color{blue}{\frac{1}{1 + e^{b}}} \]

   4. Taylor expanded in b around 0 54.0%

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

   if 1.3500000000000001e28 < b

   1. Initial program 98.4%

      \[\frac{e^{a}}{e^{a} + e^{b}} \]
   2. Add Preprocessing

   3. Taylor expanded in b around 0 34.7%

      \[\leadsto \frac{e^{a}}{\color{blue}{1 + \left(b + e^{a}\right)}} \]
   4. Step-by-step derivation

      1. associate-+r+ 34.7%

         \[\leadsto \frac{e^{a}}{\color{blue}{\left(1 + b\right) + e^{a}}} \]

      2. +-commutative 34.7%

         \[\leadsto \frac{e^{a}}{\color{blue}{e^{a} + \left(1 + b\right)}} \]

   5. Simplified 34.7%

      \[\leadsto \frac{e^{a}}{\color{blue}{e^{a} + \left(1 + b\right)}} \]

   6. Taylor expanded in b around inf 34.8%

      \[\leadsto \color{blue}{\frac{e^{a}}{b}} \]

   7. Taylor expanded in a around 0 4.7%

      \[\leadsto \color{blue}{\frac{1}{b} + \frac{a}{b}} \]

   8. Taylor expanded in a around inf 31.4%

      \[\leadsto \color{blue}{\frac{a}{b}} \]
3. Recombined 2 regimes into one program.

4. Final simplification 48.6%

   \[\leadsto \begin{array}{l} \mathbf{if}\;b \leq 1.35 \cdot 10^{+28}:\\ \;\;\;\;0.5\\ \mathbf{else}:\\ \;\;\;\;\frac{a}{b}\\ \end{array} \]
5. Add Preprocessing

Alternative 8: 40.3% accurate, 305.0× speedup

Math

\[\begin{array}{l} \\ 0.5 \end{array} \]

FPCore

(FPCore (a b) :precision binary64 0.5)

C

double code(double a, double b) {
	return 0.5;
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = 0.5d0
end function

Java

public static double code(double a, double b) {
	return 0.5;
}

Python

def code(a, b):
	return 0.5

Julia

function code(a, b)
	return 0.5
end

MATLAB

function tmp = code(a, b)
	tmp = 0.5;
end

Wolfram

code[a_, b_] := 0.5

Derivation

1. Initial program 98.4%

   \[\frac{e^{a}}{e^{a} + e^{b}} \]
2. Add Preprocessing

3. Taylor expanded in a around 0 80.3%

   \[\leadsto \color{blue}{\frac{1}{1 + e^{b}}} \]

4. Taylor expanded in b around 0 41.9%

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

5. Final simplification 41.9%

   \[\leadsto 0.5 \]
6. Add Preprocessing

Developer target: 100.0% accurate, 2.9× speedup

Math

\[\begin{array}{l} \\ \frac{1}{1 + e^{b - a}} \end{array} \]

FPCore

(FPCore (a b) :precision binary64 (/ 1.0 (+ 1.0 (exp (- b a)))))

C

double code(double a, double b) {
	return 1.0 / (1.0 + exp((b - a)));
}

Fortran

real(8) function code(a, b)
    real(8), intent (in) :: a
    real(8), intent (in) :: b
    code = 1.0d0 / (1.0d0 + exp((b - a)))
end function

Java

public static double code(double a, double b) {
	return 1.0 / (1.0 + Math.exp((b - a)));
}

Python

def code(a, b):
	return 1.0 / (1.0 + math.exp((b - a)))

Julia

function code(a, b)
	return Float64(1.0 / Float64(1.0 + exp(Float64(b - a))))
end

MATLAB

function tmp = code(a, b)
	tmp = 1.0 / (1.0 + exp((b - a)));
end

Wolfram

code[a_, b_] := N[(1.0 / N[(1.0 + N[Exp[N[(b - a), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Reproduce

herbie shell --seed 2024017 
(FPCore (a b)
  :name "Quotient of sum of exps"
  :precision binary64

  :herbie-target
  (/ 1.0 (+ 1.0 (exp (- b a))))

  (/ (exp a) (+ (exp a) (exp b))))