x / (x^2 + 1)

Percentage Accurate: 75.1% → 100.0%

Time: 2.8s

Alternatives: 6

Speedup: 1.0×

Specification

Math

\[\begin{array}{l} \\ \frac{x}{x \cdot x + 1} \end{array} \]

FPCore

(FPCore (x) :precision binary64 (/ x (+ (* x x) 1.0)))

C

double code(double x) {
	return x / ((x * x) + 1.0);
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    code = x / ((x * x) + 1.0d0)
end function

Java

public static double code(double x) {
	return x / ((x * x) + 1.0);
}

Python

def code(x):
	return x / ((x * x) + 1.0)

Julia

function code(x)
	return Float64(x / Float64(Float64(x * x) + 1.0))
end

MATLAB

function tmp = code(x)
	tmp = x / ((x * x) + 1.0);
end

Wolfram

code[x_] := N[(x / N[(N[(x * x), $MachinePrecision] + 1.0), $MachinePrecision]), $MachinePrecision]

Initial Program: 75.1% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \frac{x}{x \cdot x + 1} \end{array} \]

FPCore

(FPCore (x) :precision binary64 (/ x (+ (* x x) 1.0)))

C

double code(double x) {
	return x / ((x * x) + 1.0);
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    code = x / ((x * x) + 1.0d0)
end function

Java

public static double code(double x) {
	return x / ((x * x) + 1.0);
}

Python

def code(x):
	return x / ((x * x) + 1.0)

Julia

function code(x)
	return Float64(x / Float64(Float64(x * x) + 1.0))
end

MATLAB

function tmp = code(x)
	tmp = x / ((x * x) + 1.0);
end

Wolfram

code[x_] := N[(x / N[(N[(x * x), $MachinePrecision] + 1.0), $MachinePrecision]), $MachinePrecision]

Alternative 1: 100.0% accurate, 0.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;x \leq -4 \cdot 10^{+23}:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 1000000:\\ \;\;\;\;\frac{x}{{x}^{4} + -1} \cdot \mathsf{fma}\left(x, x, -1\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (if (<= x -4e+23)
   (/ 1.0 x)
   (if (<= x 1000000.0)
     (* (/ x (+ (pow x 4.0) -1.0)) (fma x x -1.0))
     (/ 1.0 x))))

C

double code(double x) {
	double tmp;
	if (x <= -4e+23) {
		tmp = 1.0 / x;
	} else if (x <= 1000000.0) {
		tmp = (x / (pow(x, 4.0) + -1.0)) * fma(x, x, -1.0);
	} else {
		tmp = 1.0 / x;
	}
	return tmp;
}

Julia

function code(x)
	tmp = 0.0
	if (x <= -4e+23)
		tmp = Float64(1.0 / x);
	elseif (x <= 1000000.0)
		tmp = Float64(Float64(x / Float64((x ^ 4.0) + -1.0)) * fma(x, x, -1.0));
	else
		tmp = Float64(1.0 / x);
	end
	return tmp
end

Wolfram

code[x_] := If[LessEqual[x, -4e+23], N[(1.0 / x), $MachinePrecision], If[LessEqual[x, 1000000.0], N[(N[(x / N[(N[Power[x, 4.0], $MachinePrecision] + -1.0), $MachinePrecision]), $MachinePrecision] * N[(x * x + -1.0), $MachinePrecision]), $MachinePrecision], N[(1.0 / x), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if x < -3.9999999999999997e23 or 1e6 < x

   1. Initial program 49.7%

      \[\frac{x}{x \cdot x + 1} \]

   2. Taylor expanded in x around inf 100.0%

      \[\leadsto \color{blue}{\frac{1}{x}} \]

   if -3.9999999999999997e23 < x < 1e6

   1. Initial program 100.0%

      \[\frac{x}{x \cdot x + 1} \]
   2. Step-by-step derivation

      1. flip-+ 100.0%

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

      2. associate-/r/ 100.0%

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

      3. metadata-eval 100.0%

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

      4. sub-neg 100.0%

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

      5. pow2 100.0%

         \[\leadsto \frac{x}{\color{blue}{{x}^{2}} \cdot \left(x \cdot x\right) + \left(-1\right)} \cdot \left(x \cdot x - 1\right) \]

      6. pow2 100.0%

         \[\leadsto \frac{x}{{x}^{2} \cdot \color{blue}{{x}^{2}} + \left(-1\right)} \cdot \left(x \cdot x - 1\right) \]

      7. pow-prod-up 100.0%

         \[\leadsto \frac{x}{\color{blue}{{x}^{\left(2 + 2\right)}} + \left(-1\right)} \cdot \left(x \cdot x - 1\right) \]

      8. metadata-eval 100.0%

         \[\leadsto \frac{x}{{x}^{\color{blue}{4}} + \left(-1\right)} \cdot \left(x \cdot x - 1\right) \]

      9. metadata-eval 100.0%

         \[\leadsto \frac{x}{{x}^{4} + \color{blue}{-1}} \cdot \left(x \cdot x - 1\right) \]

      10. fma-neg 100.0%

          \[\leadsto \frac{x}{{x}^{4} + -1} \cdot \color{blue}{\mathsf{fma}\left(x, x, -1\right)} \]

      11. metadata-eval 100.0%

          \[\leadsto \frac{x}{{x}^{4} + -1} \cdot \mathsf{fma}\left(x, x, \color{blue}{-1}\right) \]

   3. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{\frac{x}{{x}^{4} + -1} \cdot \mathsf{fma}\left(x, x, -1\right)} \]
3. Recombined 2 regimes into one program.

4. Final simplification 100.0%

   \[\leadsto \begin{array}{l} \mathbf{if}\;x \leq -4 \cdot 10^{+23}:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 1000000:\\ \;\;\;\;\frac{x}{{x}^{4} + -1} \cdot \mathsf{fma}\left(x, x, -1\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \]

Alternative 2: 99.7% accurate, 0.1× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;x \leq -13200:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 12500:\\ \;\;\;\;x \cdot \left(-1 + \left(1 + \frac{1}{\mathsf{fma}\left(x, x, 1\right)}\right)\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (if (<= x -13200.0)
   (/ 1.0 x)
   (if (<= x 12500.0) (* x (+ -1.0 (+ 1.0 (/ 1.0 (fma x x 1.0))))) (/ 1.0 x))))

C

double code(double x) {
	double tmp;
	if (x <= -13200.0) {
		tmp = 1.0 / x;
	} else if (x <= 12500.0) {
		tmp = x * (-1.0 + (1.0 + (1.0 / fma(x, x, 1.0))));
	} else {
		tmp = 1.0 / x;
	}
	return tmp;
}

Julia

function code(x)
	tmp = 0.0
	if (x <= -13200.0)
		tmp = Float64(1.0 / x);
	elseif (x <= 12500.0)
		tmp = Float64(x * Float64(-1.0 + Float64(1.0 + Float64(1.0 / fma(x, x, 1.0)))));
	else
		tmp = Float64(1.0 / x);
	end
	return tmp
end

Wolfram

code[x_] := If[LessEqual[x, -13200.0], N[(1.0 / x), $MachinePrecision], If[LessEqual[x, 12500.0], N[(x * N[(-1.0 + N[(1.0 + N[(1.0 / N[(x * x + 1.0), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(1.0 / x), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if x < -13200 or 12500 < x

   1. Initial program 50.9%

      \[\frac{x}{x \cdot x + 1} \]

   2. Taylor expanded in x around inf 100.0%

      \[\leadsto \color{blue}{\frac{1}{x}} \]

   if -13200 < x < 12500

   1. Initial program 100.0%

      \[\frac{x}{x \cdot x + 1} \]
   2. Step-by-step derivation

      1. clear-num 99.7%

         \[\leadsto \color{blue}{\frac{1}{\frac{x \cdot x + 1}{x}}} \]

      2. associate-/r/ 100.0%

         \[\leadsto \color{blue}{\frac{1}{x \cdot x + 1} \cdot x} \]

      3. fma-def 100.0%

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

   3. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{\frac{1}{\mathsf{fma}\left(x, x, 1\right)} \cdot x} \]
   4. Step-by-step derivation

      1. add-exp-log 100.0%

         \[\leadsto \color{blue}{e^{\log \left(\frac{1}{\mathsf{fma}\left(x, x, 1\right)}\right)}} \cdot x \]

      2. log-rec 100.0%

         \[\leadsto e^{\color{blue}{-\log \left(\mathsf{fma}\left(x, x, 1\right)\right)}} \cdot x \]

      3. fma-udef 100.0%

         \[\leadsto e^{-\log \color{blue}{\left(x \cdot x + 1\right)}} \cdot x \]

      4. pow2 100.0%

         \[\leadsto e^{-\log \left(\color{blue}{{x}^{2}} + 1\right)} \cdot x \]

      5. metadata-eval 100.0%

         \[\leadsto e^{-\log \left({x}^{\color{blue}{\left(\frac{4}{2}\right)}} + 1\right)} \cdot x \]

      6. +-commutative 100.0%

         \[\leadsto e^{-\log \color{blue}{\left(1 + {x}^{\left(\frac{4}{2}\right)}\right)}} \cdot x \]

      7. log1p-udef 100.0%

         \[\leadsto e^{-\color{blue}{\mathsf{log1p}\left({x}^{\left(\frac{4}{2}\right)}\right)}} \cdot x \]

      8. metadata-eval 100.0%

         \[\leadsto e^{-\mathsf{log1p}\left({x}^{\color{blue}{2}}\right)} \cdot x \]

      9. pow2 100.0%

         \[\leadsto e^{-\mathsf{log1p}\left(\color{blue}{x \cdot x}\right)} \cdot x \]

   5. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{e^{-\mathsf{log1p}\left(x \cdot x\right)}} \cdot x \]
   6. Step-by-step derivation

      1. exp-neg 100.0%

         \[\leadsto \color{blue}{\frac{1}{e^{\mathsf{log1p}\left(x \cdot x\right)}}} \cdot x \]

      2. log1p-udef 100.0%

         \[\leadsto \frac{1}{e^{\color{blue}{\log \left(1 + x \cdot x\right)}}} \cdot x \]

      3. add-exp-log 100.0%

         \[\leadsto \frac{1}{\color{blue}{1 + x \cdot x}} \cdot x \]

      4. +-commutative 100.0%

         \[\leadsto \frac{1}{\color{blue}{x \cdot x + 1}} \cdot x \]

   7. Applied egg-rr 100.0%

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

      1. expm1-log1p-u 100.0%

         \[\leadsto \color{blue}{\mathsf{expm1}\left(\mathsf{log1p}\left(\frac{1}{x \cdot x + 1}\right)\right)} \cdot x \]

      2. expm1-udef 100.0%

         \[\leadsto \color{blue}{\left(e^{\mathsf{log1p}\left(\frac{1}{x \cdot x + 1}\right)} - 1\right)} \cdot x \]

      3. log1p-udef 100.0%

         \[\leadsto \left(e^{\color{blue}{\log \left(1 + \frac{1}{x \cdot x + 1}\right)}} - 1\right) \cdot x \]

      4. add-exp-log 100.0%

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

      5. fma-def 100.0%

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

   9. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{\left(\left(1 + \frac{1}{\mathsf{fma}\left(x, x, 1\right)}\right) - 1\right)} \cdot x \]
3. Recombined 2 regimes into one program.

4. Final simplification 100.0%

   \[\leadsto \begin{array}{l} \mathbf{if}\;x \leq -13200:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 12500:\\ \;\;\;\;x \cdot \left(-1 + \left(1 + \frac{1}{\mathsf{fma}\left(x, x, 1\right)}\right)\right)\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \]

Alternative 3: 100.0% accurate, 0.5× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;x \leq -500000000:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 1000000:\\ \;\;\;\;x \cdot \frac{1}{1 + x \cdot x}\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (if (<= x -500000000.0)
   (/ 1.0 x)
   (if (<= x 1000000.0) (* x (/ 1.0 (+ 1.0 (* x x)))) (/ 1.0 x))))

C

double code(double x) {
	double tmp;
	if (x <= -500000000.0) {
		tmp = 1.0 / x;
	} else if (x <= 1000000.0) {
		tmp = x * (1.0 / (1.0 + (x * x)));
	} else {
		tmp = 1.0 / x;
	}
	return tmp;
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: tmp
    if (x <= (-500000000.0d0)) then
        tmp = 1.0d0 / x
    else if (x <= 1000000.0d0) then
        tmp = x * (1.0d0 / (1.0d0 + (x * x)))
    else
        tmp = 1.0d0 / x
    end if
    code = tmp
end function

Java

public static double code(double x) {
	double tmp;
	if (x <= -500000000.0) {
		tmp = 1.0 / x;
	} else if (x <= 1000000.0) {
		tmp = x * (1.0 / (1.0 + (x * x)));
	} else {
		tmp = 1.0 / x;
	}
	return tmp;
}

Python

def code(x):
	tmp = 0
	if x <= -500000000.0:
		tmp = 1.0 / x
	elif x <= 1000000.0:
		tmp = x * (1.0 / (1.0 + (x * x)))
	else:
		tmp = 1.0 / x
	return tmp

Julia

function code(x)
	tmp = 0.0
	if (x <= -500000000.0)
		tmp = Float64(1.0 / x);
	elseif (x <= 1000000.0)
		tmp = Float64(x * Float64(1.0 / Float64(1.0 + Float64(x * x))));
	else
		tmp = Float64(1.0 / x);
	end
	return tmp
end

MATLAB

function tmp_2 = code(x)
	tmp = 0.0;
	if (x <= -500000000.0)
		tmp = 1.0 / x;
	elseif (x <= 1000000.0)
		tmp = x * (1.0 / (1.0 + (x * x)));
	else
		tmp = 1.0 / x;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_] := If[LessEqual[x, -500000000.0], N[(1.0 / x), $MachinePrecision], If[LessEqual[x, 1000000.0], N[(x * N[(1.0 / N[(1.0 + N[(x * x), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(1.0 / x), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if x < -5e8 or 1e6 < x

   1. Initial program 50.5%

      \[\frac{x}{x \cdot x + 1} \]

   2. Taylor expanded in x around inf 100.0%

      \[\leadsto \color{blue}{\frac{1}{x}} \]

   if -5e8 < x < 1e6

   1. Initial program 100.0%

      \[\frac{x}{x \cdot x + 1} \]
   2. Step-by-step derivation

      1. clear-num 99.7%

         \[\leadsto \color{blue}{\frac{1}{\frac{x \cdot x + 1}{x}}} \]

      2. associate-/r/ 100.0%

         \[\leadsto \color{blue}{\frac{1}{x \cdot x + 1} \cdot x} \]

      3. fma-def 100.0%

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

   3. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{\frac{1}{\mathsf{fma}\left(x, x, 1\right)} \cdot x} \]
   4. Step-by-step derivation

      1. add-exp-log 100.0%

         \[\leadsto \color{blue}{e^{\log \left(\frac{1}{\mathsf{fma}\left(x, x, 1\right)}\right)}} \cdot x \]

      2. log-rec 100.0%

         \[\leadsto e^{\color{blue}{-\log \left(\mathsf{fma}\left(x, x, 1\right)\right)}} \cdot x \]

      3. fma-udef 100.0%

         \[\leadsto e^{-\log \color{blue}{\left(x \cdot x + 1\right)}} \cdot x \]

      4. pow2 100.0%

         \[\leadsto e^{-\log \left(\color{blue}{{x}^{2}} + 1\right)} \cdot x \]

      5. metadata-eval 100.0%

         \[\leadsto e^{-\log \left({x}^{\color{blue}{\left(\frac{4}{2}\right)}} + 1\right)} \cdot x \]

      6. +-commutative 100.0%

         \[\leadsto e^{-\log \color{blue}{\left(1 + {x}^{\left(\frac{4}{2}\right)}\right)}} \cdot x \]

      7. log1p-udef 100.0%

         \[\leadsto e^{-\color{blue}{\mathsf{log1p}\left({x}^{\left(\frac{4}{2}\right)}\right)}} \cdot x \]

      8. metadata-eval 100.0%

         \[\leadsto e^{-\mathsf{log1p}\left({x}^{\color{blue}{2}}\right)} \cdot x \]

      9. pow2 100.0%

         \[\leadsto e^{-\mathsf{log1p}\left(\color{blue}{x \cdot x}\right)} \cdot x \]

   5. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{e^{-\mathsf{log1p}\left(x \cdot x\right)}} \cdot x \]
   6. Step-by-step derivation

      1. exp-neg 100.0%

         \[\leadsto \color{blue}{\frac{1}{e^{\mathsf{log1p}\left(x \cdot x\right)}}} \cdot x \]

      2. log1p-udef 100.0%

         \[\leadsto \frac{1}{e^{\color{blue}{\log \left(1 + x \cdot x\right)}}} \cdot x \]

      3. add-exp-log 100.0%

         \[\leadsto \frac{1}{\color{blue}{1 + x \cdot x}} \cdot x \]

      4. +-commutative 100.0%

         \[\leadsto \frac{1}{\color{blue}{x \cdot x + 1}} \cdot x \]

   7. Applied egg-rr 100.0%

      \[\leadsto \color{blue}{\frac{1}{x \cdot x + 1}} \cdot x \]
3. Recombined 2 regimes into one program.

4. Final simplification 100.0%

   \[\leadsto \begin{array}{l} \mathbf{if}\;x \leq -500000000:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 1000000:\\ \;\;\;\;x \cdot \frac{1}{1 + x \cdot x}\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \]

Alternative 4: 100.0% accurate, 0.6× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;x \leq -500000000:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 500000000:\\ \;\;\;\;\frac{x}{1 + x \cdot x}\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (if (<= x -500000000.0)
   (/ 1.0 x)
   (if (<= x 500000000.0) (/ x (+ 1.0 (* x x))) (/ 1.0 x))))

C

double code(double x) {
	double tmp;
	if (x <= -500000000.0) {
		tmp = 1.0 / x;
	} else if (x <= 500000000.0) {
		tmp = x / (1.0 + (x * x));
	} else {
		tmp = 1.0 / x;
	}
	return tmp;
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: tmp
    if (x <= (-500000000.0d0)) then
        tmp = 1.0d0 / x
    else if (x <= 500000000.0d0) then
        tmp = x / (1.0d0 + (x * x))
    else
        tmp = 1.0d0 / x
    end if
    code = tmp
end function

Java

public static double code(double x) {
	double tmp;
	if (x <= -500000000.0) {
		tmp = 1.0 / x;
	} else if (x <= 500000000.0) {
		tmp = x / (1.0 + (x * x));
	} else {
		tmp = 1.0 / x;
	}
	return tmp;
}

Python

def code(x):
	tmp = 0
	if x <= -500000000.0:
		tmp = 1.0 / x
	elif x <= 500000000.0:
		tmp = x / (1.0 + (x * x))
	else:
		tmp = 1.0 / x
	return tmp

Julia

function code(x)
	tmp = 0.0
	if (x <= -500000000.0)
		tmp = Float64(1.0 / x);
	elseif (x <= 500000000.0)
		tmp = Float64(x / Float64(1.0 + Float64(x * x)));
	else
		tmp = Float64(1.0 / x);
	end
	return tmp
end

MATLAB

function tmp_2 = code(x)
	tmp = 0.0;
	if (x <= -500000000.0)
		tmp = 1.0 / x;
	elseif (x <= 500000000.0)
		tmp = x / (1.0 + (x * x));
	else
		tmp = 1.0 / x;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_] := If[LessEqual[x, -500000000.0], N[(1.0 / x), $MachinePrecision], If[LessEqual[x, 500000000.0], N[(x / N[(1.0 + N[(x * x), $MachinePrecision]), $MachinePrecision]), $MachinePrecision], N[(1.0 / x), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if x < -5e8 or 5e8 < x

   1. Initial program 50.5%

      \[\frac{x}{x \cdot x + 1} \]

   2. Taylor expanded in x around inf 100.0%

      \[\leadsto \color{blue}{\frac{1}{x}} \]

   if -5e8 < x < 5e8

   1. Initial program 100.0%

      \[\frac{x}{x \cdot x + 1} \]
3. Recombined 2 regimes into one program.

4. Final simplification 100.0%

   \[\leadsto \begin{array}{l} \mathbf{if}\;x \leq -500000000:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 500000000:\\ \;\;\;\;\frac{x}{1 + x \cdot x}\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \]

Alternative 5: 98.7% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;x \leq -1:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 1:\\ \;\;\;\;x\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \end{array} \]

FPCore

(FPCore (x)
 :precision binary64
 (if (<= x -1.0) (/ 1.0 x) (if (<= x 1.0) x (/ 1.0 x))))

C

double code(double x) {
	double tmp;
	if (x <= -1.0) {
		tmp = 1.0 / x;
	} else if (x <= 1.0) {
		tmp = x;
	} else {
		tmp = 1.0 / x;
	}
	return tmp;
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    real(8) :: tmp
    if (x <= (-1.0d0)) then
        tmp = 1.0d0 / x
    else if (x <= 1.0d0) then
        tmp = x
    else
        tmp = 1.0d0 / x
    end if
    code = tmp
end function

Java

public static double code(double x) {
	double tmp;
	if (x <= -1.0) {
		tmp = 1.0 / x;
	} else if (x <= 1.0) {
		tmp = x;
	} else {
		tmp = 1.0 / x;
	}
	return tmp;
}

Python

def code(x):
	tmp = 0
	if x <= -1.0:
		tmp = 1.0 / x
	elif x <= 1.0:
		tmp = x
	else:
		tmp = 1.0 / x
	return tmp

Julia

function code(x)
	tmp = 0.0
	if (x <= -1.0)
		tmp = Float64(1.0 / x);
	elseif (x <= 1.0)
		tmp = x;
	else
		tmp = Float64(1.0 / x);
	end
	return tmp
end

MATLAB

function tmp_2 = code(x)
	tmp = 0.0;
	if (x <= -1.0)
		tmp = 1.0 / x;
	elseif (x <= 1.0)
		tmp = x;
	else
		tmp = 1.0 / x;
	end
	tmp_2 = tmp;
end

Wolfram

code[x_] := If[LessEqual[x, -1.0], N[(1.0 / x), $MachinePrecision], If[LessEqual[x, 1.0], x, N[(1.0 / x), $MachinePrecision]]]

Derivation

1. Split input into 2 regimes

2. if x < -1 or 1 < x

   1. Initial program 51.2%

      \[\frac{x}{x \cdot x + 1} \]

   2. Taylor expanded in x around inf 99.4%

      \[\leadsto \color{blue}{\frac{1}{x}} \]

   if -1 < x < 1

   1. Initial program 100.0%

      \[\frac{x}{x \cdot x + 1} \]

   2. Taylor expanded in x around 0 99.4%

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

4. Final simplification 99.4%

   \[\leadsto \begin{array}{l} \mathbf{if}\;x \leq -1:\\ \;\;\;\;\frac{1}{x}\\ \mathbf{elif}\;x \leq 1:\\ \;\;\;\;x\\ \mathbf{else}:\\ \;\;\;\;\frac{1}{x}\\ \end{array} \]

Alternative 6: 49.6% accurate, 7.0× speedup

Math

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

FPCore

(FPCore (x) :precision binary64 x)

C

double code(double x) {
	return x;
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    code = x
end function

Java

public static double code(double x) {
	return x;
}

Python

def code(x):
	return x

Julia

function code(x)
	return x
end

MATLAB

function tmp = code(x)
	tmp = x;
end

Wolfram

code[x_] := x

Derivation

1. Initial program 74.5%

   \[\frac{x}{x \cdot x + 1} \]

2. Taylor expanded in x around 0 49.3%

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

3. Final simplification 49.3%

   \[\leadsto x \]

Developer target: 99.8% accurate, 1.0× speedup

Math

\[\begin{array}{l} \\ \frac{1}{x + \frac{1}{x}} \end{array} \]

FPCore

(FPCore (x) :precision binary64 (/ 1.0 (+ x (/ 1.0 x))))

C

double code(double x) {
	return 1.0 / (x + (1.0 / x));
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    code = 1.0d0 / (x + (1.0d0 / x))
end function

Java

public static double code(double x) {
	return 1.0 / (x + (1.0 / x));
}

Python

def code(x):
	return 1.0 / (x + (1.0 / x))

Julia

function code(x)
	return Float64(1.0 / Float64(x + Float64(1.0 / x)))
end

MATLAB

function tmp = code(x)
	tmp = 1.0 / (x + (1.0 / x));
end

Wolfram

code[x_] := N[(1.0 / N[(x + N[(1.0 / x), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

Reproduce

herbie shell --seed 2023255 
(FPCore (x)
  :name "x / (x^2 + 1)"
  :precision binary64

  :herbie-target
  (/ 1.0 (+ x (/ 1.0 x)))

  (/ x (+ (* x x) 1.0)))