Hyperbolic secant

Average Accuracy: 100.0% → 100.0%

Time: 3.0s

Precision: binary64

Cost: 19392

Math

\[\frac{2}{e^{x} + e^{-x}} \]

↓

\[{\left({\cosh x}^{-0.5}\right)}^{2} \]

FPCore

(FPCore (x) :precision binary64 (/ 2.0 (+ (exp x) (exp (- x)))))

↓

(FPCore (x) :precision binary64 (pow (pow (cosh x) -0.5) 2.0))

C

double code(double x) {
	return 2.0 / (exp(x) + exp(-x));
}

↓

double code(double x) {
	return pow(pow(cosh(x), -0.5), 2.0);
}

Fortran

real(8) function code(x)
    real(8), intent (in) :: x
    code = 2.0d0 / (exp(x) + exp(-x))
end function

↓

real(8) function code(x)
    real(8), intent (in) :: x
    code = (cosh(x) ** (-0.5d0)) ** 2.0d0
end function

Java

public static double code(double x) {
	return 2.0 / (Math.exp(x) + Math.exp(-x));
}

↓

public static double code(double x) {
	return Math.pow(Math.pow(Math.cosh(x), -0.5), 2.0);
}

Python

def code(x):
	return 2.0 / (math.exp(x) + math.exp(-x))

↓

def code(x):
	return math.pow(math.pow(math.cosh(x), -0.5), 2.0)

Julia

function code(x)
	return Float64(2.0 / Float64(exp(x) + exp(Float64(-x))))
end

↓

function code(x)
	return (cosh(x) ^ -0.5) ^ 2.0
end

MATLAB

function tmp = code(x)
	tmp = 2.0 / (exp(x) + exp(-x));
end

↓

function tmp = code(x)
	tmp = (cosh(x) ^ -0.5) ^ 2.0;
end

Wolfram

code[x_] := N[(2.0 / N[(N[Exp[x], $MachinePrecision] + N[Exp[(-x)], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]

↓

code[x_] := N[Power[N[Power[N[Cosh[x], $MachinePrecision], -0.5], $MachinePrecision], 2.0], $MachinePrecision]

Derivation

1. Initial program 100.0%

   \[\frac{2}{e^{x} + e^{-x}} \]

2. Taylor expanded in x around inf 100.0%

   \[\leadsto \color{blue}{\frac{2}{e^{-x} + e^{x}}} \]

3. Simplified 100.0%

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

   [Start] 100.0	\[ \frac{2}{e^{-x} + e^{x}} \]
   metadata-eval [<=] 100.0	\[ \frac{\color{blue}{1 \cdot 2}}{e^{-x} + e^{x}} \]
   associate-*l/ [<=] 100.0	\[ \color{blue}{\frac{1}{e^{-x} + e^{x}} \cdot 2} \]
   associate-/r/ [<=] 100.0	\[ \color{blue}{\frac{1}{\frac{e^{-x} + e^{x}}{2}}} \]
   +-commutative [=>] 100.0	\[ \frac{1}{\frac{\color{blue}{e^{x} + e^{-x}}}{2}} \]
   cosh-def [<=] 100.0	\[ \frac{1}{\color{blue}{\cosh x}} \]

4. Applied egg-rr 100.0%

   \[\leadsto \color{blue}{{\left({\cosh x}^{-0.5}\right)}^{2}} \]

5. Final simplification 100.0%

   \[\leadsto {\left({\cosh x}^{-0.5}\right)}^{2} \]

Alternatives

Alternative 1
Accuracy	100.0%
Cost	6592

\[\frac{1}{\cosh x} \]

Alternative 2
Accuracy	98.5%
Cost	841

\[\begin{array}{l} \mathbf{if}\;x \leq -17000 \lor \neg \left(x \leq 720000\right):\\ \;\;\;\;\left(1 + \frac{2}{x \cdot x}\right) + -1\\ \mathbf{else}:\\ \;\;\;\;\frac{2}{2 + x \cdot x}\\ \end{array} \]

Alternative 3
Accuracy	76.4%
Cost	713

\[\begin{array}{l} \mathbf{if}\;x \leq -1.25 \lor \neg \left(x \leq 1.2\right):\\ \;\;\;\;\frac{2}{x \cdot x}\\ \mathbf{else}:\\ \;\;\;\;1 + -0.5 \cdot \left(x \cdot x\right)\\ \end{array} \]

Alternative 4
Accuracy	76.1%
Cost	585

\[\begin{array}{l} \mathbf{if}\;x \leq -1.42 \lor \neg \left(x \leq 1.4\right):\\ \;\;\;\;\frac{2}{x \cdot x}\\ \mathbf{else}:\\ \;\;\;\;1\\ \end{array} \]

Alternative 5
Accuracy	76.4%
Cost	448

\[\frac{2}{2 + x \cdot x} \]

Alternative 6
Accuracy	51.2%
Cost	64

\[1 \]

Reproduce

herbie shell --seed 2023129 
(FPCore (x)
  :name "Hyperbolic secant"
  :precision binary64
  (/ 2.0 (+ (exp x) (exp (- x)))))