Ian Simplification

?

Percentage Accurate: 7.0% → 8.4%
Time: 27.1s
Precision: binary64
Cost: 98304

?

\[\frac{\pi}{2} - 2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right) \]
↓
\[\begin{array}{l} t_0 := \sqrt{0.5 - 0.5 \cdot x}\\ t_1 := \sin^{-1} t_0\\ \frac{1}{\frac{\mathsf{fma}\left(2 \cdot t_1, \mathsf{fma}\left(2, t_1, 0.5 \cdot \pi\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\left(0.5 \cdot \pi - \cos^{-1} t_0\right)}^{3}}} \end{array} \]
(FPCore (x)
 :precision binary64
 (- (/ PI 2.0) (* 2.0 (asin (sqrt (/ (- 1.0 x) 2.0))))))
↓
(FPCore (x)
 :precision binary64
 (let* ((t_0 (sqrt (- 0.5 (* 0.5 x)))) (t_1 (asin t_0)))
   (/
    1.0
    (/
     (fma (* 2.0 t_1) (fma 2.0 t_1 (* 0.5 PI)) (* (pow PI 2.0) 0.25))
     (- (* (pow PI 3.0) 0.125) (* 8.0 (pow (- (* 0.5 PI) (acos t_0)) 3.0)))))))
double code(double x) {
	return (((double) M_PI) / 2.0) - (2.0 * asin(sqrt(((1.0 - x) / 2.0))));
}
↓
double code(double x) {
	double t_0 = sqrt((0.5 - (0.5 * x)));
	double t_1 = asin(t_0);
	return 1.0 / (fma((2.0 * t_1), fma(2.0, t_1, (0.5 * ((double) M_PI))), (pow(((double) M_PI), 2.0) * 0.25)) / ((pow(((double) M_PI), 3.0) * 0.125) - (8.0 * pow(((0.5 * ((double) M_PI)) - acos(t_0)), 3.0))));
}

function code(x)
	return Float64(Float64(pi / 2.0) - Float64(2.0 * asin(sqrt(Float64(Float64(1.0 - x) / 2.0)))))
end
↓
function code(x)
	t_0 = sqrt(Float64(0.5 - Float64(0.5 * x)))
	t_1 = asin(t_0)
	return Float64(1.0 / Float64(fma(Float64(2.0 * t_1), fma(2.0, t_1, Float64(0.5 * pi)), Float64((pi ^ 2.0) * 0.25)) / Float64(Float64((pi ^ 3.0) * 0.125) - Float64(8.0 * (Float64(Float64(0.5 * pi) - acos(t_0)) ^ 3.0)))))
end

code[x_] := N[(N[(Pi / 2.0), $MachinePrecision] - N[(2.0 * N[ArcSin[N[Sqrt[N[(N[(1.0 - x), $MachinePrecision] / 2.0), $MachinePrecision]], $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]
↓
code[x_] := Block[{t$95$0 = N[Sqrt[N[(0.5 - N[(0.5 * x), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]}, Block[{t$95$1 = N[ArcSin[t$95$0], $MachinePrecision]}, N[(1.0 / N[(N[(N[(2.0 * t$95$1), $MachinePrecision] * N[(2.0 * t$95$1 + N[(0.5 * Pi), $MachinePrecision]), $MachinePrecision] + N[(N[Power[Pi, 2.0], $MachinePrecision] * 0.25), $MachinePrecision]), $MachinePrecision] / N[(N[(N[Power[Pi, 3.0], $MachinePrecision] * 0.125), $MachinePrecision] - N[(8.0 * N[Power[N[(N[(0.5 * Pi), $MachinePrecision] - N[ArcCos[t$95$0], $MachinePrecision]), $MachinePrecision], 3.0], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]]

\frac{\pi}{2} - 2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)
↓
\begin{array}{l}
t_0 := \sqrt{0.5 - 0.5 \cdot x}\\
t_1 := \sin^{-1} t_0\\
\frac{1}{\frac{\mathsf{fma}\left(2 \cdot t_1, \mathsf{fma}\left(2, t_1, 0.5 \cdot \pi\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\left(0.5 \cdot \pi - \cos^{-1} t_0\right)}^{3}}}
\end{array}

Local Percentage Accuracy vs ?

The average percentage accuracy by input value. Horizontal axis shows value of an input variable; the variable is choosen in the title. Vertical axis is accuracy; higher is better. Red represent the original program, while blue represents Herbie's suggestion. These can be toggled with buttons below the plot. The line is an average while dots represent individual samples.

Herbie found 8 alternatives:

AlternativeAccuracySpeedup

Accuracy vs Speed

The accuracy (vertical axis) and speed (horizontal axis) of each alternatives. Up and to the right is better. The red square shows the initial program, and each blue circle shows an alternative.The line shows the best available speed-accuracy tradeoffs.

Bogosity?

Bogosity

Target

Original7.0%
Target100.0%
Herbie8.4%
\[\sin^{-1} x \]

Derivation?

  1. Initial program 5.0%

    \[\frac{\pi}{2} - 2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right) \]

  2. Applied egg-rr5.0%

    \[\leadsto \color{blue}{\frac{1}{\frac{\mathsf{fma}\left(2 \cdot \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \mathsf{fma}\left(2, \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \pi \cdot 0.5\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right)}^{3}}}} \]
    Step-by-step derivation

    [Start]5.0%

    \[ \frac{\pi}{2} - 2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right) \]

    flip3-- [=>]5.0%

    \[ \color{blue}{\frac{{\left(\frac{\pi}{2}\right)}^{3} - {\left(2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)\right)}^{3}}{\frac{\pi}{2} \cdot \frac{\pi}{2} + \left(\left(2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)\right) \cdot \left(2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)\right) + \frac{\pi}{2} \cdot \left(2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)\right)\right)}} \]

    clear-num [=>]5.0%

    \[ \color{blue}{\frac{1}{\frac{\frac{\pi}{2} \cdot \frac{\pi}{2} + \left(\left(2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)\right) \cdot \left(2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)\right) + \frac{\pi}{2} \cdot \left(2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)\right)\right)}{{\left(\frac{\pi}{2}\right)}^{3} - {\left(2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right)\right)}^{3}}}} \]

  3. Applied egg-rr6.9%

    \[\leadsto \frac{1}{\frac{\mathsf{fma}\left(2 \cdot \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \mathsf{fma}\left(2, \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \pi \cdot 0.5\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\color{blue}{\left(\pi \cdot 0.5 - \cos^{-1} \left(\sqrt{0.5 - 0.5 \cdot x}\right)\right)}}^{3}}} \]
    Step-by-step derivation

    [Start]5.0%

    \[ \frac{1}{\frac{\mathsf{fma}\left(2 \cdot \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \mathsf{fma}\left(2, \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \pi \cdot 0.5\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right)}^{3}}} \]

    asin-acos [=>]6.9%

    \[ \frac{1}{\frac{\mathsf{fma}\left(2 \cdot \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \mathsf{fma}\left(2, \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \pi \cdot 0.5\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\color{blue}{\left(\frac{\pi}{2} - \cos^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right)\right)}}^{3}}} \]

    div-inv [=>]6.9%

    \[ \frac{1}{\frac{\mathsf{fma}\left(2 \cdot \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \mathsf{fma}\left(2, \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \pi \cdot 0.5\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\left(\color{blue}{\pi \cdot \frac{1}{2}} - \cos^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right)\right)}^{3}}} \]

    metadata-eval [=>]6.9%

    \[ \frac{1}{\frac{\mathsf{fma}\left(2 \cdot \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \mathsf{fma}\left(2, \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \pi \cdot 0.5\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\left(\pi \cdot \color{blue}{0.5} - \cos^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right)\right)}^{3}}} \]

    *-commutative [=>]6.9%

    \[ \frac{1}{\frac{\mathsf{fma}\left(2 \cdot \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \mathsf{fma}\left(2, \sin^{-1} \left(\sqrt{0.5 - x \cdot 0.5}\right), \pi \cdot 0.5\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\left(\pi \cdot 0.5 - \cos^{-1} \left(\sqrt{0.5 - \color{blue}{0.5 \cdot x}}\right)\right)}^{3}}} \]

  4. Final simplification6.9%

    \[\leadsto \frac{1}{\frac{\mathsf{fma}\left(2 \cdot \sin^{-1} \left(\sqrt{0.5 - 0.5 \cdot x}\right), \mathsf{fma}\left(2, \sin^{-1} \left(\sqrt{0.5 - 0.5 \cdot x}\right), 0.5 \cdot \pi\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\left(0.5 \cdot \pi - \cos^{-1} \left(\sqrt{0.5 - 0.5 \cdot x}\right)\right)}^{3}}} \]

Alternatives

Alternative 1
Accuracy8.4%
Cost98304
\[\begin{array}{l} t_0 := \sqrt{0.5 - 0.5 \cdot x}\\ t_1 := \sin^{-1} t_0\\ \frac{1}{\frac{\mathsf{fma}\left(2 \cdot t_1, \mathsf{fma}\left(2, t_1, 0.5 \cdot \pi\right), {\pi}^{2} \cdot 0.25\right)}{{\pi}^{3} \cdot 0.125 - 8 \cdot {\left(0.5 \cdot \pi - \cos^{-1} t_0\right)}^{3}}} \end{array} \]
Alternative 2
Accuracy8.4%
Cost45568
\[{\left(\sqrt[3]{\mathsf{fma}\left(0.5 \cdot \pi - \cos^{-1} \left(\sqrt{0.5 - 0.5 \cdot x}\right), -2, 0.5 \cdot \pi\right)}\right)}^{3} \]
Alternative 3
Accuracy8.4%
Cost26432
\[\frac{\pi}{2} + 2 \cdot \left(\cos^{-1} \left(\sqrt{0.5 - 0.5 \cdot x}\right) - 0.5 \cdot \pi\right) \]
Alternative 4
Accuracy5.4%
Cost26176
\[\frac{\pi}{2} + 2 \cdot \left(\cos^{-1} \left(\sqrt{0.5}\right) - 0.5 \cdot \pi\right) \]
Alternative 5
Accuracy7.0%
Cost19968
\[\frac{\pi}{2} - 2 \cdot \sin^{-1} \left(\frac{1}{\sqrt{\frac{2}{1 - x}}}\right) \]
Alternative 6
Accuracy7.0%
Cost19840
\[\frac{\pi}{2} - 2 \cdot \sin^{-1} \left(\sqrt{\frac{1 - x}{2}}\right) \]
Alternative 7
Accuracy4.1%
Cost19712
\[\frac{\pi}{2} - 2 \cdot \sin^{-1} \left(\frac{1}{\sqrt{2}}\right) \]
Alternative 8
Accuracy4.1%
Cost19584
\[\frac{\pi}{2} - 2 \cdot \sin^{-1} \left(\sqrt{0.5}\right) \]

Reproduce?

herbie shell --seed 2023165 
(FPCore (x)
  :name "Ian Simplification"
  :precision binary64

  :herbie-target
  (asin x)

  (- (/ PI 2.0) (* 2.0 (asin (sqrt (/ (- 1.0 x) 2.0))))))