UniformSampleCone, y

Percentage Accurate: 57.0% → 98.3%
Time: 7.3s
Alternatives: 18
Speedup: 4.6×

Specification

?
\[\left(\left(2.328306437 \cdot 10^{-10} \leq ux \land ux \leq 1\right) \land \left(2.328306437 \cdot 10^{-10} \leq uy \land uy \leq 1\right)\right) \land \left(0 \leq maxCos \land maxCos \leq 1\right)\]
\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(1 - ux\right) + ux \cdot maxCos\\ \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - t\_0 \cdot t\_0} \end{array} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (let* ((t_0 (+ (- 1.0 ux) (* ux maxCos))))
   (* (sin (* (* uy 2.0) PI)) (sqrt (- 1.0 (* t_0 t_0))))))
float code(float ux, float uy, float maxCos) {
	float t_0 = (1.0f - ux) + (ux * maxCos);
	return sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf((1.0f - (t_0 * t_0)));
}
function code(ux, uy, maxCos)
	t_0 = Float32(Float32(Float32(1.0) - ux) + Float32(ux * maxCos))
	return Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(1.0) - Float32(t_0 * t_0))))
end
function tmp = code(ux, uy, maxCos)
	t_0 = (single(1.0) - ux) + (ux * maxCos);
	tmp = sin(((uy * single(2.0)) * single(pi))) * sqrt((single(1.0) - (t_0 * t_0)));
end
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \left(1 - ux\right) + ux \cdot maxCos\\
\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - t\_0 \cdot t\_0}
\end{array}
\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.

Accuracy vs Speed?

Herbie found 18 alternatives:

AlternativeAccuracySpeedup
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.

Initial Program: 57.0% accurate, 1.0× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(1 - ux\right) + ux \cdot maxCos\\ \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - t\_0 \cdot t\_0} \end{array} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (let* ((t_0 (+ (- 1.0 ux) (* ux maxCos))))
   (* (sin (* (* uy 2.0) PI)) (sqrt (- 1.0 (* t_0 t_0))))))
float code(float ux, float uy, float maxCos) {
	float t_0 = (1.0f - ux) + (ux * maxCos);
	return sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf((1.0f - (t_0 * t_0)));
}
function code(ux, uy, maxCos)
	t_0 = Float32(Float32(Float32(1.0) - ux) + Float32(ux * maxCos))
	return Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(1.0) - Float32(t_0 * t_0))))
end
function tmp = code(ux, uy, maxCos)
	t_0 = (single(1.0) - ux) + (ux * maxCos);
	tmp = sin(((uy * single(2.0)) * single(pi))) * sqrt((single(1.0) - (t_0 * t_0)));
end
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \left(1 - ux\right) + ux \cdot maxCos\\
\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - t\_0 \cdot t\_0}
\end{array}
\end{array}

Alternative 1: 98.3% accurate, 0.8× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)} \end{array} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (let* ((t_0 (+ (- maxCos) 1.0)))
   (*
    (sin (* (* uy 2.0) PI))
    (sqrt
     (*
      (- (/ (* maxCos (- (* 2.0 (/ 1.0 maxCos)) 2.0)) ux) (* t_0 t_0))
      (* ux ux))))))
float code(float ux, float uy, float maxCos) {
	float t_0 = -maxCos + 1.0f;
	return sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf(((((maxCos * ((2.0f * (1.0f / maxCos)) - 2.0f)) / ux) - (t_0 * t_0)) * (ux * ux)));
}
function code(ux, uy, maxCos)
	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
	return Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(Float32(Float32(maxCos * Float32(Float32(Float32(2.0) * Float32(Float32(1.0) / maxCos)) - Float32(2.0))) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))))
end
function tmp = code(ux, uy, maxCos)
	t_0 = -maxCos + single(1.0);
	tmp = sin(((uy * single(2.0)) * single(pi))) * sqrt(((((maxCos * ((single(2.0) * (single(1.0) / maxCos)) - single(2.0))) / ux) - (t_0 * t_0)) * (ux * ux)));
end
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \left(-maxCos\right) + 1\\
\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}
\end{array}
\end{array}
Derivation
  1. Initial program 57.0%

    \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
  2. Taylor expanded in ux around -inf

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
  3. Step-by-step derivation
    1. *-commutativeN/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    2. lower-*.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
  4. Applied rewrites98.3%

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
  5. Taylor expanded in maxCos around inf

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
  6. Step-by-step derivation
    1. lower-*.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    2. lower--.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    3. lower-*.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    4. lower-/.f3297.8

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
  7. Applied rewrites97.8%

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
  8. Add Preprocessing

Alternative 2: 98.3% accurate, 0.9× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)} \end{array} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (let* ((t_0 (+ (- maxCos) 1.0)))
   (*
    (sin (* (* uy 2.0) PI))
    (sqrt (* (- (/ (fma -2.0 maxCos 2.0) ux) (* t_0 t_0)) (* ux ux))))))
float code(float ux, float uy, float maxCos) {
	float t_0 = -maxCos + 1.0f;
	return sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf((((fmaf(-2.0f, maxCos, 2.0f) / ux) - (t_0 * t_0)) * (ux * ux)));
}
function code(ux, uy, maxCos)
	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
	return Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))))
end
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \left(-maxCos\right) + 1\\
\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}
\end{array}
\end{array}
Derivation
  1. Initial program 57.0%

    \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
  2. Taylor expanded in ux around -inf

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
  3. Step-by-step derivation
    1. *-commutativeN/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    2. lower-*.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
  4. Applied rewrites98.3%

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
  5. Add Preprocessing

Alternative 3: 97.8% accurate, 1.0× speedup?

\[\begin{array}{l} \\ \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - \left(maxCos + maxCos\right)\right) \cdot ux} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (*
  (sin (* (* uy 2.0) PI))
  (sqrt
   (*
    (- (fma (- ux) (* (- maxCos 1.0) (- maxCos 1.0)) 2.0) (+ maxCos maxCos))
    ux))))
float code(float ux, float uy, float maxCos) {
	return sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf(((fmaf(-ux, ((maxCos - 1.0f) * (maxCos - 1.0f)), 2.0f) - (maxCos + maxCos)) * ux));
}
function code(ux, uy, maxCos)
	return Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(fma(Float32(-ux), Float32(Float32(maxCos - Float32(1.0)) * Float32(maxCos - Float32(1.0))), Float32(2.0)) - Float32(maxCos + maxCos)) * ux)))
end
\begin{array}{l}

\\
\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - \left(maxCos + maxCos\right)\right) \cdot ux}
\end{array}
Derivation
  1. Initial program 57.0%

    \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
  2. Taylor expanded in ux around 0

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{ux \cdot \left(\left(2 + -1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right)}} \]
  3. Step-by-step derivation
    1. *-commutativeN/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(2 + -1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right) \cdot \color{blue}{ux}} \]
    2. lower-*.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(2 + -1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right) \cdot \color{blue}{ux}} \]
    3. lower--.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(2 + -1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right) \cdot ux} \]
    4. +-commutativeN/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right) + 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    5. associate-*r*N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(\left(-1 \cdot ux\right) \cdot {\left(maxCos - 1\right)}^{2} + 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    6. mul-1-negN/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(\left(\mathsf{neg}\left(ux\right)\right) \cdot {\left(maxCos - 1\right)}^{2} + 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    7. lower-fma.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(\mathsf{neg}\left(ux\right), {\left(maxCos - 1\right)}^{2}, 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    8. lower-neg.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, {\left(maxCos - 1\right)}^{2}, 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    9. unpow2N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    10. lower-*.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    11. lower--.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    12. lower--.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - 2 \cdot maxCos\right) \cdot ux} \]
    13. count-2-revN/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - \left(maxCos + maxCos\right)\right) \cdot ux} \]
    14. lower-+.f3298.3

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - \left(maxCos + maxCos\right)\right) \cdot ux} \]
  4. Applied rewrites98.3%

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\mathsf{fma}\left(-ux, \left(maxCos - 1\right) \cdot \left(maxCos - 1\right), 2\right) - \left(maxCos + maxCos\right)\right) \cdot ux}} \]
  5. Add Preprocessing

Alternative 4: 97.1% accurate, 1.1× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \mathbf{if}\;uy \leq 0.008999999612569809:\\ \;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\ \mathbf{else}:\\ \;\;\;\;\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{\left(-ux\right) + 2}{ux} \cdot \left(ux \cdot ux\right)}\\ \end{array} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (let* ((t_0 (+ (- maxCos) 1.0)))
   (if (<= uy 0.008999999612569809)
     (*
      (* uy (fma -1.3333333333333333 (* (* uy uy) (* (* PI PI) PI)) (+ PI PI)))
      (sqrt (* (- (/ (fma -2.0 maxCos 2.0) ux) (* t_0 t_0)) (* ux ux))))
     (* (sin (* (* uy 2.0) PI)) (sqrt (* (/ (+ (- ux) 2.0) ux) (* ux ux)))))))
float code(float ux, float uy, float maxCos) {
	float t_0 = -maxCos + 1.0f;
	float tmp;
	if (uy <= 0.008999999612569809f) {
		tmp = (uy * fmaf(-1.3333333333333333f, ((uy * uy) * ((((float) M_PI) * ((float) M_PI)) * ((float) M_PI))), (((float) M_PI) + ((float) M_PI)))) * sqrtf((((fmaf(-2.0f, maxCos, 2.0f) / ux) - (t_0 * t_0)) * (ux * ux)));
	} else {
		tmp = sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf((((-ux + 2.0f) / ux) * (ux * ux)));
	}
	return tmp;
}
function code(ux, uy, maxCos)
	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
	tmp = Float32(0.0)
	if (uy <= Float32(0.008999999612569809))
		tmp = Float32(Float32(uy * fma(Float32(-1.3333333333333333), Float32(Float32(uy * uy) * Float32(Float32(Float32(pi) * Float32(pi)) * Float32(pi))), Float32(Float32(pi) + Float32(pi)))) * sqrt(Float32(Float32(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))));
	else
		tmp = Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(Float32(Float32(-ux) + Float32(2.0)) / ux) * Float32(ux * ux))));
	end
	return tmp
end
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \left(-maxCos\right) + 1\\
\mathbf{if}\;uy \leq 0.008999999612569809:\\
\;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\

\mathbf{else}:\\
\;\;\;\;\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{\left(-ux\right) + 2}{ux} \cdot \left(ux \cdot ux\right)}\\


\end{array}
\end{array}
Derivation
  1. Split input into 2 regimes
  2. if uy < 0.00899999961

    1. Initial program 57.2%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in ux around -inf

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
    3. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    4. Applied rewrites98.4%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
    5. Taylor expanded in uy around 0

      \[\leadsto \color{blue}{\left(uy \cdot \left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    6. Step-by-step derivation
      1. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \color{blue}{\left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      2. lower-fma.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \color{blue}{{uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      3. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, {uy}^{2} \cdot \color{blue}{{\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      4. unpow2N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      5. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      6. unpow3N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      7. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      8. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      9. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      10. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      11. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      12. count-2-revN/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \mathsf{PI}\left(\right) + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      13. lower-+.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \mathsf{PI}\left(\right) + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      14. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      15. lift-PI.f3298.4

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    7. Applied rewrites98.4%

      \[\leadsto \color{blue}{\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]

    if 0.00899999961 < uy

    1. Initial program 56.3%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in ux around -inf

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
    3. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    4. Applied rewrites97.7%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
    5. Taylor expanded in maxCos around 0

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
    6. Step-by-step derivation
      1. lower--.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      3. lift-/.f3292.2

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
    7. Applied rewrites92.2%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
    8. Taylor expanded in ux around 0

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{2 + -1 \cdot ux}{ux} \cdot \left(ux \cdot ux\right)} \]
    9. Step-by-step derivation
      1. lower-/.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{2 + -1 \cdot ux}{ux} \cdot \left(ux \cdot ux\right)} \]
      2. mul-1-negN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{2 + \left(\mathsf{neg}\left(ux\right)\right)}{ux} \cdot \left(ux \cdot ux\right)} \]
      3. +-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{\left(\mathsf{neg}\left(ux\right)\right) + 2}{ux} \cdot \left(ux \cdot ux\right)} \]
      4. lower-+.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{\left(\mathsf{neg}\left(ux\right)\right) + 2}{ux} \cdot \left(ux \cdot ux\right)} \]
      5. lower-neg.f3292.2

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{\left(-ux\right) + 2}{ux} \cdot \left(ux \cdot ux\right)} \]
    10. Applied rewrites92.2%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\frac{\left(-ux\right) + 2}{ux} \cdot \left(ux \cdot ux\right)} \]
  3. Recombined 2 regimes into one program.
  4. Add Preprocessing

Alternative 5: 97.0% accurate, 1.1× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \mathbf{if}\;uy \leq 0.008999999612569809:\\ \;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\ \mathbf{else}:\\ \;\;\;\;\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)}\\ \end{array} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (let* ((t_0 (+ (- maxCos) 1.0)))
   (if (<= uy 0.008999999612569809)
     (*
      (* uy (fma -1.3333333333333333 (* (* uy uy) (* (* PI PI) PI)) (+ PI PI)))
      (sqrt (* (- (/ (fma -2.0 maxCos 2.0) ux) (* t_0 t_0)) (* ux ux))))
     (* (sin (* (* uy 2.0) PI)) (sqrt (* (* ux ux) (- (/ 2.0 ux) 1.0)))))))
float code(float ux, float uy, float maxCos) {
	float t_0 = -maxCos + 1.0f;
	float tmp;
	if (uy <= 0.008999999612569809f) {
		tmp = (uy * fmaf(-1.3333333333333333f, ((uy * uy) * ((((float) M_PI) * ((float) M_PI)) * ((float) M_PI))), (((float) M_PI) + ((float) M_PI)))) * sqrtf((((fmaf(-2.0f, maxCos, 2.0f) / ux) - (t_0 * t_0)) * (ux * ux)));
	} else {
		tmp = sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf(((ux * ux) * ((2.0f / ux) - 1.0f)));
	}
	return tmp;
}
function code(ux, uy, maxCos)
	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
	tmp = Float32(0.0)
	if (uy <= Float32(0.008999999612569809))
		tmp = Float32(Float32(uy * fma(Float32(-1.3333333333333333), Float32(Float32(uy * uy) * Float32(Float32(Float32(pi) * Float32(pi)) * Float32(pi))), Float32(Float32(pi) + Float32(pi)))) * sqrt(Float32(Float32(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))));
	else
		tmp = Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(ux * ux) * Float32(Float32(Float32(2.0) / ux) - Float32(1.0)))));
	end
	return tmp
end
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \left(-maxCos\right) + 1\\
\mathbf{if}\;uy \leq 0.008999999612569809:\\
\;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\

\mathbf{else}:\\
\;\;\;\;\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)}\\


\end{array}
\end{array}
Derivation
  1. Split input into 2 regimes
  2. if uy < 0.00899999961

    1. Initial program 57.2%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in ux around -inf

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
    3. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    4. Applied rewrites98.4%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
    5. Taylor expanded in uy around 0

      \[\leadsto \color{blue}{\left(uy \cdot \left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    6. Step-by-step derivation
      1. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \color{blue}{\left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      2. lower-fma.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \color{blue}{{uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      3. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, {uy}^{2} \cdot \color{blue}{{\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      4. unpow2N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      5. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      6. unpow3N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      7. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      8. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      9. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      10. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      11. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      12. count-2-revN/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \mathsf{PI}\left(\right) + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      13. lower-+.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \mathsf{PI}\left(\right) + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      14. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      15. lift-PI.f3298.4

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    7. Applied rewrites98.4%

      \[\leadsto \color{blue}{\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]

    if 0.00899999961 < uy

    1. Initial program 56.3%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in ux around -inf

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
    3. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    4. Applied rewrites97.7%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
    5. Taylor expanded in maxCos around 0

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
    6. Step-by-step derivation
      1. lower--.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      3. lift-/.f3292.2

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
    7. Applied rewrites92.2%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
    8. Step-by-step derivation
      1. lift-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \color{blue}{\left(ux \cdot ux\right)}} \]
      2. lift-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot \color{blue}{ux}\right)} \]
      3. pow2N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot {ux}^{\color{blue}{2}}} \]
      4. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{{ux}^{2} \cdot \color{blue}{\left(2 \cdot \frac{1}{ux} - 1\right)}} \]
      5. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{{ux}^{2} \cdot \color{blue}{\left(2 \cdot \frac{1}{ux} - 1\right)}} \]
      6. pow2N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\color{blue}{2 \cdot \frac{1}{ux}} - 1\right)} \]
      7. lift-*.f3292.2

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\color{blue}{2 \cdot \frac{1}{ux}} - 1\right)} \]
      8. lift-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(2 \cdot \frac{1}{ux} - 1\right)} \]
      9. lift-/.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(2 \cdot \frac{1}{ux} - 1\right)} \]
      10. associate-*r/N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2 \cdot 1}{ux} - 1\right)} \]
      11. metadata-evalN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
      12. lower-/.f3292.2

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
    9. Applied rewrites92.2%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \color{blue}{\left(\frac{2}{ux} - 1\right)}} \]
  3. Recombined 2 regimes into one program.
  4. Add Preprocessing

Alternative 6: 97.0% accurate, 1.1× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \mathbf{if}\;uy \leq 0.008999999612569809:\\ \;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\ \mathbf{else}:\\ \;\;\;\;\sqrt{\left(\left(\frac{2}{ux} - 1\right) \cdot ux\right) \cdot ux} \cdot \sin \left(\left(uy + uy\right) \cdot \pi\right)\\ \end{array} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (let* ((t_0 (+ (- maxCos) 1.0)))
   (if (<= uy 0.008999999612569809)
     (*
      (* uy (fma -1.3333333333333333 (* (* uy uy) (* (* PI PI) PI)) (+ PI PI)))
      (sqrt (* (- (/ (fma -2.0 maxCos 2.0) ux) (* t_0 t_0)) (* ux ux))))
     (* (sqrt (* (* (- (/ 2.0 ux) 1.0) ux) ux)) (sin (* (+ uy uy) PI))))))
float code(float ux, float uy, float maxCos) {
	float t_0 = -maxCos + 1.0f;
	float tmp;
	if (uy <= 0.008999999612569809f) {
		tmp = (uy * fmaf(-1.3333333333333333f, ((uy * uy) * ((((float) M_PI) * ((float) M_PI)) * ((float) M_PI))), (((float) M_PI) + ((float) M_PI)))) * sqrtf((((fmaf(-2.0f, maxCos, 2.0f) / ux) - (t_0 * t_0)) * (ux * ux)));
	} else {
		tmp = sqrtf(((((2.0f / ux) - 1.0f) * ux) * ux)) * sinf(((uy + uy) * ((float) M_PI)));
	}
	return tmp;
}
function code(ux, uy, maxCos)
	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
	tmp = Float32(0.0)
	if (uy <= Float32(0.008999999612569809))
		tmp = Float32(Float32(uy * fma(Float32(-1.3333333333333333), Float32(Float32(uy * uy) * Float32(Float32(Float32(pi) * Float32(pi)) * Float32(pi))), Float32(Float32(pi) + Float32(pi)))) * sqrt(Float32(Float32(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))));
	else
		tmp = Float32(sqrt(Float32(Float32(Float32(Float32(Float32(2.0) / ux) - Float32(1.0)) * ux) * ux)) * sin(Float32(Float32(uy + uy) * Float32(pi))));
	end
	return tmp
end
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \left(-maxCos\right) + 1\\
\mathbf{if}\;uy \leq 0.008999999612569809:\\
\;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\

\mathbf{else}:\\
\;\;\;\;\sqrt{\left(\left(\frac{2}{ux} - 1\right) \cdot ux\right) \cdot ux} \cdot \sin \left(\left(uy + uy\right) \cdot \pi\right)\\


\end{array}
\end{array}
Derivation
  1. Split input into 2 regimes
  2. if uy < 0.00899999961

    1. Initial program 57.2%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in ux around -inf

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
    3. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    4. Applied rewrites98.4%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
    5. Taylor expanded in uy around 0

      \[\leadsto \color{blue}{\left(uy \cdot \left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    6. Step-by-step derivation
      1. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \color{blue}{\left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      2. lower-fma.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \color{blue}{{uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      3. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, {uy}^{2} \cdot \color{blue}{{\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      4. unpow2N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      5. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      6. unpow3N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      7. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      8. lower-*.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      9. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      10. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      11. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      12. count-2-revN/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \mathsf{PI}\left(\right) + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      13. lower-+.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \mathsf{PI}\left(\right) + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      14. lift-PI.f32N/A

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      15. lift-PI.f3298.4

        \[\leadsto \left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    7. Applied rewrites98.4%

      \[\leadsto \color{blue}{\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]

    if 0.00899999961 < uy

    1. Initial program 56.3%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in ux around -inf

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
    3. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    4. Applied rewrites97.7%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
    5. Taylor expanded in maxCos around 0

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
    6. Step-by-step derivation
      1. lower--.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      3. lift-/.f3292.2

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
    7. Applied rewrites92.2%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
    8. Step-by-step derivation
      1. lift-*.f32N/A

        \[\leadsto \color{blue}{\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)}} \]
      2. lift-sin.f32N/A

        \[\leadsto \color{blue}{\sin \left(\left(uy \cdot 2\right) \cdot \pi\right)} \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      3. lift-PI.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      4. lift-*.f32N/A

        \[\leadsto \sin \color{blue}{\left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right)} \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      5. lift-*.f32N/A

        \[\leadsto \sin \left(\color{blue}{\left(uy \cdot 2\right)} \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      6. *-commutativeN/A

        \[\leadsto \color{blue}{\sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \cdot \sin \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right)} \]
    9. Applied rewrites92.2%

      \[\leadsto \color{blue}{\sqrt{\left(\left(\frac{2}{ux} - 1\right) \cdot ux\right) \cdot ux} \cdot \sin \left(\left(uy + uy\right) \cdot \pi\right)} \]
  3. Recombined 2 regimes into one program.
  4. Add Preprocessing

Alternative 7: 96.8% accurate, 1.1× speedup?

\[\begin{array}{l} \\ \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \end{array} \]
(FPCore (ux uy maxCos)
 :precision binary32
 (*
  (sin (* (* uy 2.0) PI))
  (sqrt (* (- (/ (fma -2.0 maxCos 2.0) ux) 1.0) (* ux ux)))))
float code(float ux, float uy, float maxCos) {
	return sinf(((uy * 2.0f) * ((float) M_PI))) * sqrtf((((fmaf(-2.0f, maxCos, 2.0f) / ux) - 1.0f) * (ux * ux)));
}
function code(ux, uy, maxCos)
	return Float32(sin(Float32(Float32(uy * Float32(2.0)) * Float32(pi))) * sqrt(Float32(Float32(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) / ux) - Float32(1.0)) * Float32(ux * ux))))
end
\begin{array}{l}

\\
\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - 1\right) \cdot \left(ux \cdot ux\right)}
\end{array}
Derivation
  1. Initial program 57.0%

    \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
  2. Taylor expanded in ux around -inf

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
  3. Step-by-step derivation
    1. *-commutativeN/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    2. lower-*.f32N/A

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
  4. Applied rewrites98.3%

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
  5. Taylor expanded in maxCos around 0

    \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
  6. Step-by-step derivation
    1. Applied rewrites96.8%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
    2. Add Preprocessing

    Alternative 8: 94.2% accurate, 1.1× speedup?

    \[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \mathbf{if}\;uy \leq 0.017000000923871994:\\ \;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\ \mathbf{else}:\\ \;\;\;\;\sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos, 2\right) \cdot ux}\\ \end{array} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (let* ((t_0 (+ (- maxCos) 1.0)))
       (if (<= uy 0.017000000923871994)
         (*
          (* uy (fma -1.3333333333333333 (* (* uy uy) (* (* PI PI) PI)) (+ PI PI)))
          (sqrt (* (- (/ (fma -2.0 maxCos 2.0) ux) (* t_0 t_0)) (* ux ux))))
         (* (sin (* PI (+ uy uy))) (sqrt (* (fma -2.0 maxCos 2.0) ux))))))
    float code(float ux, float uy, float maxCos) {
    	float t_0 = -maxCos + 1.0f;
    	float tmp;
    	if (uy <= 0.017000000923871994f) {
    		tmp = (uy * fmaf(-1.3333333333333333f, ((uy * uy) * ((((float) M_PI) * ((float) M_PI)) * ((float) M_PI))), (((float) M_PI) + ((float) M_PI)))) * sqrtf((((fmaf(-2.0f, maxCos, 2.0f) / ux) - (t_0 * t_0)) * (ux * ux)));
    	} else {
    		tmp = sinf((((float) M_PI) * (uy + uy))) * sqrtf((fmaf(-2.0f, maxCos, 2.0f) * ux));
    	}
    	return tmp;
    }
    
    function code(ux, uy, maxCos)
    	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
    	tmp = Float32(0.0)
    	if (uy <= Float32(0.017000000923871994))
    		tmp = Float32(Float32(uy * fma(Float32(-1.3333333333333333), Float32(Float32(uy * uy) * Float32(Float32(Float32(pi) * Float32(pi)) * Float32(pi))), Float32(Float32(pi) + Float32(pi)))) * sqrt(Float32(Float32(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))));
    	else
    		tmp = Float32(sin(Float32(Float32(pi) * Float32(uy + uy))) * sqrt(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) * ux)));
    	end
    	return tmp
    end
    
    \begin{array}{l}
    
    \\
    \begin{array}{l}
    t_0 := \left(-maxCos\right) + 1\\
    \mathbf{if}\;uy \leq 0.017000000923871994:\\
    \;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\
    
    \mathbf{else}:\\
    \;\;\;\;\sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos, 2\right) \cdot ux}\\
    
    
    \end{array}
    \end{array}
    
    Derivation
    1. Split input into 2 regimes
    2. if uy < 0.0170000009

      1. Initial program 57.1%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in ux around -inf

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
      3. Step-by-step derivation
        1. *-commutativeN/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
        2. lower-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      4. Applied rewrites98.4%

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
      5. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{\left(uy \cdot \left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      6. Step-by-step derivation
        1. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \color{blue}{\left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        2. lower-fma.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \color{blue}{{uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        3. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, {uy}^{2} \cdot \color{blue}{{\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        4. unpow2N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        5. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        6. unpow3N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        7. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        8. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        9. lift-PI.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        10. lift-PI.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        11. lift-PI.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        12. count-2-revN/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \mathsf{PI}\left(\right) + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        13. lower-+.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \mathsf{PI}\left(\right) + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        14. lift-PI.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        15. lift-PI.f3298.3

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      7. Applied rewrites98.3%

        \[\leadsto \color{blue}{\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), \pi + \pi\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]

      if 0.0170000009 < uy

      1. Initial program 56.5%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in ux around 0

        \[\leadsto \color{blue}{\sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \cdot \sin \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right)} \]
      3. Step-by-step derivation
        1. *-commutativeN/A

          \[\leadsto \sin \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)}} \]
        2. lower-*.f32N/A

          \[\leadsto \sin \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)}} \]
        3. associate-*r*N/A

          \[\leadsto \sin \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{ux} \cdot \left(2 - 2 \cdot maxCos\right)} \]
        4. *-commutativeN/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        5. lower-sin.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{ux \cdot \left(2 - 2 \cdot maxCos\right)}} \]
        6. *-commutativeN/A

          \[\leadsto \sin \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{ux} \cdot \left(2 - 2 \cdot maxCos\right)} \]
        7. lower-*.f32N/A

          \[\leadsto \sin \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{ux} \cdot \left(2 - 2 \cdot maxCos\right)} \]
        8. lift-PI.f32N/A

          \[\leadsto \sin \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        9. *-commutativeN/A

          \[\leadsto \sin \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        10. count-2-revN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        11. lower-+.f32N/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        12. lower-sqrt.f32N/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        13. metadata-evalN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
        14. fp-cancel-sign-sub-invN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
        15. *-commutativeN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(2 + -2 \cdot maxCos\right) \cdot ux} \]
        16. lower-*.f32N/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(2 + -2 \cdot maxCos\right) \cdot ux} \]
        17. +-commutativeN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(-2 \cdot maxCos + 2\right) \cdot ux} \]
        18. lower-fma.f3277.1

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos, 2\right) \cdot ux} \]
      4. Applied rewrites77.1%

        \[\leadsto \color{blue}{\sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos, 2\right) \cdot ux}} \]
    3. Recombined 2 regimes into one program.
    4. Add Preprocessing

    Alternative 9: 90.2% accurate, 1.2× speedup?

    \[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \mathbf{if}\;uy \leq 0.0005499999970197678:\\ \;\;\;\;\left(\left(uy + uy\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\ \mathbf{else}:\\ \;\;\;\;\sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos, 2\right) \cdot ux}\\ \end{array} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (let* ((t_0 (+ (- maxCos) 1.0)))
       (if (<= uy 0.0005499999970197678)
         (*
          (* (+ uy uy) PI)
          (sqrt
           (*
            (- (/ (* maxCos (- (* 2.0 (/ 1.0 maxCos)) 2.0)) ux) (* t_0 t_0))
            (* ux ux))))
         (* (sin (* PI (+ uy uy))) (sqrt (* (fma -2.0 maxCos 2.0) ux))))))
    float code(float ux, float uy, float maxCos) {
    	float t_0 = -maxCos + 1.0f;
    	float tmp;
    	if (uy <= 0.0005499999970197678f) {
    		tmp = ((uy + uy) * ((float) M_PI)) * sqrtf(((((maxCos * ((2.0f * (1.0f / maxCos)) - 2.0f)) / ux) - (t_0 * t_0)) * (ux * ux)));
    	} else {
    		tmp = sinf((((float) M_PI) * (uy + uy))) * sqrtf((fmaf(-2.0f, maxCos, 2.0f) * ux));
    	}
    	return tmp;
    }
    
    function code(ux, uy, maxCos)
    	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
    	tmp = Float32(0.0)
    	if (uy <= Float32(0.0005499999970197678))
    		tmp = Float32(Float32(Float32(uy + uy) * Float32(pi)) * sqrt(Float32(Float32(Float32(Float32(maxCos * Float32(Float32(Float32(2.0) * Float32(Float32(1.0) / maxCos)) - Float32(2.0))) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))));
    	else
    		tmp = Float32(sin(Float32(Float32(pi) * Float32(uy + uy))) * sqrt(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) * ux)));
    	end
    	return tmp
    end
    
    \begin{array}{l}
    
    \\
    \begin{array}{l}
    t_0 := \left(-maxCos\right) + 1\\
    \mathbf{if}\;uy \leq 0.0005499999970197678:\\
    \;\;\;\;\left(\left(uy + uy\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\
    
    \mathbf{else}:\\
    \;\;\;\;\sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos, 2\right) \cdot ux}\\
    
    
    \end{array}
    \end{array}
    
    Derivation
    1. Split input into 2 regimes
    2. if uy < 5.5e-4

      1. Initial program 57.2%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in ux around -inf

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
      3. Step-by-step derivation
        1. *-commutativeN/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
        2. lower-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      4. Applied rewrites98.4%

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
      5. Taylor expanded in maxCos around inf

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      6. Step-by-step derivation
        1. lower-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        2. lower--.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        3. lower-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        4. lower-/.f3298.0

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      7. Applied rewrites98.0%

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      8. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{\left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      9. Step-by-step derivation
        1. associate-*r*N/A

          \[\leadsto \left(\left(2 \cdot uy\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        2. count-2-revN/A

          \[\leadsto \left(\left(uy + uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        3. lift-+.f32N/A

          \[\leadsto \left(\left(uy + uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        4. lower-*.f32N/A

          \[\leadsto \left(\left(uy + uy\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        5. lift-PI.f3297.3

          \[\leadsto \left(\left(uy + uy\right) \cdot \pi\right) \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      10. Applied rewrites97.3%

        \[\leadsto \color{blue}{\left(\left(uy + uy\right) \cdot \pi\right)} \cdot \sqrt{\left(\frac{maxCos \cdot \left(2 \cdot \frac{1}{maxCos} - 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]

      if 5.5e-4 < uy

      1. Initial program 56.6%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in ux around 0

        \[\leadsto \color{blue}{\sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \cdot \sin \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right)} \]
      3. Step-by-step derivation
        1. *-commutativeN/A

          \[\leadsto \sin \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)}} \]
        2. lower-*.f32N/A

          \[\leadsto \sin \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)}} \]
        3. associate-*r*N/A

          \[\leadsto \sin \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{ux} \cdot \left(2 - 2 \cdot maxCos\right)} \]
        4. *-commutativeN/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        5. lower-sin.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{ux \cdot \left(2 - 2 \cdot maxCos\right)}} \]
        6. *-commutativeN/A

          \[\leadsto \sin \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{ux} \cdot \left(2 - 2 \cdot maxCos\right)} \]
        7. lower-*.f32N/A

          \[\leadsto \sin \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{ux} \cdot \left(2 - 2 \cdot maxCos\right)} \]
        8. lift-PI.f32N/A

          \[\leadsto \sin \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        9. *-commutativeN/A

          \[\leadsto \sin \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        10. count-2-revN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        11. lower-+.f32N/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        12. lower-sqrt.f32N/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        13. metadata-evalN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
        14. fp-cancel-sign-sub-invN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
        15. *-commutativeN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(2 + -2 \cdot maxCos\right) \cdot ux} \]
        16. lower-*.f32N/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(2 + -2 \cdot maxCos\right) \cdot ux} \]
        17. +-commutativeN/A

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(-2 \cdot maxCos + 2\right) \cdot ux} \]
        18. lower-fma.f3277.0

          \[\leadsto \sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos, 2\right) \cdot ux} \]
      4. Applied rewrites77.0%

        \[\leadsto \color{blue}{\sin \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos, 2\right) \cdot ux}} \]
    3. Recombined 2 regimes into one program.
    4. Add Preprocessing

    Alternative 10: 87.5% accurate, 1.5× speedup?

    \[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \mathbf{if}\;maxCos \leq 9.999999747378752 \cdot 10^{-6}:\\ \;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \pi\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)}\\ \mathbf{else}:\\ \;\;\;\;\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\ \end{array} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (let* ((t_0 (+ (- maxCos) 1.0)))
       (if (<= maxCos 9.999999747378752e-6)
         (*
          (*
           uy
           (fma -1.3333333333333333 (* (* uy uy) (* (* PI PI) PI)) (* 2.0 PI)))
          (sqrt (* (* ux ux) (- (/ 2.0 ux) 1.0))))
         (*
          (* PI (+ uy uy))
          (sqrt (* (- (/ (fma -2.0 maxCos 2.0) ux) (* t_0 t_0)) (* ux ux)))))))
    float code(float ux, float uy, float maxCos) {
    	float t_0 = -maxCos + 1.0f;
    	float tmp;
    	if (maxCos <= 9.999999747378752e-6f) {
    		tmp = (uy * fmaf(-1.3333333333333333f, ((uy * uy) * ((((float) M_PI) * ((float) M_PI)) * ((float) M_PI))), (2.0f * ((float) M_PI)))) * sqrtf(((ux * ux) * ((2.0f / ux) - 1.0f)));
    	} else {
    		tmp = (((float) M_PI) * (uy + uy)) * sqrtf((((fmaf(-2.0f, maxCos, 2.0f) / ux) - (t_0 * t_0)) * (ux * ux)));
    	}
    	return tmp;
    }
    
    function code(ux, uy, maxCos)
    	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
    	tmp = Float32(0.0)
    	if (maxCos <= Float32(9.999999747378752e-6))
    		tmp = Float32(Float32(uy * fma(Float32(-1.3333333333333333), Float32(Float32(uy * uy) * Float32(Float32(Float32(pi) * Float32(pi)) * Float32(pi))), Float32(Float32(2.0) * Float32(pi)))) * sqrt(Float32(Float32(ux * ux) * Float32(Float32(Float32(2.0) / ux) - Float32(1.0)))));
    	else
    		tmp = Float32(Float32(Float32(pi) * Float32(uy + uy)) * sqrt(Float32(Float32(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))));
    	end
    	return tmp
    end
    
    \begin{array}{l}
    
    \\
    \begin{array}{l}
    t_0 := \left(-maxCos\right) + 1\\
    \mathbf{if}\;maxCos \leq 9.999999747378752 \cdot 10^{-6}:\\
    \;\;\;\;\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \pi\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)}\\
    
    \mathbf{else}:\\
    \;\;\;\;\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}\\
    
    
    \end{array}
    \end{array}
    
    Derivation
    1. Split input into 2 regimes
    2. if maxCos < 9.99999975e-6

      1. Initial program 57.2%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in ux around -inf

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
      3. Step-by-step derivation
        1. *-commutativeN/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
        2. lower-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      4. Applied rewrites98.3%

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
      5. Taylor expanded in maxCos around 0

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
      6. Step-by-step derivation
        1. lower--.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
        2. lower-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
        3. lift-/.f3297.8

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      7. Applied rewrites97.8%

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
      8. Step-by-step derivation
        1. lift-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \color{blue}{\left(ux \cdot ux\right)}} \]
        2. lift-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot \color{blue}{ux}\right)} \]
        3. pow2N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot {ux}^{\color{blue}{2}}} \]
        4. *-commutativeN/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{{ux}^{2} \cdot \color{blue}{\left(2 \cdot \frac{1}{ux} - 1\right)}} \]
        5. lower-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{{ux}^{2} \cdot \color{blue}{\left(2 \cdot \frac{1}{ux} - 1\right)}} \]
        6. pow2N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\color{blue}{2 \cdot \frac{1}{ux}} - 1\right)} \]
        7. lift-*.f3297.8

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\color{blue}{2 \cdot \frac{1}{ux}} - 1\right)} \]
        8. lift-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(2 \cdot \frac{1}{ux} - 1\right)} \]
        9. lift-/.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(2 \cdot \frac{1}{ux} - 1\right)} \]
        10. associate-*r/N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2 \cdot 1}{ux} - 1\right)} \]
        11. metadata-evalN/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        12. lower-/.f3297.8

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
      9. Applied rewrites97.8%

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \color{blue}{\left(\frac{2}{ux} - 1\right)}} \]
      10. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{\left(uy \cdot \left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
      11. Step-by-step derivation
        1. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \color{blue}{\left(\frac{-4}{3} \cdot \left({uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}\right) + 2 \cdot \mathsf{PI}\left(\right)\right)}\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        2. lower-fma.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \color{blue}{{uy}^{2} \cdot {\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        3. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, {uy}^{2} \cdot \color{blue}{{\mathsf{PI}\left(\right)}^{3}}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        4. unpow2N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        5. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot {\color{blue}{\mathsf{PI}\left(\right)}}^{3}, 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        6. unpow3N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        7. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\mathsf{PI}\left(\right)}\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        8. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        9. lift-PI.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        10. lift-PI.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \mathsf{PI}\left(\right)\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        11. lift-PI.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        12. lower-*.f32N/A

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(\frac{-4}{3}, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
        13. lift-PI.f3288.4

          \[\leadsto \left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \pi\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
      12. Applied rewrites88.4%

        \[\leadsto \color{blue}{\left(uy \cdot \mathsf{fma}\left(-1.3333333333333333, \left(uy \cdot uy\right) \cdot \left(\left(\pi \cdot \pi\right) \cdot \pi\right), 2 \cdot \pi\right)\right)} \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]

      if 9.99999975e-6 < maxCos

      1. Initial program 55.6%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in ux around -inf

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
      3. Step-by-step derivation
        1. *-commutativeN/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
        2. lower-*.f32N/A

          \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      4. Applied rewrites98.2%

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
      5. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{\left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      6. Step-by-step derivation
        1. count-2-revN/A

          \[\leadsto \left(uy \cdot \mathsf{PI}\left(\right) + \color{blue}{uy \cdot \mathsf{PI}\left(\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        2. distribute-rgt-inN/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \color{blue}{\left(uy + uy\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        3. lift-+.f32N/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy + \color{blue}{uy}\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        4. lift-*.f32N/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \color{blue}{\left(uy + uy\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
        5. lift-PI.f3282.3

          \[\leadsto \left(\pi \cdot \left(\color{blue}{uy} + uy\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      7. Applied rewrites82.3%

        \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    3. Recombined 2 regimes into one program.
    4. Add Preprocessing

    Alternative 11: 81.1% accurate, 1.7× speedup?

    \[\begin{array}{l} \\ \begin{array}{l} t_0 := \left(-maxCos\right) + 1\\ \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)} \end{array} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (let* ((t_0 (+ (- maxCos) 1.0)))
       (*
        (* PI (+ uy uy))
        (sqrt (* (- (/ (fma -2.0 maxCos 2.0) ux) (* t_0 t_0)) (* ux ux))))))
    float code(float ux, float uy, float maxCos) {
    	float t_0 = -maxCos + 1.0f;
    	return (((float) M_PI) * (uy + uy)) * sqrtf((((fmaf(-2.0f, maxCos, 2.0f) / ux) - (t_0 * t_0)) * (ux * ux)));
    }
    
    function code(ux, uy, maxCos)
    	t_0 = Float32(Float32(-maxCos) + Float32(1.0))
    	return Float32(Float32(Float32(pi) * Float32(uy + uy)) * sqrt(Float32(Float32(Float32(fma(Float32(-2.0), maxCos, Float32(2.0)) / ux) - Float32(t_0 * t_0)) * Float32(ux * ux))))
    end
    
    \begin{array}{l}
    
    \\
    \begin{array}{l}
    t_0 := \left(-maxCos\right) + 1\\
    \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - t\_0 \cdot t\_0\right) \cdot \left(ux \cdot ux\right)}
    \end{array}
    \end{array}
    
    Derivation
    1. Initial program 57.0%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in ux around -inf

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
    3. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    4. Applied rewrites98.3%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
    5. Taylor expanded in uy around 0

      \[\leadsto \color{blue}{\left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    6. Step-by-step derivation
      1. count-2-revN/A

        \[\leadsto \left(uy \cdot \mathsf{PI}\left(\right) + \color{blue}{uy \cdot \mathsf{PI}\left(\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      2. distribute-rgt-inN/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \color{blue}{\left(uy + uy\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      3. lift-+.f32N/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy + \color{blue}{uy}\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      4. lift-*.f32N/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \color{blue}{\left(uy + uy\right)}\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
      5. lift-PI.f3281.0

        \[\leadsto \left(\pi \cdot \left(\color{blue}{uy} + uy\right)\right) \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    7. Applied rewrites81.0%

      \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right)} \cdot \sqrt{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)} \]
    8. Add Preprocessing

    Alternative 12: 81.0% accurate, 1.9× speedup?

    \[\begin{array}{l} \\ \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - \left(maxCos + maxCos\right)\right)} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (*
      (* PI (+ uy uy))
      (sqrt
       (*
        ux
        (-
         (+ 2.0 (- (* ux (* (- maxCos 1.0) (- maxCos 1.0)))))
         (+ maxCos maxCos))))))
    float code(float ux, float uy, float maxCos) {
    	return (((float) M_PI) * (uy + uy)) * sqrtf((ux * ((2.0f + -(ux * ((maxCos - 1.0f) * (maxCos - 1.0f)))) - (maxCos + maxCos))));
    }
    
    function code(ux, uy, maxCos)
    	return Float32(Float32(Float32(pi) * Float32(uy + uy)) * sqrt(Float32(ux * Float32(Float32(Float32(2.0) + Float32(-Float32(ux * Float32(Float32(maxCos - Float32(1.0)) * Float32(maxCos - Float32(1.0)))))) - Float32(maxCos + maxCos)))))
    end
    
    function tmp = code(ux, uy, maxCos)
    	tmp = (single(pi) * (uy + uy)) * sqrt((ux * ((single(2.0) + -(ux * ((maxCos - single(1.0)) * (maxCos - single(1.0))))) - (maxCos + maxCos))));
    end
    
    \begin{array}{l}
    
    \\
    \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - \left(maxCos + maxCos\right)\right)}
    \end{array}
    
    Derivation
    1. Initial program 57.0%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in uy around 0

      \[\leadsto \color{blue}{2 \cdot \left(\left(uy \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}\right)} \]
    3. Step-by-step derivation
      1. associate-*r*N/A

        \[\leadsto \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      2. associate-*r*N/A

        \[\leadsto \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      3. *-commutativeN/A

        \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      4. lower-*.f32N/A

        \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      5. *-commutativeN/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      6. lower-*.f32N/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      7. lift-PI.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      8. *-commutativeN/A

        \[\leadsto \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      9. count-2-revN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      10. lower-+.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      11. lower-sqrt.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      12. lower--.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      13. unpow2N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
      14. lower-*.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
    4. Applied rewrites49.9%

      \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right) \cdot \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right)}} \]
    5. Taylor expanded in ux around 0

      \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + -1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right)} \]
    6. Step-by-step derivation
      1. lower-*.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + -1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right)} \]
      2. lower--.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + -1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right)} \]
      3. lower-+.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + -1 \cdot \left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right)} \]
      4. mul-1-negN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(\mathsf{neg}\left(ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right)\right) - 2 \cdot maxCos\right)} \]
      5. lower-neg.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right)} \]
      6. lower-*.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot {\left(maxCos - 1\right)}^{2}\right)\right) - 2 \cdot maxCos\right)} \]
      7. pow2N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - 2 \cdot maxCos\right)} \]
      8. lift--.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - 2 \cdot maxCos\right)} \]
      9. lift--.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - 2 \cdot maxCos\right)} \]
      10. lift-*.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - 2 \cdot maxCos\right)} \]
      11. count-2-revN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - \left(maxCos + maxCos\right)\right)} \]
      12. lift-+.f3281.1

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - \left(maxCos + maxCos\right)\right)} \]
    7. Applied rewrites81.1%

      \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(\left(2 + \left(-ux \cdot \left(\left(maxCos - 1\right) \cdot \left(maxCos - 1\right)\right)\right)\right) - \left(maxCos + maxCos\right)\right)} \]
    8. Add Preprocessing

    Alternative 13: 76.5% accurate, 2.7× speedup?

    \[\begin{array}{l} \\ \left(2 \cdot \left(uy \cdot \pi\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (* (* 2.0 (* uy PI)) (sqrt (* (* ux ux) (- (/ 2.0 ux) 1.0)))))
    float code(float ux, float uy, float maxCos) {
    	return (2.0f * (uy * ((float) M_PI))) * sqrtf(((ux * ux) * ((2.0f / ux) - 1.0f)));
    }
    
    function code(ux, uy, maxCos)
    	return Float32(Float32(Float32(2.0) * Float32(uy * Float32(pi))) * sqrt(Float32(Float32(ux * ux) * Float32(Float32(Float32(2.0) / ux) - Float32(1.0)))))
    end
    
    function tmp = code(ux, uy, maxCos)
    	tmp = (single(2.0) * (uy * single(pi))) * sqrt(((ux * ux) * ((single(2.0) / ux) - single(1.0))));
    end
    
    \begin{array}{l}
    
    \\
    \left(2 \cdot \left(uy \cdot \pi\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)}
    \end{array}
    
    Derivation
    1. Initial program 57.0%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in ux around -inf

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{{ux}^{2} \cdot \left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right)}} \]
    3. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(\left(-2 \cdot \frac{maxCos}{ux} + 2 \cdot \frac{1}{ux}\right) - {\left(1 + -1 \cdot maxCos\right)}^{2}\right) \cdot \color{blue}{{ux}^{2}}} \]
    4. Applied rewrites98.3%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\color{blue}{\left(\frac{\mathsf{fma}\left(-2, maxCos, 2\right)}{ux} - \left(\left(-maxCos\right) + 1\right) \cdot \left(\left(-maxCos\right) + 1\right)\right) \cdot \left(ux \cdot ux\right)}} \]
    5. Taylor expanded in maxCos around 0

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
    6. Step-by-step derivation
      1. lower--.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      2. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
      3. lift-/.f3292.1

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot ux\right)} \]
    7. Applied rewrites92.1%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(\color{blue}{ux} \cdot ux\right)} \]
    8. Step-by-step derivation
      1. lift-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \color{blue}{\left(ux \cdot ux\right)}} \]
      2. lift-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot \left(ux \cdot \color{blue}{ux}\right)} \]
      3. pow2N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(2 \cdot \frac{1}{ux} - 1\right) \cdot {ux}^{\color{blue}{2}}} \]
      4. *-commutativeN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{{ux}^{2} \cdot \color{blue}{\left(2 \cdot \frac{1}{ux} - 1\right)}} \]
      5. lower-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{{ux}^{2} \cdot \color{blue}{\left(2 \cdot \frac{1}{ux} - 1\right)}} \]
      6. pow2N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\color{blue}{2 \cdot \frac{1}{ux}} - 1\right)} \]
      7. lift-*.f3292.1

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\color{blue}{2 \cdot \frac{1}{ux}} - 1\right)} \]
      8. lift-*.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(2 \cdot \frac{1}{ux} - 1\right)} \]
      9. lift-/.f32N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(2 \cdot \frac{1}{ux} - 1\right)} \]
      10. associate-*r/N/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2 \cdot 1}{ux} - 1\right)} \]
      11. metadata-evalN/A

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
      12. lower-/.f3292.1

        \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
    9. Applied rewrites92.1%

      \[\leadsto \sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \color{blue}{\left(\frac{2}{ux} - 1\right)}} \]
    10. Taylor expanded in uy around 0

      \[\leadsto \color{blue}{\left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right)} \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
    11. Step-by-step derivation
      1. lower-*.f32N/A

        \[\leadsto \left(2 \cdot \color{blue}{\left(uy \cdot \mathsf{PI}\left(\right)\right)}\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
      2. lower-*.f32N/A

        \[\leadsto \left(2 \cdot \left(uy \cdot \color{blue}{\mathsf{PI}\left(\right)}\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
      3. lift-PI.f3276.5

        \[\leadsto \left(2 \cdot \left(uy \cdot \pi\right)\right) \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
    12. Applied rewrites76.5%

      \[\leadsto \color{blue}{\left(2 \cdot \left(uy \cdot \pi\right)\right)} \cdot \sqrt{\left(ux \cdot ux\right) \cdot \left(\frac{2}{ux} - 1\right)} \]
    13. Add Preprocessing

    Alternative 14: 75.2% accurate, 1.3× speedup?

    \[\begin{array}{l} \\ \begin{array}{l} t_0 := \pi \cdot \left(uy + uy\right)\\ t_1 := \left(1 - ux\right) + ux \cdot maxCos\\ \mathbf{if}\;\sqrt{1 - t\_1 \cdot t\_1} \leq 0.026000000536441803:\\ \;\;\;\;t\_0 \cdot \sqrt{\mathsf{fma}\left(-2, maxCos \cdot ux, 2 \cdot ux\right)}\\ \mathbf{else}:\\ \;\;\;\;t\_0 \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)}\\ \end{array} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (let* ((t_0 (* PI (+ uy uy))) (t_1 (+ (- 1.0 ux) (* ux maxCos))))
       (if (<= (sqrt (- 1.0 (* t_1 t_1))) 0.026000000536441803)
         (* t_0 (sqrt (fma -2.0 (* maxCos ux) (* 2.0 ux))))
         (* t_0 (sqrt (- 1.0 (* (- 1.0 ux) (- 1.0 ux))))))))
    float code(float ux, float uy, float maxCos) {
    	float t_0 = ((float) M_PI) * (uy + uy);
    	float t_1 = (1.0f - ux) + (ux * maxCos);
    	float tmp;
    	if (sqrtf((1.0f - (t_1 * t_1))) <= 0.026000000536441803f) {
    		tmp = t_0 * sqrtf(fmaf(-2.0f, (maxCos * ux), (2.0f * ux)));
    	} else {
    		tmp = t_0 * sqrtf((1.0f - ((1.0f - ux) * (1.0f - ux))));
    	}
    	return tmp;
    }
    
    function code(ux, uy, maxCos)
    	t_0 = Float32(Float32(pi) * Float32(uy + uy))
    	t_1 = Float32(Float32(Float32(1.0) - ux) + Float32(ux * maxCos))
    	tmp = Float32(0.0)
    	if (sqrt(Float32(Float32(1.0) - Float32(t_1 * t_1))) <= Float32(0.026000000536441803))
    		tmp = Float32(t_0 * sqrt(fma(Float32(-2.0), Float32(maxCos * ux), Float32(Float32(2.0) * ux))));
    	else
    		tmp = Float32(t_0 * sqrt(Float32(Float32(1.0) - Float32(Float32(Float32(1.0) - ux) * Float32(Float32(1.0) - ux)))));
    	end
    	return tmp
    end
    
    \begin{array}{l}
    
    \\
    \begin{array}{l}
    t_0 := \pi \cdot \left(uy + uy\right)\\
    t_1 := \left(1 - ux\right) + ux \cdot maxCos\\
    \mathbf{if}\;\sqrt{1 - t\_1 \cdot t\_1} \leq 0.026000000536441803:\\
    \;\;\;\;t\_0 \cdot \sqrt{\mathsf{fma}\left(-2, maxCos \cdot ux, 2 \cdot ux\right)}\\
    
    \mathbf{else}:\\
    \;\;\;\;t\_0 \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)}\\
    
    
    \end{array}
    \end{array}
    
    Derivation
    1. Split input into 2 regimes
    2. if (sqrt.f32 (-.f32 #s(literal 1 binary32) (*.f32 (+.f32 (-.f32 #s(literal 1 binary32) ux) (*.f32 ux maxCos)) (+.f32 (-.f32 #s(literal 1 binary32) ux) (*.f32 ux maxCos))))) < 0.0260000005

      1. Initial program 38.4%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{2 \cdot \left(\left(uy \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}\right)} \]
      3. Step-by-step derivation
        1. associate-*r*N/A

          \[\leadsto \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        2. associate-*r*N/A

          \[\leadsto \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        3. *-commutativeN/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        4. lower-*.f32N/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        5. *-commutativeN/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        6. lower-*.f32N/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        7. lift-PI.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        8. *-commutativeN/A

          \[\leadsto \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        9. count-2-revN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        10. lower-+.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        11. lower-sqrt.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        12. lower--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        13. unpow2N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
        14. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
      4. Applied rewrites35.2%

        \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right) \cdot \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right)}} \]
      5. Taylor expanded in ux around 0

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
      6. Step-by-step derivation
        1. metadata-evalN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
        2. fp-cancel-sign-sub-invN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
        3. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
        4. fp-cancel-sign-sub-invN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
        5. metadata-evalN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        6. lower--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        7. count-2-revN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
        8. lift-+.f3276.4

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
      7. Applied rewrites76.4%

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
      8. Taylor expanded in maxCos around 0

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{-2 \cdot \left(maxCos \cdot ux\right) + 2 \cdot ux} \]
      9. Step-by-step derivation
        1. lower-fma.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos \cdot ux, 2 \cdot ux\right)} \]
        2. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos \cdot ux, 2 \cdot ux\right)} \]
        3. lower-*.f3276.4

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos \cdot ux, 2 \cdot ux\right)} \]
      10. Applied rewrites76.4%

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{\mathsf{fma}\left(-2, maxCos \cdot ux, 2 \cdot ux\right)} \]

      if 0.0260000005 < (sqrt.f32 (-.f32 #s(literal 1 binary32) (*.f32 (+.f32 (-.f32 #s(literal 1 binary32) ux) (*.f32 ux maxCos)) (+.f32 (-.f32 #s(literal 1 binary32) ux) (*.f32 ux maxCos)))))

      1. Initial program 90.6%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{2 \cdot \left(\left(uy \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}\right)} \]
      3. Step-by-step derivation
        1. associate-*r*N/A

          \[\leadsto \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        2. associate-*r*N/A

          \[\leadsto \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        3. *-commutativeN/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        4. lower-*.f32N/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        5. *-commutativeN/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        6. lower-*.f32N/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        7. lift-PI.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        8. *-commutativeN/A

          \[\leadsto \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        9. count-2-revN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        10. lower-+.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        11. lower-sqrt.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        12. lower--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        13. unpow2N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
        14. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
      4. Applied rewrites76.5%

        \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right) \cdot \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right)}} \]
      5. Taylor expanded in maxCos around 0

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(1 - ux\right)}^{2}} \]
      6. Step-by-step derivation
        1. unpow2N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
        2. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
        3. lift--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
        4. lift--.f3273.1

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
      7. Applied rewrites73.1%

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
    3. Recombined 2 regimes into one program.
    4. Add Preprocessing

    Alternative 15: 75.2% accurate, 1.3× speedup?

    \[\begin{array}{l} \\ \begin{array}{l} t_0 := \pi \cdot \left(uy + uy\right)\\ t_1 := \left(1 - ux\right) + ux \cdot maxCos\\ \mathbf{if}\;\sqrt{1 - t\_1 \cdot t\_1} \leq 0.026000000536441803:\\ \;\;\;\;t\_0 \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)}\\ \mathbf{else}:\\ \;\;\;\;t\_0 \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)}\\ \end{array} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (let* ((t_0 (* PI (+ uy uy))) (t_1 (+ (- 1.0 ux) (* ux maxCos))))
       (if (<= (sqrt (- 1.0 (* t_1 t_1))) 0.026000000536441803)
         (* t_0 (sqrt (* ux (- 2.0 (+ maxCos maxCos)))))
         (* t_0 (sqrt (- 1.0 (* (- 1.0 ux) (- 1.0 ux))))))))
    float code(float ux, float uy, float maxCos) {
    	float t_0 = ((float) M_PI) * (uy + uy);
    	float t_1 = (1.0f - ux) + (ux * maxCos);
    	float tmp;
    	if (sqrtf((1.0f - (t_1 * t_1))) <= 0.026000000536441803f) {
    		tmp = t_0 * sqrtf((ux * (2.0f - (maxCos + maxCos))));
    	} else {
    		tmp = t_0 * sqrtf((1.0f - ((1.0f - ux) * (1.0f - ux))));
    	}
    	return tmp;
    }
    
    function code(ux, uy, maxCos)
    	t_0 = Float32(Float32(pi) * Float32(uy + uy))
    	t_1 = Float32(Float32(Float32(1.0) - ux) + Float32(ux * maxCos))
    	tmp = Float32(0.0)
    	if (sqrt(Float32(Float32(1.0) - Float32(t_1 * t_1))) <= Float32(0.026000000536441803))
    		tmp = Float32(t_0 * sqrt(Float32(ux * Float32(Float32(2.0) - Float32(maxCos + maxCos)))));
    	else
    		tmp = Float32(t_0 * sqrt(Float32(Float32(1.0) - Float32(Float32(Float32(1.0) - ux) * Float32(Float32(1.0) - ux)))));
    	end
    	return tmp
    end
    
    function tmp_2 = code(ux, uy, maxCos)
    	t_0 = single(pi) * (uy + uy);
    	t_1 = (single(1.0) - ux) + (ux * maxCos);
    	tmp = single(0.0);
    	if (sqrt((single(1.0) - (t_1 * t_1))) <= single(0.026000000536441803))
    		tmp = t_0 * sqrt((ux * (single(2.0) - (maxCos + maxCos))));
    	else
    		tmp = t_0 * sqrt((single(1.0) - ((single(1.0) - ux) * (single(1.0) - ux))));
    	end
    	tmp_2 = tmp;
    end
    
    \begin{array}{l}
    
    \\
    \begin{array}{l}
    t_0 := \pi \cdot \left(uy + uy\right)\\
    t_1 := \left(1 - ux\right) + ux \cdot maxCos\\
    \mathbf{if}\;\sqrt{1 - t\_1 \cdot t\_1} \leq 0.026000000536441803:\\
    \;\;\;\;t\_0 \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)}\\
    
    \mathbf{else}:\\
    \;\;\;\;t\_0 \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)}\\
    
    
    \end{array}
    \end{array}
    
    Derivation
    1. Split input into 2 regimes
    2. if (sqrt.f32 (-.f32 #s(literal 1 binary32) (*.f32 (+.f32 (-.f32 #s(literal 1 binary32) ux) (*.f32 ux maxCos)) (+.f32 (-.f32 #s(literal 1 binary32) ux) (*.f32 ux maxCos))))) < 0.0260000005

      1. Initial program 38.4%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{2 \cdot \left(\left(uy \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}\right)} \]
      3. Step-by-step derivation
        1. associate-*r*N/A

          \[\leadsto \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        2. associate-*r*N/A

          \[\leadsto \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        3. *-commutativeN/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        4. lower-*.f32N/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        5. *-commutativeN/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        6. lower-*.f32N/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        7. lift-PI.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        8. *-commutativeN/A

          \[\leadsto \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        9. count-2-revN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        10. lower-+.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        11. lower-sqrt.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        12. lower--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        13. unpow2N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
        14. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
      4. Applied rewrites35.2%

        \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right) \cdot \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right)}} \]
      5. Taylor expanded in ux around 0

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
      6. Step-by-step derivation
        1. metadata-evalN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
        2. fp-cancel-sign-sub-invN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
        3. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
        4. fp-cancel-sign-sub-invN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
        5. metadata-evalN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        6. lower--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
        7. count-2-revN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
        8. lift-+.f3276.4

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
      7. Applied rewrites76.4%

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]

      if 0.0260000005 < (sqrt.f32 (-.f32 #s(literal 1 binary32) (*.f32 (+.f32 (-.f32 #s(literal 1 binary32) ux) (*.f32 ux maxCos)) (+.f32 (-.f32 #s(literal 1 binary32) ux) (*.f32 ux maxCos)))))

      1. Initial program 90.6%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{2 \cdot \left(\left(uy \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}\right)} \]
      3. Step-by-step derivation
        1. associate-*r*N/A

          \[\leadsto \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        2. associate-*r*N/A

          \[\leadsto \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        3. *-commutativeN/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        4. lower-*.f32N/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        5. *-commutativeN/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        6. lower-*.f32N/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        7. lift-PI.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        8. *-commutativeN/A

          \[\leadsto \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        9. count-2-revN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        10. lower-+.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        11. lower-sqrt.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        12. lower--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        13. unpow2N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
        14. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
      4. Applied rewrites76.5%

        \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right) \cdot \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right)}} \]
      5. Taylor expanded in maxCos around 0

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(1 - ux\right)}^{2}} \]
      6. Step-by-step derivation
        1. unpow2N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
        2. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
        3. lift--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
        4. lift--.f3273.1

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
      7. Applied rewrites73.1%

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(1 - ux\right) \cdot \left(1 - ux\right)} \]
    3. Recombined 2 regimes into one program.
    4. Add Preprocessing

    Alternative 16: 65.8% accurate, 3.4× speedup?

    \[\begin{array}{l} \\ \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (* (* PI (+ uy uy)) (sqrt (* ux (- 2.0 (+ maxCos maxCos))))))
    float code(float ux, float uy, float maxCos) {
    	return (((float) M_PI) * (uy + uy)) * sqrtf((ux * (2.0f - (maxCos + maxCos))));
    }
    
    function code(ux, uy, maxCos)
    	return Float32(Float32(Float32(pi) * Float32(uy + uy)) * sqrt(Float32(ux * Float32(Float32(2.0) - Float32(maxCos + maxCos)))))
    end
    
    function tmp = code(ux, uy, maxCos)
    	tmp = (single(pi) * (uy + uy)) * sqrt((ux * (single(2.0) - (maxCos + maxCos))));
    end
    
    \begin{array}{l}
    
    \\
    \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)}
    \end{array}
    
    Derivation
    1. Initial program 57.0%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in uy around 0

      \[\leadsto \color{blue}{2 \cdot \left(\left(uy \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}\right)} \]
    3. Step-by-step derivation
      1. associate-*r*N/A

        \[\leadsto \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      2. associate-*r*N/A

        \[\leadsto \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      3. *-commutativeN/A

        \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      4. lower-*.f32N/A

        \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      5. *-commutativeN/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      6. lower-*.f32N/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      7. lift-PI.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      8. *-commutativeN/A

        \[\leadsto \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      9. count-2-revN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      10. lower-+.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      11. lower-sqrt.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      12. lower--.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      13. unpow2N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
      14. lower-*.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
    4. Applied rewrites49.9%

      \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right) \cdot \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right)}} \]
    5. Taylor expanded in ux around 0

      \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
    6. Step-by-step derivation
      1. metadata-evalN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
      2. fp-cancel-sign-sub-invN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
      3. lower-*.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
      4. fp-cancel-sign-sub-invN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
      5. metadata-evalN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
      6. lower--.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
      7. count-2-revN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
      8. lift-+.f3265.8

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
    7. Applied rewrites65.8%

      \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
    8. Add Preprocessing

    Alternative 17: 63.0% accurate, 4.6× speedup?

    \[\begin{array}{l} \\ \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot 2} \end{array} \]
    (FPCore (ux uy maxCos)
     :precision binary32
     (* (* PI (+ uy uy)) (sqrt (* ux 2.0))))
    float code(float ux, float uy, float maxCos) {
    	return (((float) M_PI) * (uy + uy)) * sqrtf((ux * 2.0f));
    }
    
    function code(ux, uy, maxCos)
    	return Float32(Float32(Float32(pi) * Float32(uy + uy)) * sqrt(Float32(ux * Float32(2.0))))
    end
    
    function tmp = code(ux, uy, maxCos)
    	tmp = (single(pi) * (uy + uy)) * sqrt((ux * single(2.0)));
    end
    
    \begin{array}{l}
    
    \\
    \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot 2}
    \end{array}
    
    Derivation
    1. Initial program 57.0%

      \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
    2. Taylor expanded in uy around 0

      \[\leadsto \color{blue}{2 \cdot \left(\left(uy \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}\right)} \]
    3. Step-by-step derivation
      1. associate-*r*N/A

        \[\leadsto \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      2. associate-*r*N/A

        \[\leadsto \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      3. *-commutativeN/A

        \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      4. lower-*.f32N/A

        \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      5. *-commutativeN/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      6. lower-*.f32N/A

        \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      7. lift-PI.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      8. *-commutativeN/A

        \[\leadsto \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      9. count-2-revN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      10. lower-+.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
      11. lower-sqrt.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      12. lower--.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
      13. unpow2N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
      14. lower-*.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
    4. Applied rewrites49.9%

      \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right) \cdot \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right)}} \]
    5. Taylor expanded in ux around 0

      \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
    6. Step-by-step derivation
      1. metadata-evalN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
      2. fp-cancel-sign-sub-invN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
      3. lower-*.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 + -2 \cdot maxCos\right)} \]
      4. fp-cancel-sign-sub-invN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(\mathsf{neg}\left(-2\right)\right) \cdot maxCos\right)} \]
      5. metadata-evalN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
      6. lower--.f32N/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - 2 \cdot maxCos\right)} \]
      7. count-2-revN/A

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
      8. lift-+.f3265.8

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
    7. Applied rewrites65.8%

      \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot \left(2 - \left(maxCos + maxCos\right)\right)} \]
    8. Taylor expanded in maxCos around 0

      \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot 2} \]
    9. Step-by-step derivation
      1. Applied rewrites63.0%

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{ux \cdot 2} \]
      2. Add Preprocessing

      Alternative 18: 7.1% accurate, 4.7× speedup?

      \[\begin{array}{l} \\ \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - 1} \end{array} \]
      (FPCore (ux uy maxCos)
       :precision binary32
       (* (* PI (+ uy uy)) (sqrt (- 1.0 1.0))))
      float code(float ux, float uy, float maxCos) {
      	return (((float) M_PI) * (uy + uy)) * sqrtf((1.0f - 1.0f));
      }
      
      function code(ux, uy, maxCos)
      	return Float32(Float32(Float32(pi) * Float32(uy + uy)) * sqrt(Float32(Float32(1.0) - Float32(1.0))))
      end
      
      function tmp = code(ux, uy, maxCos)
      	tmp = (single(pi) * (uy + uy)) * sqrt((single(1.0) - single(1.0)));
      end
      
      \begin{array}{l}
      
      \\
      \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - 1}
      \end{array}
      
      Derivation
      1. Initial program 57.0%

        \[\sin \left(\left(uy \cdot 2\right) \cdot \pi\right) \cdot \sqrt{1 - \left(\left(1 - ux\right) + ux \cdot maxCos\right) \cdot \left(\left(1 - ux\right) + ux \cdot maxCos\right)} \]
      2. Taylor expanded in uy around 0

        \[\leadsto \color{blue}{2 \cdot \left(\left(uy \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}\right)} \]
      3. Step-by-step derivation
        1. associate-*r*N/A

          \[\leadsto \left(2 \cdot \left(uy \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        2. associate-*r*N/A

          \[\leadsto \left(\left(2 \cdot uy\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        3. *-commutativeN/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        4. lower-*.f32N/A

          \[\leadsto \left(\left(uy \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \color{blue}{\sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        5. *-commutativeN/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        6. lower-*.f32N/A

          \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        7. lift-PI.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy \cdot 2\right)\right) \cdot \sqrt{\color{blue}{1} - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        8. *-commutativeN/A

          \[\leadsto \left(\pi \cdot \left(2 \cdot uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        9. count-2-revN/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        10. lower-+.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \color{blue}{{\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}}} \]
        11. lower-sqrt.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        12. lower--.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - {\left(\left(1 + maxCos \cdot ux\right) - ux\right)}^{2}} \]
        13. unpow2N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
        14. lower-*.f32N/A

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\left(1 + maxCos \cdot ux\right) - ux\right) \cdot \left(\left(1 + maxCos \cdot ux\right) - ux\right)} \]
      4. Applied rewrites49.9%

        \[\leadsto \color{blue}{\left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right) \cdot \left(\mathsf{fma}\left(maxCos, ux, 1\right) - ux\right)}} \]
      5. Taylor expanded in ux around 0

        \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - 1} \]
      6. Step-by-step derivation
        1. Applied rewrites7.1%

          \[\leadsto \left(\pi \cdot \left(uy + uy\right)\right) \cdot \sqrt{1 - 1} \]
        2. Add Preprocessing

        Reproduce

        ?
        herbie shell --seed 2025106 
        (FPCore (ux uy maxCos)
          :name "UniformSampleCone, y"
          :precision binary32
          :pre (and (and (and (<= 2.328306437e-10 ux) (<= ux 1.0)) (and (<= 2.328306437e-10 uy) (<= uy 1.0))) (and (<= 0.0 maxCos) (<= maxCos 1.0)))
          (* (sin (* (* uy 2.0) PI)) (sqrt (- 1.0 (* (+ (- 1.0 ux) (* ux maxCos)) (+ (- 1.0 ux) (* ux maxCos)))))))