2-ancestry mixing, negative discriminant

Percentage Accurate: 98.5% → 99.9%
Time: 11.4s
Alternatives: 3
Speedup: 1.1×

Specification

?
\[\begin{array}{l} \\ 2 \cdot \cos \left(\frac{2 \cdot \pi}{3} + \frac{\cos^{-1} \left(\frac{-g}{h}\right)}{3}\right) \end{array} \]
(FPCore (g h)
 :precision binary64
 (* 2.0 (cos (+ (/ (* 2.0 PI) 3.0) (/ (acos (/ (- g) h)) 3.0)))))
double code(double g, double h) {
	return 2.0 * cos((((2.0 * ((double) M_PI)) / 3.0) + (acos((-g / h)) / 3.0)));
}
public static double code(double g, double h) {
	return 2.0 * Math.cos((((2.0 * Math.PI) / 3.0) + (Math.acos((-g / h)) / 3.0)));
}
def code(g, h):
	return 2.0 * math.cos((((2.0 * math.pi) / 3.0) + (math.acos((-g / h)) / 3.0)))
function code(g, h)
	return Float64(2.0 * cos(Float64(Float64(Float64(2.0 * pi) / 3.0) + Float64(acos(Float64(Float64(-g) / h)) / 3.0))))
end
function tmp = code(g, h)
	tmp = 2.0 * cos((((2.0 * pi) / 3.0) + (acos((-g / h)) / 3.0)));
end
code[g_, h_] := N[(2.0 * N[Cos[N[(N[(N[(2.0 * Pi), $MachinePrecision] / 3.0), $MachinePrecision] + N[(N[ArcCos[N[((-g) / h), $MachinePrecision]], $MachinePrecision] / 3.0), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]
\begin{array}{l}

\\
2 \cdot \cos \left(\frac{2 \cdot \pi}{3} + \frac{\cos^{-1} \left(\frac{-g}{h}\right)}{3}\right)
\end{array}

Sampling outcomes in binary64 precision:

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 3 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: 98.5% accurate, 1.0× speedup?

\[\begin{array}{l} \\ 2 \cdot \cos \left(\frac{2 \cdot \pi}{3} + \frac{\cos^{-1} \left(\frac{-g}{h}\right)}{3}\right) \end{array} \]
(FPCore (g h)
 :precision binary64
 (* 2.0 (cos (+ (/ (* 2.0 PI) 3.0) (/ (acos (/ (- g) h)) 3.0)))))
double code(double g, double h) {
	return 2.0 * cos((((2.0 * ((double) M_PI)) / 3.0) + (acos((-g / h)) / 3.0)));
}
public static double code(double g, double h) {
	return 2.0 * Math.cos((((2.0 * Math.PI) / 3.0) + (Math.acos((-g / h)) / 3.0)));
}
def code(g, h):
	return 2.0 * math.cos((((2.0 * math.pi) / 3.0) + (math.acos((-g / h)) / 3.0)))
function code(g, h)
	return Float64(2.0 * cos(Float64(Float64(Float64(2.0 * pi) / 3.0) + Float64(acos(Float64(Float64(-g) / h)) / 3.0))))
end
function tmp = code(g, h)
	tmp = 2.0 * cos((((2.0 * pi) / 3.0) + (acos((-g / h)) / 3.0)));
end
code[g_, h_] := N[(2.0 * N[Cos[N[(N[(N[(2.0 * Pi), $MachinePrecision] / 3.0), $MachinePrecision] + N[(N[ArcCos[N[((-g) / h), $MachinePrecision]], $MachinePrecision] / 3.0), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]
\begin{array}{l}

\\
2 \cdot \cos \left(\frac{2 \cdot \pi}{3} + \frac{\cos^{-1} \left(\frac{-g}{h}\right)}{3}\right)
\end{array}

Alternative 1: 99.9% accurate, 0.4× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \cos^{-1} \left(0 - \frac{g}{h}\right)\\ t_1 := \cos \left(\mathsf{fma}\left(t\_0, -0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right)\\ t\_1 \cdot \left(\cos \left(\mathsf{fma}\left(t\_0, 0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right) \cdot \frac{2}{t\_1}\right) \end{array} \end{array} \]
(FPCore (g h)
 :precision binary64
 (let* ((t_0 (acos (- 0.0 (/ g h))))
        (t_1 (cos (fma t_0 -0.3333333333333333 (* PI 0.6666666666666666)))))
   (*
    t_1
    (*
     (cos (fma t_0 0.3333333333333333 (* PI 0.6666666666666666)))
     (/ 2.0 t_1)))))
double code(double g, double h) {
	double t_0 = acos((0.0 - (g / h)));
	double t_1 = cos(fma(t_0, -0.3333333333333333, (((double) M_PI) * 0.6666666666666666)));
	return t_1 * (cos(fma(t_0, 0.3333333333333333, (((double) M_PI) * 0.6666666666666666))) * (2.0 / t_1));
}
function code(g, h)
	t_0 = acos(Float64(0.0 - Float64(g / h)))
	t_1 = cos(fma(t_0, -0.3333333333333333, Float64(pi * 0.6666666666666666)))
	return Float64(t_1 * Float64(cos(fma(t_0, 0.3333333333333333, Float64(pi * 0.6666666666666666))) * Float64(2.0 / t_1)))
end
code[g_, h_] := Block[{t$95$0 = N[ArcCos[N[(0.0 - N[(g / h), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]}, Block[{t$95$1 = N[Cos[N[(t$95$0 * -0.3333333333333333 + N[(Pi * 0.6666666666666666), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]}, N[(t$95$1 * N[(N[Cos[N[(t$95$0 * 0.3333333333333333 + N[(Pi * 0.6666666666666666), $MachinePrecision]), $MachinePrecision]], $MachinePrecision] * N[(2.0 / t$95$1), $MachinePrecision]), $MachinePrecision]), $MachinePrecision]]]
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \cos^{-1} \left(0 - \frac{g}{h}\right)\\
t_1 := \cos \left(\mathsf{fma}\left(t\_0, -0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right)\\
t\_1 \cdot \left(\cos \left(\mathsf{fma}\left(t\_0, 0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right) \cdot \frac{2}{t\_1}\right)
\end{array}
\end{array}
Derivation
  1. Initial program 98.5%

    \[2 \cdot \cos \left(\frac{2 \cdot \pi}{3} + \frac{\cos^{-1} \left(\frac{-g}{h}\right)}{3}\right) \]
  2. Add Preprocessing
  3. Step-by-step derivation
    1. div-invN/A

      \[\leadsto 2 \cdot \cos \left(\color{blue}{\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot \frac{1}{3}} + \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right) \]
    2. div-invN/A

      \[\leadsto 2 \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot \frac{1}{3} + \color{blue}{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right) \cdot \frac{1}{3}}\right) \]
    3. distribute-rgt-outN/A

      \[\leadsto 2 \cdot \cos \color{blue}{\left(\frac{1}{3} \cdot \left(2 \cdot \mathsf{PI}\left(\right) + \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)\right)} \]
    4. *-lowering-*.f64N/A

      \[\leadsto 2 \cdot \cos \color{blue}{\left(\frac{1}{3} \cdot \left(2 \cdot \mathsf{PI}\left(\right) + \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)\right)} \]
    5. metadata-evalN/A

      \[\leadsto 2 \cdot \cos \left(\color{blue}{\frac{1}{3}} \cdot \left(2 \cdot \mathsf{PI}\left(\right) + \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)\right) \]
    6. accelerator-lowering-fma.f64N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \color{blue}{\mathsf{fma}\left(2, \mathsf{PI}\left(\right), \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)}\right) \]
    7. PI-lowering-PI.f64N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \color{blue}{\mathsf{PI}\left(\right)}, \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)\right) \]
    8. acos-lowering-acos.f64N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \mathsf{PI}\left(\right), \color{blue}{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}\right)\right) \]
    9. distribute-frac-negN/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \mathsf{PI}\left(\right), \cos^{-1} \color{blue}{\left(\mathsf{neg}\left(\frac{g}{h}\right)\right)}\right)\right) \]
    10. neg-sub0N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \mathsf{PI}\left(\right), \cos^{-1} \color{blue}{\left(0 - \frac{g}{h}\right)}\right)\right) \]
    11. --lowering--.f64N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \mathsf{PI}\left(\right), \cos^{-1} \color{blue}{\left(0 - \frac{g}{h}\right)}\right)\right) \]
    12. /-lowering-/.f6498.5

      \[\leadsto 2 \cdot \cos \left(0.3333333333333333 \cdot \mathsf{fma}\left(2, \pi, \cos^{-1} \left(0 - \color{blue}{\frac{g}{h}}\right)\right)\right) \]
  4. Applied egg-rr98.5%

    \[\leadsto 2 \cdot \cos \color{blue}{\left(0.3333333333333333 \cdot \mathsf{fma}\left(2, \pi, \cos^{-1} \left(0 - \frac{g}{h}\right)\right)\right)} \]
  5. Applied egg-rr98.4%

    \[\leadsto 2 \cdot \color{blue}{\left(\left(\cos \left(\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), 0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right) \cdot \cos \left(\frac{1}{\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), -0.3333333333333333, \pi \cdot 0.6666666666666666\right)} \cdot \mathsf{fma}\left(\pi, \pi \cdot 0.4444444444444444, {\left(\cos^{-1} \left(0 - \frac{g}{h}\right) \cdot 0.3333333333333333\right)}^{2}\right)\right)\right) \cdot \frac{1}{\cos \left(\frac{1}{\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), -0.3333333333333333, \pi \cdot 0.6666666666666666\right)} \cdot \mathsf{fma}\left(\pi, \pi \cdot 0.4444444444444444, {\left(\cos^{-1} \left(0 - \frac{g}{h}\right) \cdot 0.3333333333333333\right)}^{2}\right)\right)}\right)} \]
  6. Step-by-step derivation
    1. associate-*r*N/A

      \[\leadsto 2 \cdot \color{blue}{\left(\cos \left(\cos^{-1} \left(0 - \frac{g}{h}\right) \cdot \frac{1}{3} + \mathsf{PI}\left(\right) \cdot \frac{2}{3}\right) \cdot \left(\cos \left(\frac{1}{\cos^{-1} \left(0 - \frac{g}{h}\right) \cdot \frac{-1}{3} + \mathsf{PI}\left(\right) \cdot \frac{2}{3}} \cdot \left(\mathsf{PI}\left(\right) \cdot \left(\mathsf{PI}\left(\right) \cdot \frac{4}{9}\right) + {\left(\cos^{-1} \left(0 - \frac{g}{h}\right) \cdot \frac{1}{3}\right)}^{2}\right)\right) \cdot \frac{1}{\cos \left(\frac{1}{\cos^{-1} \left(0 - \frac{g}{h}\right) \cdot \frac{-1}{3} + \mathsf{PI}\left(\right) \cdot \frac{2}{3}} \cdot \left(\mathsf{PI}\left(\right) \cdot \left(\mathsf{PI}\left(\right) \cdot \frac{4}{9}\right) + {\left(\cos^{-1} \left(0 - \frac{g}{h}\right) \cdot \frac{1}{3}\right)}^{2}\right)\right)}\right)\right)} \]
  7. Applied egg-rr100.0%

    \[\leadsto \color{blue}{\left(\cos \left(\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), 0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right) \cdot \frac{2}{\cos \left(\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), -0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right)}\right) \cdot \cos \left(\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), -0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right)} \]
  8. Final simplification100.0%

    \[\leadsto \cos \left(\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), -0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right) \cdot \left(\cos \left(\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), 0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right) \cdot \frac{2}{\cos \left(\mathsf{fma}\left(\cos^{-1} \left(0 - \frac{g}{h}\right), -0.3333333333333333, \pi \cdot 0.6666666666666666\right)\right)}\right) \]
  9. Add Preprocessing

Alternative 2: 98.5% accurate, 1.1× speedup?

\[\begin{array}{l} \\ 2 \cdot \cos \left(\mathsf{fma}\left(\pi, 0.6666666666666666, \cos^{-1} \left(0 - \frac{g}{h}\right) \cdot 0.3333333333333333\right)\right) \end{array} \]
(FPCore (g h)
 :precision binary64
 (*
  2.0
  (cos
   (fma PI 0.6666666666666666 (* (acos (- 0.0 (/ g h))) 0.3333333333333333)))))
double code(double g, double h) {
	return 2.0 * cos(fma(((double) M_PI), 0.6666666666666666, (acos((0.0 - (g / h))) * 0.3333333333333333)));
}
function code(g, h)
	return Float64(2.0 * cos(fma(pi, 0.6666666666666666, Float64(acos(Float64(0.0 - Float64(g / h))) * 0.3333333333333333))))
end
code[g_, h_] := N[(2.0 * N[Cos[N[(Pi * 0.6666666666666666 + N[(N[ArcCos[N[(0.0 - N[(g / h), $MachinePrecision]), $MachinePrecision]], $MachinePrecision] * 0.3333333333333333), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]
\begin{array}{l}

\\
2 \cdot \cos \left(\mathsf{fma}\left(\pi, 0.6666666666666666, \cos^{-1} \left(0 - \frac{g}{h}\right) \cdot 0.3333333333333333\right)\right)
\end{array}
Derivation
  1. Initial program 98.5%

    \[2 \cdot \cos \left(\frac{2 \cdot \pi}{3} + \frac{\cos^{-1} \left(\frac{-g}{h}\right)}{3}\right) \]
  2. Add Preprocessing
  3. Step-by-step derivation
    1. div-invN/A

      \[\leadsto 2 \cdot \cos \left(\color{blue}{\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot \frac{1}{3}} + \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right) \]
    2. *-commutativeN/A

      \[\leadsto 2 \cdot \cos \left(\color{blue}{\left(\mathsf{PI}\left(\right) \cdot 2\right)} \cdot \frac{1}{3} + \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right) \]
    3. associate-*l*N/A

      \[\leadsto 2 \cdot \cos \left(\color{blue}{\mathsf{PI}\left(\right) \cdot \left(2 \cdot \frac{1}{3}\right)} + \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right) \]
    4. accelerator-lowering-fma.f64N/A

      \[\leadsto 2 \cdot \cos \color{blue}{\left(\mathsf{fma}\left(\mathsf{PI}\left(\right), 2 \cdot \frac{1}{3}, \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right)\right)} \]
    5. PI-lowering-PI.f64N/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\color{blue}{\mathsf{PI}\left(\right)}, 2 \cdot \frac{1}{3}, \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right)\right) \]
    6. metadata-evalN/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), 2 \cdot \color{blue}{\frac{1}{3}}, \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right)\right) \]
    7. metadata-evalN/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), \color{blue}{\frac{2}{3}}, \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right)\right) \]
    8. div-invN/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), \frac{2}{3}, \color{blue}{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right) \cdot \frac{1}{3}}\right)\right) \]
    9. *-lowering-*.f64N/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), \frac{2}{3}, \color{blue}{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right) \cdot \frac{1}{3}}\right)\right) \]
    10. acos-lowering-acos.f64N/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), \frac{2}{3}, \color{blue}{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)} \cdot \frac{1}{3}\right)\right) \]
    11. distribute-frac-negN/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), \frac{2}{3}, \cos^{-1} \color{blue}{\left(\mathsf{neg}\left(\frac{g}{h}\right)\right)} \cdot \frac{1}{3}\right)\right) \]
    12. neg-sub0N/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), \frac{2}{3}, \cos^{-1} \color{blue}{\left(0 - \frac{g}{h}\right)} \cdot \frac{1}{3}\right)\right) \]
    13. --lowering--.f64N/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), \frac{2}{3}, \cos^{-1} \color{blue}{\left(0 - \frac{g}{h}\right)} \cdot \frac{1}{3}\right)\right) \]
    14. /-lowering-/.f64N/A

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\mathsf{PI}\left(\right), \frac{2}{3}, \cos^{-1} \left(0 - \color{blue}{\frac{g}{h}}\right) \cdot \frac{1}{3}\right)\right) \]
    15. metadata-eval98.5

      \[\leadsto 2 \cdot \cos \left(\mathsf{fma}\left(\pi, 0.6666666666666666, \cos^{-1} \left(0 - \frac{g}{h}\right) \cdot \color{blue}{0.3333333333333333}\right)\right) \]
  4. Applied egg-rr98.5%

    \[\leadsto 2 \cdot \cos \color{blue}{\left(\mathsf{fma}\left(\pi, 0.6666666666666666, \cos^{-1} \left(0 - \frac{g}{h}\right) \cdot 0.3333333333333333\right)\right)} \]
  5. Add Preprocessing

Alternative 3: 98.5% accurate, 1.1× speedup?

\[\begin{array}{l} \\ 2 \cdot \cos \left(0.3333333333333333 \cdot \mathsf{fma}\left(2, \pi, \cos^{-1} \left(0 - \frac{g}{h}\right)\right)\right) \end{array} \]
(FPCore (g h)
 :precision binary64
 (* 2.0 (cos (* 0.3333333333333333 (fma 2.0 PI (acos (- 0.0 (/ g h))))))))
double code(double g, double h) {
	return 2.0 * cos((0.3333333333333333 * fma(2.0, ((double) M_PI), acos((0.0 - (g / h))))));
}
function code(g, h)
	return Float64(2.0 * cos(Float64(0.3333333333333333 * fma(2.0, pi, acos(Float64(0.0 - Float64(g / h)))))))
end
code[g_, h_] := N[(2.0 * N[Cos[N[(0.3333333333333333 * N[(2.0 * Pi + N[ArcCos[N[(0.0 - N[(g / h), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]), $MachinePrecision]], $MachinePrecision]), $MachinePrecision]
\begin{array}{l}

\\
2 \cdot \cos \left(0.3333333333333333 \cdot \mathsf{fma}\left(2, \pi, \cos^{-1} \left(0 - \frac{g}{h}\right)\right)\right)
\end{array}
Derivation
  1. Initial program 98.5%

    \[2 \cdot \cos \left(\frac{2 \cdot \pi}{3} + \frac{\cos^{-1} \left(\frac{-g}{h}\right)}{3}\right) \]
  2. Add Preprocessing
  3. Step-by-step derivation
    1. div-invN/A

      \[\leadsto 2 \cdot \cos \left(\color{blue}{\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot \frac{1}{3}} + \frac{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}{3}\right) \]
    2. div-invN/A

      \[\leadsto 2 \cdot \cos \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot \frac{1}{3} + \color{blue}{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right) \cdot \frac{1}{3}}\right) \]
    3. distribute-rgt-outN/A

      \[\leadsto 2 \cdot \cos \color{blue}{\left(\frac{1}{3} \cdot \left(2 \cdot \mathsf{PI}\left(\right) + \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)\right)} \]
    4. *-lowering-*.f64N/A

      \[\leadsto 2 \cdot \cos \color{blue}{\left(\frac{1}{3} \cdot \left(2 \cdot \mathsf{PI}\left(\right) + \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)\right)} \]
    5. metadata-evalN/A

      \[\leadsto 2 \cdot \cos \left(\color{blue}{\frac{1}{3}} \cdot \left(2 \cdot \mathsf{PI}\left(\right) + \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)\right) \]
    6. accelerator-lowering-fma.f64N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \color{blue}{\mathsf{fma}\left(2, \mathsf{PI}\left(\right), \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)}\right) \]
    7. PI-lowering-PI.f64N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \color{blue}{\mathsf{PI}\left(\right)}, \cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)\right)\right) \]
    8. acos-lowering-acos.f64N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \mathsf{PI}\left(\right), \color{blue}{\cos^{-1} \left(\frac{\mathsf{neg}\left(g\right)}{h}\right)}\right)\right) \]
    9. distribute-frac-negN/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \mathsf{PI}\left(\right), \cos^{-1} \color{blue}{\left(\mathsf{neg}\left(\frac{g}{h}\right)\right)}\right)\right) \]
    10. neg-sub0N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \mathsf{PI}\left(\right), \cos^{-1} \color{blue}{\left(0 - \frac{g}{h}\right)}\right)\right) \]
    11. --lowering--.f64N/A

      \[\leadsto 2 \cdot \cos \left(\frac{1}{3} \cdot \mathsf{fma}\left(2, \mathsf{PI}\left(\right), \cos^{-1} \color{blue}{\left(0 - \frac{g}{h}\right)}\right)\right) \]
    12. /-lowering-/.f6498.5

      \[\leadsto 2 \cdot \cos \left(0.3333333333333333 \cdot \mathsf{fma}\left(2, \pi, \cos^{-1} \left(0 - \color{blue}{\frac{g}{h}}\right)\right)\right) \]
  4. Applied egg-rr98.5%

    \[\leadsto 2 \cdot \cos \color{blue}{\left(0.3333333333333333 \cdot \mathsf{fma}\left(2, \pi, \cos^{-1} \left(0 - \frac{g}{h}\right)\right)\right)} \]
  5. Add Preprocessing

Reproduce

?
herbie shell --seed 2024194 
(FPCore (g h)
  :name "2-ancestry mixing, negative discriminant"
  :precision binary64
  (* 2.0 (cos (+ (/ (* 2.0 PI) 3.0) (/ (acos (/ (- g) h)) 3.0)))))