Migdal et al, Equation (51)

Percentage Accurate: 99.4% → 99.4%
Time: 6.3s
Alternatives: 8
Speedup: 1.1×

Specification

?
\[\begin{array}{l} \\ \frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \end{array} \]
(FPCore (k n)
 :precision binary64
 (* (/ 1.0 (sqrt k)) (pow (* (* 2.0 (PI)) n) (/ (- 1.0 k) 2.0))))
\begin{array}{l}

\\
\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\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 8 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: 99.4% accurate, 1.0× speedup?

\[\begin{array}{l} \\ \frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \end{array} \]
(FPCore (k n)
 :precision binary64
 (* (/ 1.0 (sqrt k)) (pow (* (* 2.0 (PI)) n) (/ (- 1.0 k) 2.0))))
\begin{array}{l}

\\
\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)}
\end{array}

Alternative 1: 99.4% accurate, 1.0× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\\ \frac{\sqrt{t\_0}}{\sqrt{{t\_0}^{k} \cdot k}} \end{array} \end{array} \]
(FPCore (k n)
 :precision binary64
 (let* ((t_0 (* n (* (PI) 2.0)))) (/ (sqrt t_0) (sqrt (* (pow t_0 k) k)))))
\begin{array}{l}

\\
\begin{array}{l}
t_0 := n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\\
\frac{\sqrt{t\_0}}{\sqrt{{t\_0}^{k} \cdot k}}
\end{array}
\end{array}
Derivation
  1. Initial program 99.5%

    \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
  2. Add Preprocessing
  3. Step-by-step derivation
    1. lift-pow.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \color{blue}{{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)}} \]
    2. rem-exp-logN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot {\color{blue}{\left(e^{\log \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}\right)}}^{\left(\frac{1 - k}{2}\right)} \]
    3. lift-/.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot {\left(e^{\log \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}\right)}^{\color{blue}{\left(\frac{1 - k}{2}\right)}} \]
    4. frac-2negN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot {\left(e^{\log \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}\right)}^{\color{blue}{\left(\frac{\mathsf{neg}\left(\left(1 - k\right)\right)}{\mathsf{neg}\left(2\right)}\right)}} \]
    5. distribute-frac-negN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot {\left(e^{\log \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}\right)}^{\color{blue}{\left(\mathsf{neg}\left(\frac{1 - k}{\mathsf{neg}\left(2\right)}\right)\right)}} \]
    6. pow-negN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \color{blue}{\frac{1}{{\left(e^{\log \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}\right)}^{\left(\frac{1 - k}{\mathsf{neg}\left(2\right)}\right)}}} \]
    7. lower-/.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \color{blue}{\frac{1}{{\left(e^{\log \left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}\right)}^{\left(\frac{1 - k}{\mathsf{neg}\left(2\right)}\right)}}} \]
    8. rem-exp-logN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\color{blue}{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}}^{\left(\frac{1 - k}{\mathsf{neg}\left(2\right)}\right)}} \]
    9. distribute-neg-frac2N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\color{blue}{\left(\mathsf{neg}\left(\frac{1 - k}{2}\right)\right)}}} \]
    10. lift-/.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\mathsf{neg}\left(\color{blue}{\frac{1 - k}{2}}\right)\right)}} \]
    11. lower-pow.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{\color{blue}{{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\mathsf{neg}\left(\frac{1 - k}{2}\right)\right)}}} \]
    12. lift-*.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\color{blue}{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}}^{\left(\mathsf{neg}\left(\frac{1 - k}{2}\right)\right)}} \]
    13. *-commutativeN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\color{blue}{\left(n \cdot \left(2 \cdot \mathsf{PI}\left(\right)\right)\right)}}^{\left(\mathsf{neg}\left(\frac{1 - k}{2}\right)\right)}} \]
    14. lower-*.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\color{blue}{\left(n \cdot \left(2 \cdot \mathsf{PI}\left(\right)\right)\right)}}^{\left(\mathsf{neg}\left(\frac{1 - k}{2}\right)\right)}} \]
    15. lift-*.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(n \cdot \color{blue}{\left(2 \cdot \mathsf{PI}\left(\right)\right)}\right)}^{\left(\mathsf{neg}\left(\frac{1 - k}{2}\right)\right)}} \]
    16. *-commutativeN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(n \cdot \color{blue}{\left(\mathsf{PI}\left(\right) \cdot 2\right)}\right)}^{\left(\mathsf{neg}\left(\frac{1 - k}{2}\right)\right)}} \]
    17. lower-*.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(n \cdot \color{blue}{\left(\mathsf{PI}\left(\right) \cdot 2\right)}\right)}^{\left(\mathsf{neg}\left(\frac{1 - k}{2}\right)\right)}} \]
    18. lift-/.f64N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(\mathsf{neg}\left(\color{blue}{\frac{1 - k}{2}}\right)\right)}} \]
    19. clear-numN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(\mathsf{neg}\left(\color{blue}{\frac{1}{\frac{2}{1 - k}}}\right)\right)}} \]
    20. associate-/r/N/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(\mathsf{neg}\left(\color{blue}{\frac{1}{2} \cdot \left(1 - k\right)}\right)\right)}} \]
    21. metadata-evalN/A

      \[\leadsto \frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(\mathsf{neg}\left(\color{blue}{\frac{1}{2}} \cdot \left(1 - k\right)\right)\right)}} \]
  4. Applied rewrites99.5%

    \[\leadsto \frac{1}{\sqrt{k}} \cdot \color{blue}{\frac{1}{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(-0.5 \cdot \left(1 - k\right)\right)}}} \]
  5. Step-by-step derivation
    1. lift-*.f64N/A

      \[\leadsto \color{blue}{\frac{1}{\sqrt{k}} \cdot \frac{1}{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(\frac{-1}{2} \cdot \left(1 - k\right)\right)}}} \]
    2. lift-/.f64N/A

      \[\leadsto \color{blue}{\frac{1}{\sqrt{k}}} \cdot \frac{1}{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(\frac{-1}{2} \cdot \left(1 - k\right)\right)}} \]
    3. associate-*l/N/A

      \[\leadsto \color{blue}{\frac{1 \cdot \frac{1}{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(\frac{-1}{2} \cdot \left(1 - k\right)\right)}}}{\sqrt{k}}} \]
  6. Applied rewrites99.7%

    \[\leadsto \color{blue}{\frac{\sqrt{\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n}}{\sqrt{k \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{k}}}} \]
  7. Final simplification99.7%

    \[\leadsto \frac{\sqrt{n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)}}{\sqrt{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{k} \cdot k}} \]
  8. Add Preprocessing

Alternative 2: 97.8% accurate, 1.1× speedup?

\[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;k \leq 1:\\ \;\;\;\;\sqrt{n \cdot 2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right)}{k}}\\ \mathbf{else}:\\ \;\;\;\;\frac{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(-0.5 \cdot k\right)}}{\sqrt{k}}\\ \end{array} \end{array} \]
(FPCore (k n)
 :precision binary64
 (if (<= k 1.0)
   (* (sqrt (* n 2.0)) (sqrt (/ (PI) k)))
   (/ (pow (* n (* (PI) 2.0)) (* -0.5 k)) (sqrt k))))
\begin{array}{l}

\\
\begin{array}{l}
\mathbf{if}\;k \leq 1:\\
\;\;\;\;\sqrt{n \cdot 2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right)}{k}}\\

\mathbf{else}:\\
\;\;\;\;\frac{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(-0.5 \cdot k\right)}}{\sqrt{k}}\\


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

    1. Initial program 99.0%

      \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
    2. Add Preprocessing
    3. Taylor expanded in k around 0

      \[\leadsto \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \cdot \sqrt{2}} \]
    4. Step-by-step derivation
      1. *-commutativeN/A

        \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
      2. lower-*.f64N/A

        \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
      3. lower-sqrt.f64N/A

        \[\leadsto \color{blue}{\sqrt{2}} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \]
      4. lower-sqrt.f64N/A

        \[\leadsto \sqrt{2} \cdot \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
      5. lower-/.f64N/A

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

        \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
      7. lower-*.f64N/A

        \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
      8. lower-PI.f6475.2

        \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right)} \cdot n}{k}} \]
    5. Applied rewrites75.2%

      \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k}}} \]
    6. Step-by-step derivation
      1. Applied rewrites75.4%

        \[\leadsto \color{blue}{\sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k} \cdot 2}} \]
      2. Step-by-step derivation
        1. Applied rewrites75.5%

          \[\leadsto \sqrt{n \cdot \frac{\mathsf{PI}\left(\right) \cdot 2}{k}} \]
        2. Step-by-step derivation
          1. Applied rewrites95.7%

            \[\leadsto \sqrt{\frac{\mathsf{PI}\left(\right)}{k}} \cdot \color{blue}{\sqrt{n \cdot 2}} \]

          if 1 < k

          1. Initial program 100.0%

            \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
          2. Add Preprocessing
          3. Taylor expanded in k around inf

            \[\leadsto \frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\color{blue}{\left(\frac{-1}{2} \cdot k\right)}} \]
          4. Step-by-step derivation
            1. lower-*.f64100.0

              \[\leadsto \frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\color{blue}{\left(-0.5 \cdot k\right)}} \]
          5. Applied rewrites100.0%

            \[\leadsto \frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\color{blue}{\left(-0.5 \cdot k\right)}} \]
          6. Step-by-step derivation
            1. lift-*.f64N/A

              \[\leadsto \color{blue}{\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{-1}{2} \cdot k\right)}} \]
            2. *-commutativeN/A

              \[\leadsto \color{blue}{{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{-1}{2} \cdot k\right)} \cdot \frac{1}{\sqrt{k}}} \]
            3. lift-/.f64N/A

              \[\leadsto {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{-1}{2} \cdot k\right)} \cdot \color{blue}{\frac{1}{\sqrt{k}}} \]
            4. un-div-invN/A

              \[\leadsto \color{blue}{\frac{{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{-1}{2} \cdot k\right)}}{\sqrt{k}}} \]
            5. lower-/.f64100.0

              \[\leadsto \color{blue}{\frac{{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(-0.5 \cdot k\right)}}{\sqrt{k}}} \]
          7. Applied rewrites100.0%

            \[\leadsto \color{blue}{\frac{{\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(-0.5 \cdot k\right)}}{\sqrt{k}}} \]
        3. Recombined 2 regimes into one program.
        4. Final simplification97.9%

          \[\leadsto \begin{array}{l} \mathbf{if}\;k \leq 1:\\ \;\;\;\;\sqrt{n \cdot 2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right)}{k}}\\ \mathbf{else}:\\ \;\;\;\;\frac{{\left(n \cdot \left(\mathsf{PI}\left(\right) \cdot 2\right)\right)}^{\left(-0.5 \cdot k\right)}}{\sqrt{k}}\\ \end{array} \]
        5. Add Preprocessing

        Alternative 3: 99.4% accurate, 1.1× speedup?

        \[\begin{array}{l} \\ \sqrt{\frac{1}{k}} \cdot {\left(\left(n \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)} \end{array} \]
        (FPCore (k n)
         :precision binary64
         (* (sqrt (/ 1.0 k)) (pow (* (* n 2.0) (PI)) (fma -0.5 k 0.5))))
        \begin{array}{l}
        
        \\
        \sqrt{\frac{1}{k}} \cdot {\left(\left(n \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)}
        \end{array}
        
        Derivation
        1. Initial program 99.5%

          \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
        2. Add Preprocessing
        3. Taylor expanded in k around inf

          \[\leadsto \color{blue}{\sqrt{\frac{1}{k}} \cdot e^{\frac{1}{2} \cdot \left(\log \left(2 \cdot \left(n \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(1 - k\right)\right)}} \]
        4. Step-by-step derivation
          1. *-commutativeN/A

            \[\leadsto \color{blue}{e^{\frac{1}{2} \cdot \left(\log \left(2 \cdot \left(n \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(1 - k\right)\right)} \cdot \sqrt{\frac{1}{k}}} \]
          2. lower-*.f64N/A

            \[\leadsto \color{blue}{e^{\frac{1}{2} \cdot \left(\log \left(2 \cdot \left(n \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(1 - k\right)\right)} \cdot \sqrt{\frac{1}{k}}} \]
        5. Applied rewrites99.6%

          \[\leadsto \color{blue}{{\left(\left(2 \cdot n\right) \cdot \mathsf{PI}\left(\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)} \cdot \sqrt{\frac{1}{k}}} \]
        6. Final simplification99.6%

          \[\leadsto \sqrt{\frac{1}{k}} \cdot {\left(\left(n \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)} \]
        7. Add Preprocessing

        Alternative 4: 99.4% accurate, 1.1× speedup?

        \[\begin{array}{l} \\ \frac{{\left(\left(n \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)}}{\sqrt{k}} \end{array} \]
        (FPCore (k n)
         :precision binary64
         (/ (pow (* (* n 2.0) (PI)) (fma -0.5 k 0.5)) (sqrt k)))
        \begin{array}{l}
        
        \\
        \frac{{\left(\left(n \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)}}{\sqrt{k}}
        \end{array}
        
        Derivation
        1. Initial program 99.5%

          \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
        2. Add Preprocessing
        3. Taylor expanded in k around inf

          \[\leadsto \color{blue}{\sqrt{\frac{1}{k}} \cdot e^{\frac{1}{2} \cdot \left(\log \left(2 \cdot \left(n \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(1 - k\right)\right)}} \]
        4. Step-by-step derivation
          1. *-commutativeN/A

            \[\leadsto \color{blue}{e^{\frac{1}{2} \cdot \left(\log \left(2 \cdot \left(n \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(1 - k\right)\right)} \cdot \sqrt{\frac{1}{k}}} \]
          2. lower-*.f64N/A

            \[\leadsto \color{blue}{e^{\frac{1}{2} \cdot \left(\log \left(2 \cdot \left(n \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(1 - k\right)\right)} \cdot \sqrt{\frac{1}{k}}} \]
        5. Applied rewrites99.6%

          \[\leadsto \color{blue}{{\left(\left(2 \cdot n\right) \cdot \mathsf{PI}\left(\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)} \cdot \sqrt{\frac{1}{k}}} \]
        6. Step-by-step derivation
          1. Applied rewrites99.6%

            \[\leadsto \color{blue}{\frac{{\left(\mathsf{PI}\left(\right) \cdot \left(n \cdot 2\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)}}{\sqrt{k}}} \]
          2. Final simplification99.6%

            \[\leadsto \frac{{\left(\left(n \cdot 2\right) \cdot \mathsf{PI}\left(\right)\right)}^{\left(\mathsf{fma}\left(-0.5, k, 0.5\right)\right)}}{\sqrt{k}} \]
          3. Add Preprocessing

          Alternative 5: 48.9% accurate, 3.6× speedup?

          \[\begin{array}{l} \\ \sqrt{n \cdot 2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right)}{k}} \end{array} \]
          (FPCore (k n) :precision binary64 (* (sqrt (* n 2.0)) (sqrt (/ (PI) k))))
          \begin{array}{l}
          
          \\
          \sqrt{n \cdot 2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right)}{k}}
          \end{array}
          
          Derivation
          1. Initial program 99.5%

            \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
          2. Add Preprocessing
          3. Taylor expanded in k around 0

            \[\leadsto \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \cdot \sqrt{2}} \]
          4. Step-by-step derivation
            1. *-commutativeN/A

              \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
            2. lower-*.f64N/A

              \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
            3. lower-sqrt.f64N/A

              \[\leadsto \color{blue}{\sqrt{2}} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \]
            4. lower-sqrt.f64N/A

              \[\leadsto \sqrt{2} \cdot \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
            5. lower-/.f64N/A

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

              \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
            7. lower-*.f64N/A

              \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
            8. lower-PI.f6437.7

              \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right)} \cdot n}{k}} \]
          5. Applied rewrites37.7%

            \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k}}} \]
          6. Step-by-step derivation
            1. Applied rewrites37.9%

              \[\leadsto \color{blue}{\sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k} \cdot 2}} \]
            2. Step-by-step derivation
              1. Applied rewrites37.9%

                \[\leadsto \sqrt{n \cdot \frac{\mathsf{PI}\left(\right) \cdot 2}{k}} \]
              2. Step-by-step derivation
                1. Applied rewrites47.7%

                  \[\leadsto \sqrt{\frac{\mathsf{PI}\left(\right)}{k}} \cdot \color{blue}{\sqrt{n \cdot 2}} \]
                2. Final simplification47.7%

                  \[\leadsto \sqrt{n \cdot 2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right)}{k}} \]
                3. Add Preprocessing

                Alternative 6: 48.9% accurate, 3.6× speedup?

                \[\begin{array}{l} \\ \sqrt{\frac{\mathsf{PI}\left(\right) \cdot 2}{k}} \cdot \sqrt{n} \end{array} \]
                (FPCore (k n) :precision binary64 (* (sqrt (/ (* (PI) 2.0) k)) (sqrt n)))
                \begin{array}{l}
                
                \\
                \sqrt{\frac{\mathsf{PI}\left(\right) \cdot 2}{k}} \cdot \sqrt{n}
                \end{array}
                
                Derivation
                1. Initial program 99.5%

                  \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
                2. Add Preprocessing
                3. Taylor expanded in k around 0

                  \[\leadsto \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \cdot \sqrt{2}} \]
                4. Step-by-step derivation
                  1. *-commutativeN/A

                    \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                  2. lower-*.f64N/A

                    \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                  3. lower-sqrt.f64N/A

                    \[\leadsto \color{blue}{\sqrt{2}} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \]
                  4. lower-sqrt.f64N/A

                    \[\leadsto \sqrt{2} \cdot \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                  5. lower-/.f64N/A

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

                    \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
                  7. lower-*.f64N/A

                    \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
                  8. lower-PI.f6437.7

                    \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right)} \cdot n}{k}} \]
                5. Applied rewrites37.7%

                  \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k}}} \]
                6. Step-by-step derivation
                  1. Applied rewrites47.7%

                    \[\leadsto \sqrt{n} \cdot \color{blue}{\sqrt{\frac{2 \cdot \mathsf{PI}\left(\right)}{k}}} \]
                  2. Final simplification47.7%

                    \[\leadsto \sqrt{\frac{\mathsf{PI}\left(\right) \cdot 2}{k}} \cdot \sqrt{n} \]
                  3. Add Preprocessing

                  Alternative 7: 37.3% accurate, 4.8× speedup?

                  \[\begin{array}{l} \\ \sqrt{\frac{\mathsf{PI}\left(\right) \cdot 2}{k} \cdot n} \end{array} \]
                  (FPCore (k n) :precision binary64 (sqrt (* (/ (* (PI) 2.0) k) n)))
                  \begin{array}{l}
                  
                  \\
                  \sqrt{\frac{\mathsf{PI}\left(\right) \cdot 2}{k} \cdot n}
                  \end{array}
                  
                  Derivation
                  1. Initial program 99.5%

                    \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
                  2. Add Preprocessing
                  3. Taylor expanded in k around 0

                    \[\leadsto \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \cdot \sqrt{2}} \]
                  4. Step-by-step derivation
                    1. *-commutativeN/A

                      \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                    2. lower-*.f64N/A

                      \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                    3. lower-sqrt.f64N/A

                      \[\leadsto \color{blue}{\sqrt{2}} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \]
                    4. lower-sqrt.f64N/A

                      \[\leadsto \sqrt{2} \cdot \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                    5. lower-/.f64N/A

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

                      \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
                    7. lower-*.f64N/A

                      \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
                    8. lower-PI.f6437.7

                      \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right)} \cdot n}{k}} \]
                  5. Applied rewrites37.7%

                    \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k}}} \]
                  6. Step-by-step derivation
                    1. Applied rewrites37.9%

                      \[\leadsto \color{blue}{\sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k} \cdot 2}} \]
                    2. Step-by-step derivation
                      1. Applied rewrites37.9%

                        \[\leadsto \sqrt{n \cdot \frac{\mathsf{PI}\left(\right) \cdot 2}{k}} \]
                      2. Final simplification37.9%

                        \[\leadsto \sqrt{\frac{\mathsf{PI}\left(\right) \cdot 2}{k} \cdot n} \]
                      3. Add Preprocessing

                      Alternative 8: 37.3% accurate, 4.8× speedup?

                      \[\begin{array}{l} \\ \sqrt{\left(\frac{2}{k} \cdot \mathsf{PI}\left(\right)\right) \cdot n} \end{array} \]
                      (FPCore (k n) :precision binary64 (sqrt (* (* (/ 2.0 k) (PI)) n)))
                      \begin{array}{l}
                      
                      \\
                      \sqrt{\left(\frac{2}{k} \cdot \mathsf{PI}\left(\right)\right) \cdot n}
                      \end{array}
                      
                      Derivation
                      1. Initial program 99.5%

                        \[\frac{1}{\sqrt{k}} \cdot {\left(\left(2 \cdot \mathsf{PI}\left(\right)\right) \cdot n\right)}^{\left(\frac{1 - k}{2}\right)} \]
                      2. Add Preprocessing
                      3. Taylor expanded in k around 0

                        \[\leadsto \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \cdot \sqrt{2}} \]
                      4. Step-by-step derivation
                        1. *-commutativeN/A

                          \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                        2. lower-*.f64N/A

                          \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                        3. lower-sqrt.f64N/A

                          \[\leadsto \color{blue}{\sqrt{2}} \cdot \sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}} \]
                        4. lower-sqrt.f64N/A

                          \[\leadsto \sqrt{2} \cdot \color{blue}{\sqrt{\frac{n \cdot \mathsf{PI}\left(\right)}{k}}} \]
                        5. lower-/.f64N/A

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

                          \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
                        7. lower-*.f64N/A

                          \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right) \cdot n}}{k}} \]
                        8. lower-PI.f6437.7

                          \[\leadsto \sqrt{2} \cdot \sqrt{\frac{\color{blue}{\mathsf{PI}\left(\right)} \cdot n}{k}} \]
                      5. Applied rewrites37.7%

                        \[\leadsto \color{blue}{\sqrt{2} \cdot \sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k}}} \]
                      6. Step-by-step derivation
                        1. Applied rewrites37.9%

                          \[\leadsto \color{blue}{\sqrt{\frac{\mathsf{PI}\left(\right) \cdot n}{k} \cdot 2}} \]
                        2. Step-by-step derivation
                          1. Applied rewrites37.9%

                            \[\leadsto \sqrt{n \cdot \frac{\mathsf{PI}\left(\right) \cdot 2}{k}} \]
                          2. Step-by-step derivation
                            1. Applied rewrites37.9%

                              \[\leadsto \sqrt{n \cdot \left(\frac{2}{k} \cdot \mathsf{PI}\left(\right)\right)} \]
                            2. Final simplification37.9%

                              \[\leadsto \sqrt{\left(\frac{2}{k} \cdot \mathsf{PI}\left(\right)\right) \cdot n} \]
                            3. Add Preprocessing

                            Reproduce

                            ?
                            herbie shell --seed 2024332 
                            (FPCore (k n)
                              :name "Migdal et al, Equation (51)"
                              :precision binary64
                              (* (/ 1.0 (sqrt k)) (pow (* (* 2.0 (PI)) n) (/ (- 1.0 k) 2.0))))