ab-angle->ABCF A

Percentage Accurate: 79.5% → 79.5%
Time: 6.7s
Alternatives: 8
Speedup: 2.0×

Specification

?
\[\begin{array}{l} \\ \begin{array}{l} t_0 := \frac{angle}{180} \cdot \mathsf{PI}\left(\right)\\ {\left(a \cdot \sin t\_0\right)}^{2} + {\left(b \cdot \cos t\_0\right)}^{2} \end{array} \end{array} \]
(FPCore (a b angle)
 :precision binary64
 (let* ((t_0 (* (/ angle 180.0) (PI))))
   (+ (pow (* a (sin t_0)) 2.0) (pow (* b (cos t_0)) 2.0))))
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \frac{angle}{180} \cdot \mathsf{PI}\left(\right)\\
{\left(a \cdot \sin t\_0\right)}^{2} + {\left(b \cdot \cos t\_0\right)}^{2}
\end{array}
\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: 79.5% accurate, 1.0× speedup?

\[\begin{array}{l} \\ \begin{array}{l} t_0 := \frac{angle}{180} \cdot \mathsf{PI}\left(\right)\\ {\left(a \cdot \sin t\_0\right)}^{2} + {\left(b \cdot \cos t\_0\right)}^{2} \end{array} \end{array} \]
(FPCore (a b angle)
 :precision binary64
 (let* ((t_0 (* (/ angle 180.0) (PI))))
   (+ (pow (* a (sin t_0)) 2.0) (pow (* b (cos t_0)) 2.0))))
\begin{array}{l}

\\
\begin{array}{l}
t_0 := \frac{angle}{180} \cdot \mathsf{PI}\left(\right)\\
{\left(a \cdot \sin t\_0\right)}^{2} + {\left(b \cdot \cos t\_0\right)}^{2}
\end{array}
\end{array}

Alternative 1: 79.5% accurate, 2.0× speedup?

\[\begin{array}{l} \\ \mathsf{fma}\left(b, b, {\left(\sin \left(0.005555555555555556 \cdot \left(\mathsf{PI}\left(\right) \cdot angle\right)\right) \cdot a\right)}^{2}\right) \end{array} \]
(FPCore (a b angle)
 :precision binary64
 (fma b b (pow (* (sin (* 0.005555555555555556 (* (PI) angle))) a) 2.0)))
\begin{array}{l}

\\
\mathsf{fma}\left(b, b, {\left(\sin \left(0.005555555555555556 \cdot \left(\mathsf{PI}\left(\right) \cdot angle\right)\right) \cdot a\right)}^{2}\right)
\end{array}
Derivation
  1. Initial program 77.7%

    \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
  2. Add Preprocessing
  3. Taylor expanded in angle around 0

    \[\leadsto {\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \color{blue}{1}\right)}^{2} \]
  4. Step-by-step derivation
    1. Applied rewrites77.9%

      \[\leadsto {\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \color{blue}{1}\right)}^{2} \]
    2. Taylor expanded in angle around inf

      \[\leadsto {\left(a \cdot \color{blue}{\sin \left(\frac{1}{180} \cdot \left(angle \cdot \mathsf{PI}\left(\right)\right)\right)}\right)}^{2} + {\left(b \cdot 1\right)}^{2} \]
    3. Step-by-step derivation
      1. Applied rewrites77.9%

        \[\leadsto {\left(a \cdot \color{blue}{\sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot 0.005555555555555556\right)}\right)}^{2} + {\left(b \cdot 1\right)}^{2} \]
      2. Step-by-step derivation
        1. lift-+.f64N/A

          \[\leadsto \color{blue}{{\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} + {\left(b \cdot 1\right)}^{2}} \]
        2. +-commutativeN/A

          \[\leadsto \color{blue}{{\left(b \cdot 1\right)}^{2} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2}} \]
        3. lift-pow.f64N/A

          \[\leadsto \color{blue}{{\left(b \cdot 1\right)}^{2}} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
        4. unpow2N/A

          \[\leadsto \color{blue}{\left(b \cdot 1\right) \cdot \left(b \cdot 1\right)} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
        5. lift-*.f64N/A

          \[\leadsto \left(b \cdot 1\right) \cdot \color{blue}{\left(b \cdot 1\right)} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
        6. *-commutativeN/A

          \[\leadsto \left(b \cdot 1\right) \cdot \color{blue}{\left(1 \cdot b\right)} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
        7. associate-*r*N/A

          \[\leadsto \color{blue}{\left(\left(b \cdot 1\right) \cdot 1\right) \cdot b} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
        8. lower-fma.f64N/A

          \[\leadsto \color{blue}{\mathsf{fma}\left(\left(b \cdot 1\right) \cdot 1, b, {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2}\right)} \]
      3. Applied rewrites77.9%

        \[\leadsto \color{blue}{\mathsf{fma}\left(\left(1 \cdot b\right) \cdot 1, b, {\left(\sin \left(0.005555555555555556 \cdot \left(\mathsf{PI}\left(\right) \cdot angle\right)\right) \cdot a\right)}^{2}\right)} \]
      4. Taylor expanded in angle around 0

        \[\leadsto \mathsf{fma}\left(\color{blue}{b}, b, {\left(\sin \left(\frac{1}{180} \cdot \left(\mathsf{PI}\left(\right) \cdot angle\right)\right) \cdot a\right)}^{2}\right) \]
      5. Step-by-step derivation
        1. Applied rewrites77.9%

          \[\leadsto \mathsf{fma}\left(\color{blue}{b}, b, {\left(\sin \left(0.005555555555555556 \cdot \left(\mathsf{PI}\left(\right) \cdot angle\right)\right) \cdot a\right)}^{2}\right) \]
        2. Add Preprocessing

        Alternative 2: 67.1% accurate, 3.2× speedup?

        \[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;a \leq 410000000:\\ \;\;\;\;b \cdot b\\ \mathbf{else}:\\ \;\;\;\;\mathsf{fma}\left(\left(1 \cdot b\right) \cdot 1, b, {\left(\left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot 0.005555555555555556\right) \cdot a\right)}^{2}\right)\\ \end{array} \end{array} \]
        (FPCore (a b angle)
         :precision binary64
         (if (<= a 410000000.0)
           (* b b)
           (fma
            (* (* 1.0 b) 1.0)
            b
            (pow (* (* (* (PI) angle) 0.005555555555555556) a) 2.0))))
        \begin{array}{l}
        
        \\
        \begin{array}{l}
        \mathbf{if}\;a \leq 410000000:\\
        \;\;\;\;b \cdot b\\
        
        \mathbf{else}:\\
        \;\;\;\;\mathsf{fma}\left(\left(1 \cdot b\right) \cdot 1, b, {\left(\left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot 0.005555555555555556\right) \cdot a\right)}^{2}\right)\\
        
        
        \end{array}
        \end{array}
        
        Derivation
        1. Split input into 2 regimes
        2. if a < 4.1e8

          1. Initial program 77.0%

            \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
          2. Add Preprocessing
          3. Taylor expanded in angle around 0

            \[\leadsto \color{blue}{{b}^{2}} \]
          4. Step-by-step derivation
            1. Applied rewrites65.3%

              \[\leadsto \color{blue}{b \cdot b} \]

            if 4.1e8 < a

            1. Initial program 79.9%

              \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
            2. Add Preprocessing
            3. Taylor expanded in angle around 0

              \[\leadsto {\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \color{blue}{1}\right)}^{2} \]
            4. Step-by-step derivation
              1. Applied rewrites80.0%

                \[\leadsto {\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \color{blue}{1}\right)}^{2} \]
              2. Taylor expanded in angle around inf

                \[\leadsto {\left(a \cdot \color{blue}{\sin \left(\frac{1}{180} \cdot \left(angle \cdot \mathsf{PI}\left(\right)\right)\right)}\right)}^{2} + {\left(b \cdot 1\right)}^{2} \]
              3. Step-by-step derivation
                1. Applied rewrites80.0%

                  \[\leadsto {\left(a \cdot \color{blue}{\sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot 0.005555555555555556\right)}\right)}^{2} + {\left(b \cdot 1\right)}^{2} \]
                2. Step-by-step derivation
                  1. lift-+.f64N/A

                    \[\leadsto \color{blue}{{\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} + {\left(b \cdot 1\right)}^{2}} \]
                  2. +-commutativeN/A

                    \[\leadsto \color{blue}{{\left(b \cdot 1\right)}^{2} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2}} \]
                  3. lift-pow.f64N/A

                    \[\leadsto \color{blue}{{\left(b \cdot 1\right)}^{2}} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
                  4. unpow2N/A

                    \[\leadsto \color{blue}{\left(b \cdot 1\right) \cdot \left(b \cdot 1\right)} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
                  5. lift-*.f64N/A

                    \[\leadsto \left(b \cdot 1\right) \cdot \color{blue}{\left(b \cdot 1\right)} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
                  6. *-commutativeN/A

                    \[\leadsto \left(b \cdot 1\right) \cdot \color{blue}{\left(1 \cdot b\right)} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
                  7. associate-*r*N/A

                    \[\leadsto \color{blue}{\left(\left(b \cdot 1\right) \cdot 1\right) \cdot b} + {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2} \]
                  8. lower-fma.f64N/A

                    \[\leadsto \color{blue}{\mathsf{fma}\left(\left(b \cdot 1\right) \cdot 1, b, {\left(a \cdot \sin \left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \frac{1}{180}\right)\right)}^{2}\right)} \]
                3. Applied rewrites80.0%

                  \[\leadsto \color{blue}{\mathsf{fma}\left(\left(1 \cdot b\right) \cdot 1, b, {\left(\sin \left(0.005555555555555556 \cdot \left(\mathsf{PI}\left(\right) \cdot angle\right)\right) \cdot a\right)}^{2}\right)} \]
                4. Taylor expanded in angle around 0

                  \[\leadsto \mathsf{fma}\left(\left(1 \cdot b\right) \cdot 1, b, {\left(\left(\frac{1}{180} \cdot \color{blue}{\left(angle \cdot \mathsf{PI}\left(\right)\right)}\right) \cdot a\right)}^{2}\right) \]
                5. Step-by-step derivation
                  1. Applied rewrites76.2%

                    \[\leadsto \mathsf{fma}\left(\left(1 \cdot b\right) \cdot 1, b, {\left(\left(\left(\mathsf{PI}\left(\right) \cdot angle\right) \cdot \color{blue}{0.005555555555555556}\right) \cdot a\right)}^{2}\right) \]
                6. Recombined 2 regimes into one program.
                7. Add Preprocessing

                Alternative 3: 64.7% accurate, 9.1× speedup?

                \[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;a \leq 410000000:\\ \;\;\;\;b \cdot b\\ \mathbf{elif}\;a \leq 6.7 \cdot 10^{+165}:\\ \;\;\;\;\mathsf{fma}\left(\left(\left(3.08641975308642 \cdot 10^{-5} \cdot \left(a \cdot a\right)\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right), angle \cdot angle, b \cdot b\right)\\ \mathbf{else}:\\ \;\;\;\;\left(\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right)\right) \cdot \left(a \cdot angle\right)\\ \end{array} \end{array} \]
                (FPCore (a b angle)
                 :precision binary64
                 (if (<= a 410000000.0)
                   (* b b)
                   (if (<= a 6.7e+165)
                     (fma
                      (* (* (* 3.08641975308642e-5 (* a a)) (PI)) (PI))
                      (* angle angle)
                      (* b b))
                     (* (* (* 3.08641975308642e-5 angle) (* (* (PI) (PI)) a)) (* a angle)))))
                \begin{array}{l}
                
                \\
                \begin{array}{l}
                \mathbf{if}\;a \leq 410000000:\\
                \;\;\;\;b \cdot b\\
                
                \mathbf{elif}\;a \leq 6.7 \cdot 10^{+165}:\\
                \;\;\;\;\mathsf{fma}\left(\left(\left(3.08641975308642 \cdot 10^{-5} \cdot \left(a \cdot a\right)\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right), angle \cdot angle, b \cdot b\right)\\
                
                \mathbf{else}:\\
                \;\;\;\;\left(\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right)\right) \cdot \left(a \cdot angle\right)\\
                
                
                \end{array}
                \end{array}
                
                Derivation
                1. Split input into 3 regimes
                2. if a < 4.1e8

                  1. Initial program 77.0%

                    \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                  2. Add Preprocessing
                  3. Taylor expanded in angle around 0

                    \[\leadsto \color{blue}{{b}^{2}} \]
                  4. Step-by-step derivation
                    1. Applied rewrites65.3%

                      \[\leadsto \color{blue}{b \cdot b} \]

                    if 4.1e8 < a < 6.70000000000000037e165

                    1. Initial program 68.3%

                      \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                    2. Add Preprocessing
                    3. Taylor expanded in angle around 0

                      \[\leadsto \color{blue}{{angle}^{2} \cdot \left(\frac{-1}{32400} \cdot \left({b}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right) + \frac{1}{32400} \cdot \left({a}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right) + {b}^{2}} \]
                    4. Step-by-step derivation
                      1. Applied rewrites32.8%

                        \[\leadsto \color{blue}{\mathsf{fma}\left(\left(-3.08641975308642 \cdot 10^{-5} \cdot \left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(b \cdot b - a \cdot a\right), angle \cdot angle, b \cdot b\right)} \]
                      2. Taylor expanded in a around inf

                        \[\leadsto \mathsf{fma}\left(\frac{1}{32400} \cdot \left({a}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right), \color{blue}{angle} \cdot angle, b \cdot b\right) \]
                      3. Step-by-step derivation
                        1. Applied rewrites57.9%

                          \[\leadsto \mathsf{fma}\left(\left(\left(3.08641975308642 \cdot 10^{-5} \cdot \left(a \cdot a\right)\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \mathsf{PI}\left(\right), \color{blue}{angle} \cdot angle, b \cdot b\right) \]

                        if 6.70000000000000037e165 < a

                        1. Initial program 99.8%

                          \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                        2. Add Preprocessing
                        3. Taylor expanded in angle around 0

                          \[\leadsto \color{blue}{{angle}^{2} \cdot \left(\frac{-1}{32400} \cdot \left({b}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right) + \frac{1}{32400} \cdot \left({a}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right) + {b}^{2}} \]
                        4. Step-by-step derivation
                          1. Applied rewrites42.5%

                            \[\leadsto \color{blue}{\mathsf{fma}\left(\left(-3.08641975308642 \cdot 10^{-5} \cdot \left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(b \cdot b - a \cdot a\right), angle \cdot angle, b \cdot b\right)} \]
                          2. Taylor expanded in a around inf

                            \[\leadsto \frac{1}{32400} \cdot \color{blue}{\left({a}^{2} \cdot \left({angle}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right)} \]
                          3. Step-by-step derivation
                            1. Applied rewrites72.2%

                              \[\leadsto \left(\left(\left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right) \cdot a\right) \cdot angle\right) \cdot \color{blue}{\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)} \]
                            2. Step-by-step derivation
                              1. Applied rewrites75.5%

                                \[\leadsto \left(\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right)\right) \cdot \left(a \cdot \color{blue}{angle}\right) \]
                            3. Recombined 3 regimes into one program.
                            4. Add Preprocessing

                            Alternative 4: 63.4% accurate, 12.1× speedup?

                            \[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;a \leq 2 \cdot 10^{+128}:\\ \;\;\;\;b \cdot b\\ \mathbf{else}:\\ \;\;\;\;\left(\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right)\right) \cdot \left(a \cdot angle\right)\\ \end{array} \end{array} \]
                            (FPCore (a b angle)
                             :precision binary64
                             (if (<= a 2e+128)
                               (* b b)
                               (* (* (* 3.08641975308642e-5 angle) (* (* (PI) (PI)) a)) (* a angle))))
                            \begin{array}{l}
                            
                            \\
                            \begin{array}{l}
                            \mathbf{if}\;a \leq 2 \cdot 10^{+128}:\\
                            \;\;\;\;b \cdot b\\
                            
                            \mathbf{else}:\\
                            \;\;\;\;\left(\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right)\right) \cdot \left(a \cdot angle\right)\\
                            
                            
                            \end{array}
                            \end{array}
                            
                            Derivation
                            1. Split input into 2 regimes
                            2. if a < 2.0000000000000002e128

                              1. Initial program 75.1%

                                \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                              2. Add Preprocessing
                              3. Taylor expanded in angle around 0

                                \[\leadsto \color{blue}{{b}^{2}} \]
                              4. Step-by-step derivation
                                1. Applied rewrites62.4%

                                  \[\leadsto \color{blue}{b \cdot b} \]

                                if 2.0000000000000002e128 < a

                                1. Initial program 93.1%

                                  \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                                2. Add Preprocessing
                                3. Taylor expanded in angle around 0

                                  \[\leadsto \color{blue}{{angle}^{2} \cdot \left(\frac{-1}{32400} \cdot \left({b}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right) + \frac{1}{32400} \cdot \left({a}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right) + {b}^{2}} \]
                                4. Step-by-step derivation
                                  1. Applied rewrites36.4%

                                    \[\leadsto \color{blue}{\mathsf{fma}\left(\left(-3.08641975308642 \cdot 10^{-5} \cdot \left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(b \cdot b - a \cdot a\right), angle \cdot angle, b \cdot b\right)} \]
                                  2. Taylor expanded in a around inf

                                    \[\leadsto \frac{1}{32400} \cdot \color{blue}{\left({a}^{2} \cdot \left({angle}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right)} \]
                                  3. Step-by-step derivation
                                    1. Applied rewrites69.3%

                                      \[\leadsto \left(\left(\left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right) \cdot a\right) \cdot angle\right) \cdot \color{blue}{\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)} \]
                                    2. Step-by-step derivation
                                      1. Applied rewrites68.8%

                                        \[\leadsto \left(\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right)\right) \cdot \left(a \cdot \color{blue}{angle}\right) \]
                                    3. Recombined 2 regimes into one program.
                                    4. Add Preprocessing

                                    Alternative 5: 62.5% accurate, 12.1× speedup?

                                    \[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;a \leq 2 \cdot 10^{+128}:\\ \;\;\;\;b \cdot b\\ \mathbf{else}:\\ \;\;\;\;\left(\mathsf{PI}\left(\right) \cdot \left(\left(a \cdot \mathsf{PI}\left(\right)\right) \cdot \left(a \cdot angle\right)\right)\right) \cdot \left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)\\ \end{array} \end{array} \]
                                    (FPCore (a b angle)
                                     :precision binary64
                                     (if (<= a 2e+128)
                                       (* b b)
                                       (* (* (PI) (* (* a (PI)) (* a angle))) (* 3.08641975308642e-5 angle))))
                                    \begin{array}{l}
                                    
                                    \\
                                    \begin{array}{l}
                                    \mathbf{if}\;a \leq 2 \cdot 10^{+128}:\\
                                    \;\;\;\;b \cdot b\\
                                    
                                    \mathbf{else}:\\
                                    \;\;\;\;\left(\mathsf{PI}\left(\right) \cdot \left(\left(a \cdot \mathsf{PI}\left(\right)\right) \cdot \left(a \cdot angle\right)\right)\right) \cdot \left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)\\
                                    
                                    
                                    \end{array}
                                    \end{array}
                                    
                                    Derivation
                                    1. Split input into 2 regimes
                                    2. if a < 2.0000000000000002e128

                                      1. Initial program 75.1%

                                        \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                                      2. Add Preprocessing
                                      3. Taylor expanded in angle around 0

                                        \[\leadsto \color{blue}{{b}^{2}} \]
                                      4. Step-by-step derivation
                                        1. Applied rewrites62.4%

                                          \[\leadsto \color{blue}{b \cdot b} \]

                                        if 2.0000000000000002e128 < a

                                        1. Initial program 93.1%

                                          \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                                        2. Add Preprocessing
                                        3. Taylor expanded in angle around 0

                                          \[\leadsto \color{blue}{{angle}^{2} \cdot \left(\frac{-1}{32400} \cdot \left({b}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right) + \frac{1}{32400} \cdot \left({a}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right) + {b}^{2}} \]
                                        4. Step-by-step derivation
                                          1. Applied rewrites36.4%

                                            \[\leadsto \color{blue}{\mathsf{fma}\left(\left(-3.08641975308642 \cdot 10^{-5} \cdot \left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(b \cdot b - a \cdot a\right), angle \cdot angle, b \cdot b\right)} \]
                                          2. Taylor expanded in a around inf

                                            \[\leadsto \frac{1}{32400} \cdot \color{blue}{\left({a}^{2} \cdot \left({angle}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right)} \]
                                          3. Step-by-step derivation
                                            1. Applied rewrites69.3%

                                              \[\leadsto \left(\left(\left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right) \cdot a\right) \cdot angle\right) \cdot \color{blue}{\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)} \]
                                            2. Step-by-step derivation
                                              1. Applied rewrites66.3%

                                                \[\leadsto \left(\mathsf{PI}\left(\right) \cdot \left(\left(a \cdot \mathsf{PI}\left(\right)\right) \cdot \left(a \cdot angle\right)\right)\right) \cdot \left(3.08641975308642 \cdot 10^{-5} \cdot angle\right) \]
                                            3. Recombined 2 regimes into one program.
                                            4. Add Preprocessing

                                            Alternative 6: 62.5% accurate, 12.1× speedup?

                                            \[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;a \leq 2 \cdot 10^{+128}:\\ \;\;\;\;b \cdot b\\ \mathbf{else}:\\ \;\;\;\;\left(a \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \left(a \cdot angle\right)\right)\right) \cdot \left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)\\ \end{array} \end{array} \]
                                            (FPCore (a b angle)
                                             :precision binary64
                                             (if (<= a 2e+128)
                                               (* b b)
                                               (* (* a (* (* (PI) (PI)) (* a angle))) (* 3.08641975308642e-5 angle))))
                                            \begin{array}{l}
                                            
                                            \\
                                            \begin{array}{l}
                                            \mathbf{if}\;a \leq 2 \cdot 10^{+128}:\\
                                            \;\;\;\;b \cdot b\\
                                            
                                            \mathbf{else}:\\
                                            \;\;\;\;\left(a \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \left(a \cdot angle\right)\right)\right) \cdot \left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)\\
                                            
                                            
                                            \end{array}
                                            \end{array}
                                            
                                            Derivation
                                            1. Split input into 2 regimes
                                            2. if a < 2.0000000000000002e128

                                              1. Initial program 75.1%

                                                \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                                              2. Add Preprocessing
                                              3. Taylor expanded in angle around 0

                                                \[\leadsto \color{blue}{{b}^{2}} \]
                                              4. Step-by-step derivation
                                                1. Applied rewrites62.4%

                                                  \[\leadsto \color{blue}{b \cdot b} \]

                                                if 2.0000000000000002e128 < a

                                                1. Initial program 93.1%

                                                  \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                                                2. Add Preprocessing
                                                3. Taylor expanded in angle around 0

                                                  \[\leadsto \color{blue}{{angle}^{2} \cdot \left(\frac{-1}{32400} \cdot \left({b}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right) + \frac{1}{32400} \cdot \left({a}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right) + {b}^{2}} \]
                                                4. Step-by-step derivation
                                                  1. Applied rewrites36.4%

                                                    \[\leadsto \color{blue}{\mathsf{fma}\left(\left(-3.08641975308642 \cdot 10^{-5} \cdot \left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(b \cdot b - a \cdot a\right), angle \cdot angle, b \cdot b\right)} \]
                                                  2. Taylor expanded in a around inf

                                                    \[\leadsto \frac{1}{32400} \cdot \color{blue}{\left({a}^{2} \cdot \left({angle}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right)} \]
                                                  3. Step-by-step derivation
                                                    1. Applied rewrites69.3%

                                                      \[\leadsto \left(\left(\left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right) \cdot a\right) \cdot angle\right) \cdot \color{blue}{\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)} \]
                                                    2. Step-by-step derivation
                                                      1. Applied rewrites66.3%

                                                        \[\leadsto \left(a \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \left(a \cdot angle\right)\right)\right) \cdot \left(3.08641975308642 \cdot 10^{-5} \cdot angle\right) \]
                                                    3. Recombined 2 regimes into one program.
                                                    4. Add Preprocessing

                                                    Alternative 7: 61.0% accurate, 12.1× speedup?

                                                    \[\begin{array}{l} \\ \begin{array}{l} \mathbf{if}\;a \leq 1.05 \cdot 10^{+152}:\\ \;\;\;\;b \cdot b\\ \mathbf{else}:\\ \;\;\;\;\left(3.08641975308642 \cdot 10^{-5} \cdot \left(a \cdot a\right)\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \left(angle \cdot angle\right)\right)\\ \end{array} \end{array} \]
                                                    (FPCore (a b angle)
                                                     :precision binary64
                                                     (if (<= a 1.05e+152)
                                                       (* b b)
                                                       (* (* 3.08641975308642e-5 (* a a)) (* (* (PI) (PI)) (* angle angle)))))
                                                    \begin{array}{l}
                                                    
                                                    \\
                                                    \begin{array}{l}
                                                    \mathbf{if}\;a \leq 1.05 \cdot 10^{+152}:\\
                                                    \;\;\;\;b \cdot b\\
                                                    
                                                    \mathbf{else}:\\
                                                    \;\;\;\;\left(3.08641975308642 \cdot 10^{-5} \cdot \left(a \cdot a\right)\right) \cdot \left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \left(angle \cdot angle\right)\right)\\
                                                    
                                                    
                                                    \end{array}
                                                    \end{array}
                                                    
                                                    Derivation
                                                    1. Split input into 2 regimes
                                                    2. if a < 1.0500000000000001e152

                                                      1. Initial program 74.9%

                                                        \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                                                      2. Add Preprocessing
                                                      3. Taylor expanded in angle around 0

                                                        \[\leadsto \color{blue}{{b}^{2}} \]
                                                      4. Step-by-step derivation
                                                        1. Applied rewrites62.0%

                                                          \[\leadsto \color{blue}{b \cdot b} \]

                                                        if 1.0500000000000001e152 < a

                                                        1. Initial program 99.8%

                                                          \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                                                        2. Add Preprocessing
                                                        3. Taylor expanded in angle around 0

                                                          \[\leadsto \color{blue}{{angle}^{2} \cdot \left(\frac{-1}{32400} \cdot \left({b}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right) + \frac{1}{32400} \cdot \left({a}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right) + {b}^{2}} \]
                                                        4. Step-by-step derivation
                                                          1. Applied rewrites42.0%

                                                            \[\leadsto \color{blue}{\mathsf{fma}\left(\left(-3.08641975308642 \cdot 10^{-5} \cdot \left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right)\right) \cdot \left(b \cdot b - a \cdot a\right), angle \cdot angle, b \cdot b\right)} \]
                                                          2. Taylor expanded in a around inf

                                                            \[\leadsto \frac{1}{32400} \cdot \color{blue}{\left({a}^{2} \cdot \left({angle}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right)} \]
                                                          3. Step-by-step derivation
                                                            1. Applied rewrites77.0%

                                                              \[\leadsto \left(\left(\left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot a\right) \cdot a\right) \cdot angle\right) \cdot \color{blue}{\left(3.08641975308642 \cdot 10^{-5} \cdot angle\right)} \]
                                                            2. Taylor expanded in a around inf

                                                              \[\leadsto \frac{1}{32400} \cdot \color{blue}{\left({a}^{2} \cdot \left({angle}^{2} \cdot {\mathsf{PI}\left(\right)}^{2}\right)\right)} \]
                                                            3. Step-by-step derivation
                                                              1. Applied rewrites59.3%

                                                                \[\leadsto \left(3.08641975308642 \cdot 10^{-5} \cdot \left(a \cdot a\right)\right) \cdot \color{blue}{\left(\left(\mathsf{PI}\left(\right) \cdot \mathsf{PI}\left(\right)\right) \cdot \left(angle \cdot angle\right)\right)} \]
                                                            4. Recombined 2 regimes into one program.
                                                            5. Add Preprocessing

                                                            Alternative 8: 57.5% accurate, 74.7× speedup?

                                                            \[\begin{array}{l} \\ b \cdot b \end{array} \]
                                                            (FPCore (a b angle) :precision binary64 (* b b))
                                                            double code(double a, double b, double angle) {
                                                            	return b * b;
                                                            }
                                                            
                                                            module fmin_fmax_functions
                                                                implicit none
                                                                private
                                                                public fmax
                                                                public fmin
                                                            
                                                                interface fmax
                                                                    module procedure fmax88
                                                                    module procedure fmax44
                                                                    module procedure fmax84
                                                                    module procedure fmax48
                                                                end interface
                                                                interface fmin
                                                                    module procedure fmin88
                                                                    module procedure fmin44
                                                                    module procedure fmin84
                                                                    module procedure fmin48
                                                                end interface
                                                            contains
                                                                real(8) function fmax88(x, y) result (res)
                                                                    real(8), intent (in) :: x
                                                                    real(8), intent (in) :: y
                                                                    res = merge(y, merge(x, max(x, y), y /= y), x /= x)
                                                                end function
                                                                real(4) function fmax44(x, y) result (res)
                                                                    real(4), intent (in) :: x
                                                                    real(4), intent (in) :: y
                                                                    res = merge(y, merge(x, max(x, y), y /= y), x /= x)
                                                                end function
                                                                real(8) function fmax84(x, y) result(res)
                                                                    real(8), intent (in) :: x
                                                                    real(4), intent (in) :: y
                                                                    res = merge(dble(y), merge(x, max(x, dble(y)), y /= y), x /= x)
                                                                end function
                                                                real(8) function fmax48(x, y) result(res)
                                                                    real(4), intent (in) :: x
                                                                    real(8), intent (in) :: y
                                                                    res = merge(y, merge(dble(x), max(dble(x), y), y /= y), x /= x)
                                                                end function
                                                                real(8) function fmin88(x, y) result (res)
                                                                    real(8), intent (in) :: x
                                                                    real(8), intent (in) :: y
                                                                    res = merge(y, merge(x, min(x, y), y /= y), x /= x)
                                                                end function
                                                                real(4) function fmin44(x, y) result (res)
                                                                    real(4), intent (in) :: x
                                                                    real(4), intent (in) :: y
                                                                    res = merge(y, merge(x, min(x, y), y /= y), x /= x)
                                                                end function
                                                                real(8) function fmin84(x, y) result(res)
                                                                    real(8), intent (in) :: x
                                                                    real(4), intent (in) :: y
                                                                    res = merge(dble(y), merge(x, min(x, dble(y)), y /= y), x /= x)
                                                                end function
                                                                real(8) function fmin48(x, y) result(res)
                                                                    real(4), intent (in) :: x
                                                                    real(8), intent (in) :: y
                                                                    res = merge(y, merge(dble(x), min(dble(x), y), y /= y), x /= x)
                                                                end function
                                                            end module
                                                            
                                                            real(8) function code(a, b, angle)
                                                            use fmin_fmax_functions
                                                                real(8), intent (in) :: a
                                                                real(8), intent (in) :: b
                                                                real(8), intent (in) :: angle
                                                                code = b * b
                                                            end function
                                                            
                                                            public static double code(double a, double b, double angle) {
                                                            	return b * b;
                                                            }
                                                            
                                                            def code(a, b, angle):
                                                            	return b * b
                                                            
                                                            function code(a, b, angle)
                                                            	return Float64(b * b)
                                                            end
                                                            
                                                            function tmp = code(a, b, angle)
                                                            	tmp = b * b;
                                                            end
                                                            
                                                            code[a_, b_, angle_] := N[(b * b), $MachinePrecision]
                                                            
                                                            \begin{array}{l}
                                                            
                                                            \\
                                                            b \cdot b
                                                            \end{array}
                                                            
                                                            Derivation
                                                            1. Initial program 77.7%

                                                              \[{\left(a \cdot \sin \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} + {\left(b \cdot \cos \left(\frac{angle}{180} \cdot \mathsf{PI}\left(\right)\right)\right)}^{2} \]
                                                            2. Add Preprocessing
                                                            3. Taylor expanded in angle around 0

                                                              \[\leadsto \color{blue}{{b}^{2}} \]
                                                            4. Step-by-step derivation
                                                              1. Applied rewrites59.5%

                                                                \[\leadsto \color{blue}{b \cdot b} \]
                                                              2. Add Preprocessing

                                                              Reproduce

                                                              ?
                                                              herbie shell --seed 2025019 
                                                              (FPCore (a b angle)
                                                                :name "ab-angle->ABCF A"
                                                                :precision binary64
                                                                (+ (pow (* a (sin (* (/ angle 180.0) (PI)))) 2.0) (pow (* b (cos (* (/ angle 180.0) (PI)))) 2.0)))