Average Error: 58.7 → 0.5
Time: 14.8s
Precision: 64
\[-1.700000000000000122124532708767219446599 \cdot 10^{-4} \lt x\]
\[e^{x} - 1\]
↓
\[\mathsf{fma}\left(\mathsf{fma}\left(x, \frac{1}{6}, \frac{1}{2}\right) \cdot x, x, x\right)\]
e^{x} - 1
↓
\mathsf{fma}\left(\mathsf{fma}\left(x, \frac{1}{6}, \frac{1}{2}\right) \cdot x, x, x\right)
double f(double x) {
        double r4687508 = x;
        double r4687509 = exp(r4687508);
        double r4687510 = 1.0;
        double r4687511 = r4687509 - r4687510;
        return r4687511;
}
↓
double f(double x) {
        double r4687512 = x;
        double r4687513 = 0.16666666666666666;
        double r4687514 = 0.5;
        double r4687515 = fma(r4687512, r4687513, r4687514);
        double r4687516 = r4687515 * r4687512;
        double r4687517 = fma(r4687516, r4687512, r4687512);
        return r4687517;
}

Error

Bits error versus x

Target

Original58.7
Target0.5
Herbie0.5
\[x \cdot \left(\left(1 + \frac{x}{2}\right) + \frac{x \cdot x}{6}\right)\]

Derivation

  1. Initial program 58.7

    \[e^{x} - 1\]

  2. Taylor expanded around 0 0.5

    \[\leadsto \color{blue}{x + \left(\frac{1}{6} \cdot {x}^{3} + \frac{1}{2} \cdot {x}^{2}\right)}\]

  3. Simplified0.5

    \[\leadsto \color{blue}{\mathsf{fma}\left(x \cdot \mathsf{fma}\left(x, \frac{1}{6}, \frac{1}{2}\right), x, x\right)}\]

  4. Final simplification0.5

    \[\leadsto \mathsf{fma}\left(\mathsf{fma}\left(x, \frac{1}{6}, \frac{1}{2}\right) \cdot x, x, x\right)\]

Reproduce

herbie shell --seed 2019170 +o rules:numerics
(FPCore (x)
  :name "expm1 (example 3.7)"
  :pre (< -0.00017 x)

  :herbie-target
  (* x (+ (+ 1.0 (/ x 2.0)) (/ (* x x) 6.0)))

  (- (exp x) 1.0))