Expression 4, p15

Average Error: 0.0 → 0

Time: 12.1s

Precision: 64

\[5 \le a \le 10 \land 0.0 \le b \le 0.001000000000000000020816681711721685132943\]

Math

\[\left(a + b\right) \cdot \left(a + b\right)\]

↓

\[\mathsf{fma}\left(a, a, b \cdot \mathsf{fma}\left(a, 2, b\right)\right)\]

C

double f(double a, double b) {
        double r50581 = a;
        double r50582 = b;
        double r50583 = r50581 + r50582;
        double r50584 = r50583 * r50583;
        return r50584;
}

↓

double f(double a, double b) {
        double r50585 = a;
        double r50586 = b;
        double r50587 = 2.0;
        double r50588 = fma(r50585, r50587, r50586);
        double r50589 = r50586 * r50588;
        double r50590 = fma(r50585, r50585, r50589);
        return r50590;
}

Error

Bits error versus a

Bits error versus b

Target

Original	0.0
Target	0.0
Herbie	0

\[\left(\left(b \cdot a + b \cdot b\right) + b \cdot a\right) + a \cdot a\]

Derivation

1. Initial program 0.0

   \[\left(a + b\right) \cdot \left(a + b\right)\]
2. Using strategy rm

3. Applied distribute-lft-in 0.0

   \[\leadsto \color{blue}{\left(a + b\right) \cdot a + \left(a + b\right) \cdot b}\]

4. Taylor expanded around 0 0.0

   \[\leadsto \color{blue}{{a}^{2} + \left(2 \cdot \left(a \cdot b\right) + {b}^{2}\right)}\]

5. Simplified 0

   \[\leadsto \color{blue}{\mathsf{fma}\left(a, a, b \cdot \mathsf{fma}\left(a, 2, b\right)\right)}\]

6. Final simplification 0

   \[\leadsto \mathsf{fma}\left(a, a, b \cdot \mathsf{fma}\left(a, 2, b\right)\right)\]

Reproduce

herbie shell --seed 2019323 +o rules:numerics
(FPCore (a b)
  :name "Expression 4, p15"
  :precision binary64
  :pre (and (<= 5 a 10) (<= 0.0 b 0.001))

  :herbie-target
  (+ (+ (+ (* b a) (* b b)) (* b a)) (* a a))

  (* (+ a b) (+ a b)))