Expression, p6

Average Error: 3.7 → 2.8

Time: 3.0s

Precision: 64

\[-14 \le a \le -13 \land -3 \le b \le -2 \land 3 \le c \le 3.5 \land 12.5 \le d \le 13.5\]

Math

\[\left(a + \left(b + \left(c + d\right)\right)\right) \cdot 2\]

↓

\[\left(\left(a + \left(b + c\right)\right) + d\right) \cdot 2\]

C

double f(double a, double b, double c, double d) {
        double r96296 = a;
        double r96297 = b;
        double r96298 = c;
        double r96299 = d;
        double r96300 = r96298 + r96299;
        double r96301 = r96297 + r96300;
        double r96302 = r96296 + r96301;
        double r96303 = 2.0;
        double r96304 = r96302 * r96303;
        return r96304;
}

↓

double f(double a, double b, double c, double d) {
        double r96305 = a;
        double r96306 = b;
        double r96307 = c;
        double r96308 = r96306 + r96307;
        double r96309 = r96305 + r96308;
        double r96310 = d;
        double r96311 = r96309 + r96310;
        double r96312 = 2.0;
        double r96313 = r96311 * r96312;
        return r96313;
}

Error

Bits error versus a

Bits error versus b

Bits error versus c

Bits error versus d

Target

Original	3.7
Target	3.9
Herbie	2.8

\[\left(a + b\right) \cdot 2 + \left(c + d\right) \cdot 2\]

Derivation

1. Initial program 3.7

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

3. Applied associate-+r+ 2.8

   \[\leadsto \left(a + \color{blue}{\left(\left(b + c\right) + d\right)}\right) \cdot 2\]
4. Using strategy rm

5. Applied associate-+r+ 2.8

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

6. Final simplification 2.8

   \[\leadsto \left(\left(a + \left(b + c\right)\right) + d\right) \cdot 2\]

Reproduce

herbie shell --seed 2020100 +o rules:numerics
(FPCore (a b c d)
  :name "Expression, p6"
  :precision binary64
  :pre (and (<= -14 a -13) (<= -3 b -2) (<= 3 c 3.5) (<= 12.5 d 13.5))

  :herbie-target
  (+ (* (+ a b) 2) (* (+ c d) 2))

  (* (+ a (+ b (+ c d))) 2))