Average Error: 41.6 → 41.6
Time: 3.2s
Precision: binary64
\[e^{\frac{x}{2}} - \frac{x}{e^{x} - 1}\]
↓
\[e^{\frac{x}{2}} - \frac{x}{e^{x} - 1}\]
e^{\frac{x}{2}} - \frac{x}{e^{x} - 1}
↓
e^{\frac{x}{2}} - \frac{x}{e^{x} - 1}
double code(double x) {
	return ((double) (((double) exp(((double) (x / 2.0)))) - ((double) (x / ((double) (((double) exp(x)) - 1.0))))));
}
↓
double code(double x) {
	return ((double) (((double) exp(((double) (x / 2.0)))) - ((double) (x / ((double) (((double) exp(x)) - 1.0))))));
}

Error

Bits error versus x

Try it out

Your Program's Arguments

Results

Enter valid numbers for all inputs

Derivation

  1. Initial program 41.6

    \[e^{\frac{x}{2}} - \frac{x}{e^{x} - 1}\]

  2. Final simplification41.6

    \[\leadsto e^{\frac{x}{2}} - \frac{x}{e^{x} - 1}\]

Reproduce

herbie shell --seed 2020153 
(FPCore (x)
  :name "(- (exp (/ x 2)) (/ x (- (exp x) 1)))"
  :precision binary64
  (- (exp (/ x 2.0)) (/ x (- (exp x) 1.0))))