Average Error: 0.2 → 0.2
Time: 1.3s
Precision: binary64
\[\frac{A}{\mathsf{max}\left(sigma, \frac{1}{\sqrt{N}}\right)}\]
↓
\[\frac{A}{\mathsf{max}\left(sigma, \frac{1}{\sqrt{N}}\right)}\]
\frac{A}{\mathsf{max}\left(sigma, \frac{1}{\sqrt{N}}\right)}
↓
\frac{A}{\mathsf{max}\left(sigma, \frac{1}{\sqrt{N}}\right)}
double code(double A, double sigma, double N) {
	return ((double) (A / ((double) fmax(sigma, ((double) (1.0 / ((double) sqrt(N))))))));
}
↓
double code(double A, double sigma, double N) {
	return ((double) (A / ((double) fmax(sigma, ((double) (1.0 / ((double) sqrt(N))))))));
}

Error

Bits error versus A

Bits error versus sigma

Bits error versus N

Try it out

Your Program's Arguments

Results

Enter valid numbers for all inputs

Derivation

  1. Initial program 0.2

    \[\frac{A}{\mathsf{max}\left(sigma, \frac{1}{\sqrt{N}}\right)}\]

  2. Final simplification0.2

    \[\leadsto \frac{A}{\mathsf{max}\left(sigma, \frac{1}{\sqrt{N}}\right)}\]

Reproduce

herbie shell --seed 2020152 
(FPCore (A sigma N)
  :name "(/ A (fmax sigma (/ 1 (sqrt N))))"
  :precision binary64
  (/ A (fmax sigma (/ 1.0 (sqrt N)))))