Automatic Differentiation
 
Loading...
Searching...
No Matches

◆ finite_diff_stepsize()

double stan::math::finite_diff_stepsize ( double  u)
inline

Return the stepsize for finite difference evaluations at the specified scalar.

The formula used is stepsize(u) = cbrt(epsilon) * max(1, abs(u)).

Parameters
uinitial value to increment
Returns
stepsize away from u for finite differences

Definition at line 21 of file finite_diff_stepsize.hpp.