The tolerance for computed continued fractions, cf_tol, should be related to the desired tolerance on omega, tol, by the gradient of the "objective function" (continued fraction equation). We can estimate it with a finite difference derivative.
If cf_tol ends up being smaller than DBL_EPSILON, we probably won't be able to polish omega to the desired tolerance.
The tolerance for computed continued fractions,
cf_tol
, should be related to the desired tolerance on omega,tol
, by the gradient of the "objective function" (continued fraction equation). We can estimate it with a finite difference derivative.If cf_tol ends up being smaller than DBL_EPSILON, we probably won't be able to polish omega to the desired tolerance.