Closed hsloot closed 3 years ago
For most of the problematic integrals, problems will probably only be present if difference_order == 1
. For this case, we can use the first-order differences directly without problems. Nevertheless, integration errors should still be dealt with.
For some Bernstein Functions, using an alternative integral (e.g. with the Stieltjes density) might be an option.
Description
Various methods make use of
stats::integrate
, but do not check whether the integration was successful or handles the error if that is not the case.