This adds back the verify_solution_completeness routine which was already in HC v1. This should be the last missing functionality from v1.
On the way I found that we had no function to get the certificates corresponding to distinct solutions. So I added distinct_certificates.
I also observed that our epsilon inflation procedure could fail if a coordinate of the true solution is 0 since we considered only a relative value for the inflation. I improved the situation by requiring that we inflate at least by eps(prec) * prec^(-1/4).
This adds back the
verify_solution_completeness
routine which was already in HC v1. This should be the last missing functionality from v1.On the way I found that we had no function to get the certificates corresponding to distinct solutions. So I added
distinct_certificates
.I also observed that our epsilon inflation procedure could fail if a coordinate of the true solution is 0 since we considered only a relative value for the inflation. I improved the situation by requiring that we inflate at least by
eps(prec) * prec^(-1/4)
.