LLNL / hiop

HPC solver for nonlinear optimization problems
Other
210 stars 42 forks source link

Fix the approach used to update mu #664

Closed nychiang closed 1 year ago

nychiang commented 1 year ago

When the given convergence tolerance is large, the existing implementation will prevent us further reducing mu and consequently the algorithm cannot converge.

This PR aims to fix this issue and updates the solutions to the unit tests.

@cameronrutherford @abhyshr This PR fixes the opflow issue.