Charlie Vanaret
Charlie Vanaret
Add an interface to [HiGHS](https://ergo-code.github.io/HiGHS/stable/) as a QP solver for positive semi-definite Hessians. We'll then be able to run Uno's SQP method with convexification. It was interfaced as LP solver...
See https://github.com/cvanaret/Uno/issues/432#issuecomment-3544285116 The strategy I'm going for: Each subproblem solver should have two copies of its `EvaluationSpace`: one at the current iterate and one at the trial iterate. In the...
- [x] added `Scaling` class - [x] added `EvaluationSpace::compute_constraint_jacobian_norms()` to compute the inf norm of each Jacobian row - [x] created an optional scaling object in the constraint relaxation strategies...
The Python bindings `unopy` are now merged into main. This PR is dedicated to the packaging of `unopy`. At the moment, `setup.py` fetches artifacts of external dependencies, compiles and installs...
Added a function `uno_get_solver_method()` in C/Fortran/Python interfaces for the user to check which method is used by the solver, based on the strategy combination. This is especially useful because some...
Function scaling a la IPOPT (section 3.8 of the implementation paper).
Uno generates a description of the strategy combination like: `TR Fletcher-filter restoration inequality-constrained method with exact Hessian and no regularization` or `LS Waechter-filter restoration primal-dual interior-point method with exact Hessian...
https://github.com/cvanaret/Uno/pull/414/ suggests that feasibility restoration may be useful when the line search fails (step length too small). Leaving it as is for now.