sschnug / pymiqp Goto Github PK

View Code? Open in Web Editor NEW

13.0 13.0 3.0 31 KB

Mixed Integer Quadratic Programming for Python (using MINLP-solver Bonmin)

Python 16.74% C++ 83.26%

pymiqp's People

Contributors

Stargazers

Watchers

Forkers

dedekinds epoxx-arch lalajiayou

pymiqp's Issues

Bonmin-internals: binary-bounds

Commit a755df9 introduced explicit bound-information for binary-variables as those were missing in Bonmin's relaxation-process.

As Stefan Vigerske said:

In your implementation of TMINLP::get_bounds_info you should specify
bounds [0,1] for (unfixed) binary variables, though it may seem redundant.

i should check the docs / internals if there is any chance, i'm overwriting bounds for already fixed variables (which would be very bad!).

Edit

Check what Bonmin provides as arguments: if there are already valid (more restricting) bounds set, don't touch them!

Refactor / Design: options-passing

Currently only some options are supported and each of those has it's own specific class-variables.

Refactor this to make it more generic!

tests/test_1.py: very much variation in sol dep. on algorithm

B-BB:
sol-status:  0
sol-obj:  -5510222.415947652
sol-x:  [33.  5. 75.  0. 83.  0. 80. 61. 22.  0.]

B-Hyb
sol-status:  0
sol-obj:  -2664503.314829478
sol-x:  [209.   0.   0.   0. 184.   0.   0.   0.   0.   0.]

B-OA
sol-status:  0
sol-obj:  -5510196.4890296
sol-x:  [34.  5. 75.  0. 82.  1. 80. 61. 22.  0.]

B-QG
sol-status:  0
sol-obj:  -2664503.314829478
sol-x:  [209.   0.   0.   0. 184.   0.   0.   0.   0.   0.]

Scaling-problems or something else?

Recommend Projects