Franz Wesselmann comments

Results 33 comments of


                                            Franz Wesselmann

Wrongly claimed infeasible model

This seems to be fixed by #1730 as well.

Potential issues with presolve: incorrect 'optimal' solution and incorrect infeasibiliy detection

@jajhall I am having a look, if that's Ok

Assertion `model->col_upper_[col] == kHighsInf || (model->col_upper_[col] >= implColUpper[col] - primal_feastol && colUpperSource[col] == row)' failed.

Interesting, could you provide the MPS file?

Assertion `model->col_upper_[col] == kHighsInf || (model->col_upper_[col] >= implColUpper[col] - primal_feastol && colUpperSource[col] == row)' failed.

Unfortunately, I am not able to reproduce this yet. I suspect that its due to the fact that the problematic assertions do not take into account that the implied column...

Model status goes from infeasible to optimal when shutting down presolve

I think reduction number 5144 is causing this. The model is solved to optimality with reduction limit 5143: `highs.setOptionValue("presolve_reduction_limit", 5143);` But it is infeasible when setting it to 5144: `highs.setOptionValue("presolve_reduction_limit",...

instance found to be infeasible but it isn't if run without presolve or run with cbc

Some of the equality constraints in the model are poorly scaled. I think this may lead the presolve code to fix variables to incorrect values. For example, equality constraint 'r726'...

Crash with unbounded columns

Hello! My understanding of the code is very limited, but it looks like the problematic assertion will always fail if there are free variables in the objective function. It looks...

Potential issues with presolve: incorrect 'optimal' solution and incorrect infeasibiliy detection

> Thanks. There are bugs in the MIP presolve that we are steadily fixing. This can be added to the instances @fwesselm I just tried the `latest` branch, and this...

Potential issues with presolve: incorrect 'optimal' solution and incorrect infeasibiliy detection

> This is great news @fwesselm > > This might be due to differences in compute environment but one thing that I want to flag is that the total solver...

Incorrectly detected infeasibility with pre-solve turned off

I can reproduce this using the `latest` branch. However, the problem is solved to optimality when I reduce the `mip_feasibility_tolerance`: `highs.setOptionValue("mip_feasibility_tolerance", 1e-09);`