Infeasibility on SCIP - Problem feasible in other solvers

Specific Domains Optimization (Mathematical)
,
1

Dear Team,

I have been trying SCIP but whatever I do it always returns an infeasibility.

Other solvers I have tried do deliver a solution.

The message I keep getting on an on for different problems is:

violation: right hand side is violated by 0.00533581525087357
best solution is not feasible in original problem

What would you recommend?

2

This web may provide a valuable hint:

https://stackoverflow.com/questions/22452332/obtain-best-feasible-solution-with-scip

However, the SCIP.jl wrapper does not indicate how to enter “set emphasis feasibility”. Any suggestion?

3

Take a read of the first post in Please read: make it easier to help you - #11.

It’s easier to help if you provide a minimal piece of code that demonstrates the problem.

2 Likes
4

Yeah, so SCIP.jl only supports setting parameters. To set an emphasis (which then sets multiple parameters), you would have to call the underlying C function. That is, it’s not straight-forward to do currently.

5

Let me then share the code I am using:

using JuMP # used for mathematical programming
using MathOptFormat
using MathOptInterface
using SCIP

mps_model = MathOptFormat.MPS.Model()
MOI.read_from_file(mps_model,string(path,"test_1.mps"))

ucf = Model(with_optimizer(SCIP.Optimizer))
MOI.copy_to(JuMP.backend(ucf),mps_model)
@time optimize!(ucf)

I have placed a set of three MPS files in WeTransfer to be able to reproduce the problem:

https://we.tl/t-nHrmMNhben

Thank you

6

Any other idea to avoid SCIP to find the problem infeasible? It runs well on other solvers.

7

It might be best to ask on the SCIP mailing list.
But from the message that you quote in your original post, it looks like SCIP actually finds a solution, but then discards it because the violation is beyond the tolerances.

So playing with feastol might help here.

8

Already did play with feastol, unluckily it was infeasible always

Will reach out to SCIP mailing list

Thanks