kibaekkim / dspopt.jl Goto Github PK

Hello,

I am having an issue using different methods with DSPopt. I obtain the correct solution with the following code:

status = optimize!(model,
        is_stochastic = false, # Needs to indicate that the model is NOT a stochastic program.
        solve_type = DSPopt.ExtensiveForm, # see instances(DSPopt.Methods) for other methods
    )

However, when I use solve_type = DSPopt.DW, I do not obtain the correct optimal solution.

Any help would be greatly appreciated.

Binary variables do not have lower and upper bounds in MOI. Needs to explicitly pass to DSP the lower and upper bounds for binary variables.

This issue is used to trigger TagBot; feel free to unsubscribe.

If you haven't already, you should update your TagBot.yml to include issue comment triggers.
Please see this post on Discourse for instructions and more details.

This function is available in DSP, but not implemented in this interface. The function can still be called by

DSPopt.@dsp_ccall("writeMps", Cvoid, (Ptr{Cvoid}, Ptr{UInt8}), DSPopt.dspenv.p, "myfile")

Hi,

I'm sorry for opening another issue, but I'm not able to understand why there is a segmentation fault when launching some examples. I tried to ran different examples always obtaining the following message (in this particular case I am running this example https://github.com/kibaekkim/DSPopt.jl/blob/0812121ea8031c3f67a8cf93a8f59ecdfe0ba897/examples/farmer_mpi_stoc.jl):

Am I doing something wrong?

Thank you for your help

Consider to have a feature to pass the ambiguity set information to DSP: https://github.com/Argonne-National-Laboratory/DSP/blob/master/examples/dro/DRDCAP.jl

Hi,
I'm trying to solve a large-scale model with DSP, and have tried the stochastic and non-stochastic formulations by following the farmer examples included. Also, I have successfully solved the problem with straight JuMP + Gurobi by modeling it manually in its explicit form, so I know it is feasible in all constraints. Nevertheless, when trying to solve it with DSPopt, ALPS always throws the "No primal ray is available" message, in stochastic and non-stochastic formulation, Dual or Extensive forms. Is it something which may be documented in the past? I could eventually share lps files of the model.

Kind regards!

Need to implement quadratic objective function

Hello,
Sorry for opening an issue during Christmas and New Year, but when I use DSPopt.jl, I failed to pass some of the tests and occurred the segmentation fault (received signal: 11).

I installed DSP and DSPopt in two places:

1. When I installed DSP and DSPopt on a 2020 Intel Macbook Pro with the solver of SCIP.
   When I ran "] test DSPopt", it couldn't execute the optimization procedure and got the segmentation fault (received signal: 11)

I also ran "ctest" for DSP and it didn't pass two tests on problems 6 and 24, I don't know whether it related to the failure of the DSPopt test.

2. I also install DSP and DSP on a Linux HPC server (ComputeCanada) with the solver CPLEX.
   Here when I ran "] test DSPopt", the test looks fine but it still pop out the segmentation fault (received signal: 11) at the end of the process.
   

Here "ctest" on DSP has no problem

I will be glab to provide more details if needed.

Any help would be appreciated.

The QCQP interface is available for stochastic programming only. We need to implement one of the following:

• QCQP interface for general structured models
• the display of warning message when user tries to solve generic structured QCQP models.

While we can run DSP in parallel, the scenario subproblems are not created in parallel.