Stop Julia at certain optimiality gap and return decision variables

[Duaa]

I have a MILP problem and I am trying to solve it using CPLEX package through JuMP in Julia Language. However, the problem takes a long time to be solved. Is there a way to stop the program and return decision variables when the optimality gap reaches for example 10%. Thank you.

[rdeits]

I haven’t used CPLEX, but I believe you can do:

m = Model(solver=CplexSolver(CPX_PARAM_EPGAP=1e-2))

or

setsolver(m, CplexSolver(CPX_PARAM_EPGAP=1e-2))

[rdeits]

(the above assumes you are using CPLEX through JuMP, which I just realized you didn’t actually say).

[Duaa]

Thank you, it works.

[Duaa]

Yes, I am using CPLEX through JuMP.
Sorry for not mentioning it.

[Rajat_Sanyal]

Hi @rdeits,

I am using Julia 1.1.0 and following packages:

“Clp” => v"0.6.1"
“Atom” => v"0.7.14"
“CPLEX” => v"0.4.3"
“JuMP” => v"0.19.0"
“Juno” => v"0.5.4"
“Conda” => v"1.2.0"
“ExcelReaders” => v"0.10.3"
“MathProgBase” => v"0.7.7"
“XLSX” => v"0.4.6"
“Cbc” => v"0.6.0"
“DataFrames” => v"0.17.1".

I am trying to set m = Model(solver=CplexSolver(CPX_PARAM_EPGAP=1e-2)), but I am getting the following error:

The solver= keyword is no longer available in JuMP 0.19 and later.

However, they are suggesting a link for the reference which is somehow not a valid link.

Any help in this regard would be highly appreciated. Also, could you please share any link where I can find how to set different stopping criteria for the CPLEX solver through JuMP?

[odow]

JuMP has been updated.

JuMP v0.19 has been released

We are thrilled to announce the release of JuMP v0.19. Work on this release started nearly two years ago at the first JuMP-dev meetup. As introduced in an earlier post and at the second JuMP-dev meetup, JuMP has undergone a major refactoring to replace its solver interface from MathProgBase (MPB) to the new MathOptInterface (MOI), receiving many new features and improvements along the way. While MPB had three different interfaces: linear/quadratic, conic and nonlinear, MOI is completely generi…

The new syntax is

m = Model(with_optimizer(CPLEX.Optimizer, CPX_PARAM_EPGAP=1e-2))

[Rajat_Sanyal]

Thanks, @odow . Any idea where can I find the list of keywords to set the stopping criteria, example: absolute MIP gap, relative MIP gap, time etc.? Also, is there a way to set a warm start?

[odow]

The list of CPLEX parameters can be found here: IBM Documentation

Warmstarts can set set using
https://github.com/JuliaOpt/JuMP.jl/blob/fbc5c80763a17067199e95b4e2e1554823fadbce/src/variables.jl#L692-L712

These were mistakenly missing from the docs: start_value and set_start_value are not listed in the documentation · Issue #1871 · jump-dev/JuMP.jl · GitHub

[NiclasMattsson]

You’ll soon discover that it’s extremely clunky to navigate the parameter list using the web interface. Luckily IBM also makes a nice easily searchable PDF of the whole thing, but to my knowledge they don’t actually link to it anywhere. Some time ago I googled my way to an old version and then found the latest one by just modifying the URL:

paramcplex.pdf

By the way, I think the old parameter names are being deprecated. So if you want to future proof your code you may want to use the newer more verbose names, e.g. CPXPARAM_MIP_Tolerances_MIPGap instead of CPX_PARAM_EPGAP.

[lausilvag]

Hi!
I have the same problem. I am using Julia 1.2.0 and the following packages:
CPLEX=V.12.9
JuMP =V. 0.20.1

I am trying with this code:
using JuMP, CPLEX
m= Model(with_optimizer(CPLEX.Optimizer, CPXPARAM_MIP_Strategy_VariableSelect=1))

and this is the error: " no method matching CPLEX.Optimizer (;CPXPARAM_MIP_Strategy_VariableSelect=1)"

Please, can you help me?
Thanks!

[odow]

Use:

model = Model(with_optimizer(CPLEX.Optimizer))
MOI.set(model, MOI.RawParameter("CPXPARAM_MIP_Strategy_VariableSelect"), 1)

We should improve the user-experience of this.

We already have:

set_parameter(model, "CPXPARAM_MIP_Strategy_VariableSelect", 1)

[lausilvag]

Thank you so much!
it works perfectly!