[SCIP] possible memory leak in SCIP with JuMP Interface

Felipe Serrano fserranom5 at gmail.com
Fri Apr 21 04:56:03 CEST 2017


Dear Yankai

We already investigated a possible memory leak. Please check
https://github.com/SCIP-Interfaces/SCIP.jl/issues/33

Please reopen the issue or create a new one there with this problem.

Best,
Felipe

On Thu, Apr 20, 2017 at 11:22 PM, Yankai Cao <yankai.cao at gmail.com> wrote:

> Hi,
>
> I am using scipoptsuite-4.0.0 with julia interface SCIP.jl. I  find that
> if I create and solve models in a for loop using SCIP with models built
> with JuMP, then the peak memory usage increases dramatically as the number
> of iterations increases. I would expect the peak memory usage keeps almost
> constant.
>
> Here is the test code test.jl:
>
> using JuMP, Gurobi, SCIP
>
> function test(iter)
> for i = 1:iter
>      m= Model()
>      nvar = 1000
>      @variable(m, 0<=x[1:nvar]<=1)
>      @constraint(m, con[i in 1:(nvar-2)], x[i]+x[i+1]<=1)
>      @objective(m, :Min, sum{(x[i]+x[i+2])^2, i in 1:(nvar-2)})
>      m.solver=SCIPSolver("display/verblevel", 0)
>      #m.solver=GurobiSolver(LogToConsole=0,OutputFlag=0)
>      solve(m)
> end
> end
> test(10)
>
> I monitored the peak memory usage using /usr/bin/time -v julia test.jl
>
> For SCIP, I got
> test(1)
> Maximum resident set size (kbytes): 234312
> test(10)
> Maximum resident set size (kbytes): 803168
> test(100)
> Maximum resident set size (kbytes): 6441800
>
> For Gurobi, I got
> test(1)
> Maximum resident set size (kbytes): 174972
> test(10)
> Maximum resident set size (kbytes): 187388
> test(100)
> Maximum resident set size (kbytes): 263780
> So for Gurobi, the result is as expected, but for SCIP the peak memory
> usage do increases dramatically. Can anyone check this issue? Thanks.
>
>
>
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20170420/6bf46e25/attachment.html>


More information about the Scip mailing list