Re: [Help-glpk] [Fwd: Re: Future improvements for GLPK]

[Noli Sicad]

Hi,

I have good result of using RINS in CBC solver. It was able to solve my MIP
forestry model with 5,000 stands / polygons x 10-year planning period (i.e.
more than 50,000 columns / variable in 4.5 hours.

This more than 50,000 columns MIP model could not be solved with
proximity search and Fpump in GLPK solver at the moment. Cuts
algorithms could not solve this problem as well.

I only managed to solve more than 20,0000 variables in GLPK Solver
using proximity search and it took more than 3 hours to solve it.

Hope to see RINS incorporated in GLPK solver.

Thanks.

Noli

On 3/2/16, Andrew Makhorin <address@hidden> wrote:
> -------- Forwarded Message --------
> From: Giorgio Sartor <address@hidden>
> To: Haroldo Gambini Santos <address@hidden>
> Cc: Renan Silva <address@hidden>, address@hidden
> Subject: Re: [Help-glpk] Future improvements for GLPK
> Date: Wed, 2 Mar 2016 08:36:18 +0800
>
> Hi,
>
>
> I actually sent the code for RINS to Andrew a while ago after writing
> Proximity Search (aka proxy). My guess is that including RINS in a
> stable release is a bit tricky since it has to work directly inside the
> b&c. However, the results in some preliminary tests were quite good as
> you may expect.
>
> Regards,
>
>
> Giorgio
>
>
>
> On 2 Mar 2016, at 1:53 AM, Haroldo Gambini Santos
> <address@hidden> wrote:
>
>
>> Hi Renan,
>>
>> If you want to work with cutting planes or heuristics, I would
>> suggest:
>>
>> cuts:
>> - zero half cuts:  these cuts provided very good results in other
>> solvers, are not implemented yet in GLPK
>>
>> heuristics:
>> - RINS is quite simple and AFAIK is not implemented in GLPK yet
>>
>> Cheers
>>
>>
>> On 01-03-2016 14:46, Renan Silva wrote:
>>
>> > Hello,
>> >
>> >
>> > I am a computer science undergraduate student and currently I am
>> > researching on Integer Linear Programming and I am using GLPK as my
>> > primary solver.
>> >
>> > As a future project I want to make some contribution to GLPK and I
>> > would like someone to point some (easy for an undergrad student)
>> > points where GLPK could be improved. I think I lack the expertise to
>> > mess around with the simplex, but I might be able to work with some
>> > cutting plane generation or some heuristics.
>> >
>> >
>> >
>> > Thank you for you time.
>> >
>> > Renan S. Silva
>> >
>> >
>> >
>> > _______________________________________________
>> > Help-glpk mailing list
>> > address@hidden
>> > https://lists.gnu.org/mailman/listinfo/help-glpk
>>
>> --
>> ==================================================
>> Haroldo Gambini Santos
>> D.Sc, Computer Science
>> Universidade Federal de Ouro Preto
>> http://www.decom.ufop.br/haroldo/
>> _______________________________________________
>> Help-glpk mailing list
>> address@hidden
>> https://lists.gnu.org/mailman/listinfo/help-glpk
>>
>
>
>
> _______________________________________________
> Help-glpk mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/help-glpk
>