Re: specially ordered sets variables in Mathprog

[Erwin Kalvelagen]

Binary variables are usually faster than SOS1 sets, especially when the solver can apply some good cuts.Of course here are cases that can be modeled with SOS1 sets and not with binary variables, but that is not what you are dealing with..

----------------------------------------------------------------
Erwin Kalvelagen
Amsterdam Optimization Modeling Group
erwin@amsterdamoptimization.com
https://www.amsterdamoptimization.com
----------------------------------------------------------------

On Tue, Mar 15, 2022 at 6:48 AM Tommaso Cucinotta <tommaso.cucinotta@santannapisa.it> wrote:

Hi,

I'd like to know whether Mathprog supports (or is there any plan to support) SOS (special ordered sets) variables, at
least type 1.

I often use mathprog and glpsol to convert MILP problems (actually, BLP mostly) for other solvers (CBC, CPLEX), and I'm
seeing these other solvers support these types of variables, what makes me think that there might be some advantages in
formalizing a problem this way, as perhaps the solver might be quicker in scanning through the possible branches etc...
I mean, using the SOS1 native formalization, instead of equivalent formulations as in

https://en.wikibooks.org/wiki/GLPK/Modeling_tips#Special_ordered_sets

Thanks,

     T.

--
Tommaso Cucinotta, Computer Engineering PhD
Associate Professor at the Real-Time Systems Laboratory (ReTiS)
Scuola Superiore Sant'Anna, Pisa, Italy
http://retis.sssup.it/people/tommaso