Originally posted by: CarlosSuazoM.

Hey guys,

I am dealing with the following MIP problem, which I want to solve up top MIPGap start 1%:

http://www.spec.cl/downloads/Tx_INVX_lp.zip

Currently is taking hours to find the root relaxation, so I am wondering if there is any trick, parameter setting that could help me to solve it faster.

Could distributed optimization accelerate convergence time?

Many thanks in advance,

Carlos Suazo

Originally posted by: BoJensen

I suggest you first read this tech note. Especially the section about our automatic tuner.

Originally posted by: CarlosSuazoM.

I tried automatic tuner, and it recommend me to use Primal as startalgorithm and subalgorithm to solve the MIP problem.

With default options is taking almost 15000 sec to solve Root relaxation problem!

My problem is a sequentially dependent decision model, so the decisions made at a particular time affect decisions made later. I am wondering if I could improve formulation to reduce convergence time.
For this multi-period (n periods) problem, we introduced a binary decision variable called I_t into each period t to show how a decision should be made in each period.

Let n = 4, then dependency is expressed as follows:

I_1 <= I_2
I_2 <= I_3
I_3 <= I_4

How will affect convergence time if we add following constraints??? Is it recommendable? 

I_1 <= I_3
I_1 <= I_4
I_2 <= I_4

Will these additional constraints help convergence during branch and cut execution???

Many thanks in advance,

Carlos