Originally posted by: benhamou

I try to give to give an intial solution while solving a linear problem using distributed MIP bu it is not accepted. I got this message:

Processing 1 MIP starts.
MIP start 'm1' defined solution with objective 331558.7500.
1 of 1 MIP starts provided solutions.
MIP start 'm1' defined initial solution with objective 331558.7500.
Tried aggregator 2 times.
MIP Presolve eliminated 0 rows and 364776 columns.
Aggregator did 783 substitutions.
Reduced MIP has 1340 rows, 695688 columns, and 5664395 nonzeros.
Reduced MIP has 695688 binaries, 0 generals, 0 SOSs, and 0 indicators.
Presolve time = 5.52 sec. (3119.41 ticks)
Running distributed MIP on 2 solvers.
Setting up 2 distributed solvers.
Setup time = 5.87 sec. (9.04 ticks)
Starting ramp-up.

Processing 1 MIP starts.
MIP start 'm1' defined no solution.
Retaining values of one MIP start for possible repair.

BUT  when I use MIP and not distributed MIP the initial solution is accepted. Is there any parameterization of cplex that enable distributed MIP to accept initial solution as it is the case in ordinary MIP ?

Thanks.

In your log you can see two sets of output lines that relate to your MIP starts:

Processing 1 MIP starts.
MIP start 'm1' defined solution with objective 331558.7500.
1 of 1 MIP starts provided solutions.
MIP start 'm1' defined initial solution with objective 331558.7500.
...

Starting ramp-up.

Processing 1 MIP starts.
MIP start 'm1' defined no solution.
Retaining values of one MIP start for possible repair.

As you can see, your MIP start is accepted by the master solver before it starts presolve. Later, when the presolved model is launched on a parallel worker then a MIP start is rejected on the parallel worker. That could be due to numerics or other problems but it should actually not be an issue. The worker should still use 331558.7500 as best known feasible objective value. What is the "best integer" value you see in the rampup log? It should be 331558.7500 or better.

Originally posted by: benhamou

hello Daniel,

here you are the ramp up log. As you can see worker begin with a bad value start solution !!!

Processing 1 MIP starts.
MIP start 'm1' defined solution with objective 331558.7500.
1 of 1 MIP starts provided solutions.
MIP start 'm1' defined initial solution with objective 331558.7500.
Tried aggregator 2 times.
MIP Presolve eliminated 0 rows and 364776 columns.
Aggregator did 783 substitutions.
Reduced MIP has 1340 rows, 695688 columns, and 5664395 nonzeros.
Reduced MIP has 695688 binaries, 0 generals, 0 SOSs, and 0 indicators.
Presolve time = 5.52 sec. (3119.41 ticks)
Running distributed MIP on 2 solvers.
Setting up 2 distributed solvers.
Setup time = 5.87 sec. (9.04 ticks)
Starting ramp-up.

Processing 1 MIP starts.
MIP start 'm1' defined no solution.
Retaining values of one MIP start for possible repair.
Found incumbent of value 1.5030301e+08 after 1.42 sec. (768.38 ticks)
Dual steepest-edge pricing selected.
Probing time = 1.57 sec. (263.99 ticks)
Clique table members: 1414.
MIP emphasis: balance optimality and feasibility.
MIP search method: dynamic search.
Parallel mode: deterministic, using up to 8 threads.
Root relaxation solution time = 25.87 sec. (10977.62 ticks)

        Nodes                                         Cuts/
   Node  Left     Objective  IInf  Best Integer    Best Bound    ItCnt     Gap

*     0+    0                       1.50303e+08    81098.5000            99.95%
Found incumbent of value 1.5030301e+08 after 39.44 sec. (17463.15 ticks)
*     0+    0                       1.50303e+08    81098.5000            99.95%
Found incumbent of value 1.5030269e+08 after 39.45 sec. (17476.41 ticks)
*     0+    0                       9.03403e+07    81098.5000            99.91%
Found incumbent of value 9.0340340e+07 after 39.46 sec. (17484.38 ticks)

OK, this is unexpected. Can you share the offending model either here or with daniel(dot)junglas(at)de(dot)ibm(dot)com so that I can take a look?

Originally posted by: benhamou

hello daniel 

I can't send you files through this site ( limit problem ). I sent them to your ibm mail. mail subject is: initial solution using distMip

Thanks