Network simplex does not use a warm-start basis

Dear Gurobi Support team,

Hello! I noticed when using network simplex by setting

model.setParam("NetworkAlg", 1)

to solve a min-cost-flow problem, Guroobi will not apply my warm-start basis even if it is feasible.

However, when I use the Simplex method by setting

model.setParam("Method", 0)

it accepts my warm-start basis.

The log files are as follows:

Optimize a model with 16385 rows, 147456 columns and 294912 nonzeros
Coefficient statistics:
Matrix range [1e+00, 1e+00]
Objective range [1e-04, 4e+08]
Bounds range [1e+00, 3e+03]
RHS range [2e+00, 4e+03]
Starting network simplex...
Iteration Objective Primal Inf. Dual Inf. Time
0 8.5217352e+17 1.280000e+05 8.521735e+14 0s
Extra simplex iterations after uncrush: 35
92768 1.7720569e+05 0.000000e+00 0.000000e+00 0s

Solved in 92768 iterations and 0.49 seconds (0.49 work units)
Optimal objective 1.772056888e+05

Optimize a model with 16385 rows, 147456 columns and 294912 nonzeros
Coefficient statistics:
Matrix range [1e+00, 1e+00]
Objective range [1e-04, 4e+08]
Bounds range [1e+00, 3e+03]
RHS range [2e+00, 4e+03]
Iteration Objective Primal Inf. Dual Inf. Time
0 1.7723106e+05 0.000000e+00 7.547598e+01 0s
1098 1.7720569e+05 0.000000e+00 0.000000e+00 0s

Solved in 1098 iterations and 0.07 seconds (0.09 work units)
Optimal objective 1.772056888e+05]

Could you tell me if this is really the case?

In addition, the parameter "LPWarmStart" is always set to 2 and I also turn off the presolve:

model.setParam("LPWarmStart", 2)
model.setParam("Presolve", 0)

Many thanks for your time and consideration!

Best regards,

Cheng

Network simplex does not use a warm-start basis

Comments

Didn't find what you were looking for?