Dear Gurobiers,
I have a multi-commodity flow model and two variants: one with continuous flow variables (0 \leq x^{k}_{ij} \leq 1 )and one with integer flow variables (x^{k}_{ij} \in {0,1}).
I observe that solver spends much more time at the root node with the continuous model is much than the integer model. I do not understand why?
Continuous model:
Presolved: 10155 rows, 298771 columns, 892859 nonzeros
Variable types: 297044 continuous, 1727 integer (0 binary)
....
176692 2.5817410e+07 0.000000e+00 0.000000e+00 451s
Root relaxation: objective 2.581741e+07, 176692 iterations, 449.27 seconds
Integer model:
Presolved: 10155 rows, 298771 columns, 892859 nonzeros
Variable types: 0 continuous, 298771 integer (297044 binary)
....
89210 2.5817410e+07 0.000000e+00 0.000000e+00 24s
Root relaxation: objective 2.581741e+07, 89210 iterations, 22.47 seconds
It seems that it not only happens with this particular instance but also other instances. So I want to know what is the possible reason.
Thanks,
Lessi.