Advice on how to speed up of MILP (Pyomo with Gurobi)
AnsweredHi,
I want to solve a MILP, which optimizes the operation of an energy system. It is constructed in pyomo and includes some binary variables for electrolyzer, fuel cell, battery operation and simple linear constraints for linearization of component power-dependent performance. Currently the model is constructed for 1 year with an hourly resolution. Later I want to increase the temporal resolution to 15min, which will increase model complexity further.
Unfortunately the model runtime is yet quite long:
What I already did:
- Use Gurobi Python interface
- Reduction of Matrix range and RHS range through model unit adaption
- Set MIPGap: 1e-3
Following you can find the Gurobi log:
Operational optimization: 2023-03-31 10:29:16 Create model
[ 0.00] Start
[+ 22.29] Built model
45.15 seconds required for presolve
Set parameter MIPGap to value 0.001
Gurobi Optimizer version 10.0.1 build v10.0.1rc0 (win64)
CPU model: Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz, instruction set [SSE2|AVX|AVX2]
Thread count: 4 physical cores, 8 logical processors, using up to 8 threads
Optimize a model with 394203 rows, 280320 columns and 849720 nonzeros
Model fingerprint: 0x96fb3a6f
Variable types: 236520 continuous, 43800 integer (43800 binary)
Coefficient statistics:
Matrix range [1e-04, 1e+02]
Objective range [8e-02, 4e-01]
Bounds range [1e-01, 1e+00]
RHS range [1e-03, 2e+02]
Presolve removed 175833 rows and 105753 columns
Presolve time: 2.47s
Presolved: 218370 rows, 174567 columns, 631053 nonzeros
Variable types: 139527 continuous, 35040 integer (35040 binary)
Deterministic concurrent LP optimizer: primal simplex, dual simplex, and barrier
Showing barrier log only...
Root barrier log...
Ordering time: 0.09s
Barrier statistics:
AA' NZ : 7.711e+05
Factor NZ : 2.037e+06 (roughly 170 MB of memory)
Factor Ops : 2.347e+07 (less than 1 second per iteration)
Threads : 1
Objective Residual
Iter Primal Dual Primal Dual Compl Time
0 1.76555678e+05 -1.34829695e+06 1.56e+01 6.40e-01 1.70e+01 4s
1 4.88230074e+04 -4.94777143e+05 9.42e-01 8.57e-13 1.84e+00 5s
2 1.77898636e+04 -1.16781344e+05 1.53e-01 5.51e-13 3.50e-01 5s
3 1.32373991e+04 -4.16257869e+04 4.91e-02 7.52e-13 1.37e-01 5s
4 7.77317370e+03 -1.73993245e+04 1.53e-02 6.99e-13 5.87e-02 5s
5 5.62039141e+03 -5.76944462e+03 9.37e-03 6.52e-13 2.69e-02 6s
6 4.42107260e+03 1.36252422e+02 4.20e-03 3.94e-13 1.07e-02 6s
7 3.35967175e+03 1.60088655e+03 1.25e-03 3.10e-13 4.14e-03 6s
8 3.03861308e+03 2.26901367e+03 5.46e-04 3.02e-13 1.81e-03 6s
9 2.97625395e+03 2.40450356e+03 4.03e-04 3.73e-13 1.34e-03 6s
10 2.91438280e+03 2.54059631e+03 2.68e-04 3.65e-13 8.80e-04 7s
11 2.88680196e+03 2.63809845e+03 2.08e-04 3.68e-13 5.99e-04 7s
12 2.85812332e+03 2.69519809e+03 1.52e-04 3.61e-13 4.00e-04 7s
13 2.84592379e+03 2.72184771e+03 1.30e-04 4.30e-13 3.12e-04 7s
14 2.83691210e+03 2.74429451e+03 1.10e-04 4.01e-13 2.39e-04 8s
15 2.82297062e+03 2.75688416e+03 8.12e-05 4.02e-13 1.72e-04 8s
16 2.81399491e+03 2.76506030e+03 6.09e-05 4.56e-13 1.27e-04 8s
17 2.81015861e+03 2.77142690e+03 5.21e-05 4.35e-13 1.03e-04 8s
18 2.80616493e+03 2.77377590e+03 4.28e-05 4.96e-13 8.55e-05 9s
19 2.80211064e+03 2.77821025e+03 3.34e-05 5.16e-13 6.40e-05 9s
20 2.79890466e+03 2.78076670e+03 2.59e-05 5.34e-13 4.88e-05 9s
21 2.79684065e+03 2.78306590e+03 2.09e-05 4.50e-13 3.77e-05 9s
22 2.79567391e+03 2.78458844e+03 1.81e-05 5.07e-13 3.10e-05 10s
23 2.79433760e+03 2.78556191e+03 1.49e-05 4.92e-13 2.48e-05 10s
24 2.79370105e+03 2.78596852e+03 1.33e-05 4.99e-13 2.19e-05 10s
25 2.79297049e+03 2.78630025e+03 1.17e-05 5.95e-13 1.90e-05 11s
26 2.79277439e+03 2.78688053e+03 1.11e-05 5.90e-13 1.72e-05 11s
27 2.79229161e+03 2.78714042e+03 1.02e-05 6.19e-13 1.52e-05 12s
28 2.79151945e+03 2.78736682e+03 8.32e-06 5.59e-13 1.23e-05 12s
29 2.79146575e+03 2.78748858e+03 8.20e-06 6.77e-13 1.19e-05 12s
30 2.79118906e+03 2.78774945e+03 7.35e-06 5.37e-13 1.04e-05 13s
31 2.79069586e+03 2.78795931e+03 6.12e-06 5.21e-13 8.37e-06 13s
32 2.79061962e+03 2.78802919e+03 5.87e-06 6.27e-13 7.96e-06 13s
33 2.79010511e+03 2.78830812e+03 4.46e-06 5.27e-13 5.69e-06 13s
34 2.78989833e+03 2.78851256e+03 3.81e-06 4.05e-13 4.56e-06 14s
35 2.78961240e+03 2.78858437e+03 2.98e-06 4.19e-13 3.44e-06 14s
36 2.78939062e+03 2.78865718e+03 2.35e-06 4.02e-13 2.55e-06 14s
37 2.78928644e+03 2.78868137e+03 2.05e-06 4.58e-13 2.15e-06 14s
38 2.78923191e+03 2.78870471e+03 1.85e-06 4.93e-13 1.90e-06 15s
39 2.78911299e+03 2.78873596e+03 1.33e-06 4.39e-13 1.36e-06 15s
40 2.78908296e+03 2.78876560e+03 1.19e-06 3.40e-13 1.18e-06 15s
41 2.78901208e+03 2.78878442e+03 9.97e-07 3.13e-13 9.09e-07 15s
42 2.78895533e+03 2.78878558e+03 7.90e-07 2.84e-13 6.97e-07 16s
43 2.78892351e+03 2.78879122e+03 6.33e-07 2.89e-13 5.50e-07 16s
44 2.78888523e+03 2.78879649e+03 4.25e-07 3.62e-13 3.69e-07 16s
45 2.78888295e+03 2.78879839e+03 4.12e-07 4.01e-13 3.55e-07 16s
46 2.78884046e+03 2.78880074e+03 1.74e-07 3.38e-13 1.58e-07 17s
47 2.78882116e+03 2.78880274e+03 9.26e-08 2.45e-13 7.82e-08 17s
48 2.78880741e+03 2.78880645e+03 5.41e-09 2.25e-13 4.35e-09 17s
49 2.78880653e+03 2.78880653e+03 3.10e-11 2.25e-13 6.79e-13 17s
Barrier solved model in 49 iterations and 17.13 seconds (7.53 work units)
Optimal objective 2.78880653e+03
Root crossover log...
74590 DPushes remaining with DInf 0.0000000e+00 17s
0 DPushes remaining with DInf 0.0000000e+00 18s
6707 PPushes remaining with PInf 0.0000000e+00 18s
0 PPushes remaining with PInf 0.0000000e+00 19s
Push phase complete: Pinf 0.0000000e+00, Dinf 8.9258473e-12 19s
Root simplex log...
Iteration Objective Primal Inf. Dual Inf. Time
53893 2.7888065e+03 0.000000e+00 0.000000e+00 19s
Concurrent spin time: 0.00s
Solved with barrier
53893 2.7888065e+03 0.000000e+00 0.000000e+00 19s
Root relaxation: objective 2.788807e+03, 53893 iterations, 15.39 seconds (7.59 work units)
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time
0 0 2788.80653 0 524 - 2788.80653 - - 22s
H 0 0 2834.0840604 2788.80653 1.60% - 25s
H 0 0 2832.0327700 2788.80653 1.53% - 27s
0 0 2797.57721 0 235 2832.03277 2797.57721 1.22% - 29s
H 0 0 2802.4356704 2797.57721 0.17% - 30s
0 0 2797.57847 0 233 2802.43567 2797.57847 0.17% - 30s
0 0 2798.12463 0 145 2802.43567 2798.12463 0.15% - 31s
H 0 0 2799.9059793 2798.12463 0.06% - 32s
Cutting planes:
Gomory: 251
Cover: 2
Implied bound: 59
MIR: 196
Flow cover: 1082
RLT: 38
Relax-and-lift: 140
Explored 1 nodes (56142 simplex iterations) in 32.58 seconds (23.12 work units)
Thread count was 8 (of 8 available processors)
Solution count 4: 2799.91 2802.44 2832.03 2834.08
Optimal solution found (tolerance 1.00e-03)
Best objective 2.799905979318e+03, best bound 2.798124630783e+03, gap 0.0636%
32.91 seconds required for solver
2.80 seconds required for postsolve
[+ 101.47] Wrote LP file and solved
[+ 0.08] Loaded results
[+ 61.15] Save model txt
Any further hints to reduce runtime?
- Are coefficeint statistics looking good enough/could they still have an impact on runtime? It would be possible to change the objective range to make it bigger than 1.
Further, I wonder why model runtime is heavily dependent on energy system component sizes. It can change by the factor of 10-100 between a system with a small and a big hydrogen or battery storage. This is therefore relevant because the MILP shall be later included into another sizing model, which will run the MILP for many different component sizes.
Do you have any recommendations how to handle this application of the MILP and how to deal with Gurbobi parameter setting in such a case?
Thank you very much for your support.
Fabian
0
-
Following the gurobi log for the model with larger energy system component sizes.
I increase MIPGap to 2e-3 to reduce runtime. But it is still obvious that Gurobi needs most of the time to close the final steps of MIPGap.Operational optimization: 2023-03-31 11:15:15 Create model
[ 0.00] Start
[+ 21.17] Built model
44.18 seconds required for presolve
Set parameter MIPGap to value 0.002
Gurobi Optimizer version 10.0.1 build v10.0.1rc0 (win64)
CPU model: Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz, instruction set [SSE2|AVX|AVX2]
Thread count: 4 physical cores, 8 logical processors, using up to 8 threads
Optimize a model with 394203 rows, 280320 columns and 849720 nonzeros
Model fingerprint: 0xec6edcb1
Variable types: 236520 continuous, 43800 integer (43800 binary)
Coefficient statistics:
Matrix range [5e-05, 1e+02]
Objective range [8e-02, 4e-01]
Bounds range [1e-01, 1e+00]
RHS range [1e-03, 2e+02]
Presolve removed 177271 rows and 106921 columns
Presolve time: 1.91s
Presolved: 216932 rows, 173399 columns, 627906 nonzeros
Variable types: 138359 continuous, 35040 integer (35040 binary)
Deterministic concurrent LP optimizer: primal simplex, dual simplex, and barrier
Showing barrier log only...
Root barrier log...
Ordering time: 0.06s
Barrier statistics:
AA' NZ : 7.675e+05
Factor NZ : 1.994e+06 (roughly 170 MB of memory)
Factor Ops : 2.265e+07 (less than 1 second per iteration)
Threads : 1
Objective Residual
Iter Primal Dual Primal Dual Compl Time
0 2.45125532e+05 -1.39920366e+06 1.64e+01 6.40e-01 1.94e+01 4s
1 4.96627402e+04 -5.69950068e+05 9.31e-01 3.24e-12 1.99e+00 4s
2 5.18358830e+03 -1.26332669e+05 1.78e-01 2.09e-12 3.60e-01 4s
3 3.82356060e+03 -6.20434619e+04 9.58e-02 2.41e-12 1.95e-01 4s
4 3.45107074e+03 -3.57925084e+04 3.97e-02 1.93e-12 1.08e-01 4s
5 3.27937901e+03 -1.41215494e+04 2.11e-02 1.64e-12 5.05e-02 5s
6 2.58142443e+03 -3.68566402e+03 9.58e-03 1.28e-12 1.94e-02 5s
7 2.03210208e+03 -1.57522820e+02 3.04e-03 1.20e-12 6.74e-03 5s
8 1.81230783e+03 6.45056512e+02 1.38e-03 1.31e-12 3.39e-03 5s
9 1.73008747e+03 1.04709209e+03 9.33e-04 1.36e-12 2.02e-03 5s
10 1.67573648e+03 1.31598653e+03 4.98e-04 1.42e-12 1.07e-03 5s
11 1.63629627e+03 1.42403637e+03 2.10e-04 1.43e-12 5.78e-04 6s
12 1.61619020e+03 1.52188124e+03 8.65e-05 1.30e-12 2.54e-04 6s
13 1.60792931e+03 1.56116172e+03 4.46e-05 1.24e-12 1.28e-04 6s
14 1.60392054e+03 1.57377593e+03 2.72e-05 1.39e-12 8.13e-05 6s
15 1.60130056e+03 1.58409986e+03 1.60e-05 1.42e-12 4.67e-05 7s
16 1.59993945e+03 1.59118177e+03 1.05e-05 1.35e-12 2.56e-05 7s
17 1.59887686e+03 1.59300101e+03 6.50e-06 1.51e-12 1.68e-05 7s
18 1.59812399e+03 1.59499507e+03 3.86e-06 1.34e-12 9.22e-06 7s
19 1.59772259e+03 1.59589958e+03 2.39e-06 1.31e-12 5.48e-06 7s
20 1.59755894e+03 1.59633353e+03 1.82e-06 1.38e-12 3.84e-06 8s
21 1.59738047e+03 1.59654978e+03 1.27e-06 1.51e-12 2.63e-06 8s
22 1.59727934e+03 1.59673521e+03 9.02e-07 1.42e-12 1.78e-06 8s
23 1.59712908e+03 1.59688683e+03 3.94e-07 1.18e-12 7.86e-07 8s
24 1.59708170e+03 1.59695465e+03 2.32e-07 1.18e-12 4.32e-07 8s
25 1.59706959e+03 1.59699278e+03 1.91e-07 1.10e-12 3.00e-07 9s
26 1.59704888e+03 1.59699591e+03 1.23e-07 1.57e-12 2.00e-07 9s
27 1.59704216e+03 1.59700084e+03 1.00e-07 1.64e-12 1.59e-07 9s
28 1.59703995e+03 1.59700458e+03 9.30e-08 1.53e-12 1.42e-07 9s
29 1.59701348e+03 1.59700955e+03 9.49e-09 9.02e-13 1.52e-08 9s
30 1.59701020e+03 1.59700991e+03 1.36e-09 9.01e-13 6.72e-10 10s
31 1.59701003e+03 1.59701002e+03 5.75e-11 8.94e-13 2.10e-11 10s
Barrier solved model in 31 iterations and 9.97 seconds (5.40 work units)
Optimal objective 1.59701003e+03
Root crossover log...
74619 DPushes remaining with DInf 0.0000000e+00 10s
0 DPushes remaining with DInf 0.0000000e+00 12s
5629 PPushes remaining with PInf 1.0123747e-04 12s
0 PPushes remaining with PInf 0.0000000e+00 12s
Push phase complete: Pinf 0.0000000e+00, Dinf 1.3812786e-11 12s
Root simplex log...
Iteration Objective Primal Inf. Dual Inf. Time
53802 1.5970100e+03 0.000000e+00 0.000000e+00 12s
Concurrent spin time: 0.02s
Solved with barrier
53802 1.5970100e+03 0.000000e+00 0.000000e+00 12s
Root relaxation: objective 1.597010e+03, 53802 iterations, 9.45 seconds (6.48 work units)
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time
0 0 1597.01002 0 870 - 1597.01002 - - 15s
H 0 0 1638.4332602 1597.01002 2.53% - 16s
H 0 0 1636.6899077 1597.01002 2.42% - 18s
0 0 1602.09632 0 294 1636.68991 1602.09632 2.11% - 21s
H 0 0 1612.6719822 1602.09632 0.66% - 22s
0 0 1602.11074 0 292 1612.67198 1602.11074 0.65% - 23s
0 0 1602.51257 0 253 1612.67198 1602.51257 0.63% - 24s
H 0 0 1606.9854239 1602.51257 0.28% - 25s
0 0 1602.51474 0 251 1606.98542 1602.51474 0.28% - 25s
0 0 1602.63747 0 244 1606.98542 1602.63747 0.27% - 29s
0 0 1602.64233 0 244 1606.98542 1602.64233 0.27% - 29s
0 0 1602.65455 0 244 1606.98542 1602.65455 0.27% - 32s
H 0 0 1606.6696528 1602.65455 0.25% - 33s
0 0 1602.65941 0 242 1606.66965 1602.65941 0.25% - 34s
0 0 1602.65941 0 242 1606.66965 1602.65941 0.25% - 37s
0 2 1602.65941 0 242 1606.66965 1602.65941 0.25% - 40s
23 28 1602.78294 6 232 1606.66965 1602.74693 0.24% 16.3 45s
51 63 1602.84775 12 225 1606.66965 1602.74693 0.24% 8.6 50s
72 92 1602.90630 18 216 1606.66965 1602.74693 0.24% 6.8 55s
114 140 1603.00445 28 207 1606.66965 1602.74693 0.24% 5.5 60s
155 184 1603.04658 37 195 1606.66965 1602.74693 0.24% 5.1 68s
183 229 1603.05965 45 188 1606.66965 1602.74693 0.24% 4.7 71s
306 354 1603.10793 76 161 1606.66965 1602.74693 0.24% 3.8 76s
390 423 1603.12230 90 156 1606.66965 1602.74693 0.24% 3.6 81s
462 501 1603.13117 108 151 1606.66965 1602.74693 0.24% 3.2 86s
527 579 1603.16362 127 144 1606.66965 1602.74693 0.24% 3.2 91s
H 578 583 1606.6018997 1602.74693 0.24% 3.0 94s
H 579 583 1606.5835226 1602.74693 0.24% 3.0 94s
582 625 1603.18185 141 142 1606.58352 1602.74693 0.24% 3.0 97s
624 675 1603.18219 152 143 1606.58352 1602.74693 0.24% 3.0 100s
708 773 1603.18375 173 144 1606.58352 1602.74693 0.24% 2.8 107s
772 811 1603.19702 192 141 1606.58352 1602.74693 0.24% 2.7 110s
873 930 1603.26482 223 132 1606.58352 1602.74693 0.24% 2.6 118s
929 987 1603.27464 239 131 1606.58352 1602.74693 0.24% 2.6 121s
986 1082 1603.27547 247 128 1606.58352 1602.74693 0.24% 2.5 125s
1137 1258 1603.31566 285 118 1606.58352 1602.74693 0.24% 2.3 131s
1257 1343 1603.34173 305 115 1606.58352 1602.74693 0.24% 2.3 135s
1396 1451 1603.42187 324 107 1606.58352 1602.74693 0.24% 2.3 142s
1450 1503 1603.46607 335 106 1606.58352 1602.74693 0.24% 2.3 157s
1502 1595 1603.46653 348 107 1606.58352 1602.74693 0.24% 2.4 161s
1594 1599 1603.46770 364 104 1606.58352 1602.74693 0.24% 2.4 225s
H 1598 1632 1606.5576683 1602.74693 0.24% 2.4 230s
1631 1704 1603.47128 373 103 1606.55767 1602.74693 0.24% 2.4 235s
H 1647 1704 1606.5521773 1602.74693 0.24% 2.4 235s
H 1685 1704 1606.5413203 1602.74693 0.24% 2.3 235s
1703 1782 1603.47187 389 102 1606.54132 1602.74693 0.24% 2.3 241s
1781 1865 1603.47244 401 100 1606.54132 1602.74693 0.24% 2.3 247s
1864 1949 1603.47268 418 100 1606.54132 1602.74693 0.24% 2.3 253s
1948 2061 1603.49309 433 100 1606.54132 1602.74693 0.24% 2.3 261s
H 1975 2061 1606.5299138 1602.74693 0.24% 2.3 261s
2060 2169 1603.56768 456 97 1606.52991 1602.74693 0.24% 2.3 268s
2168 2278 1603.71159 480 94 1606.52991 1602.74693 0.24% 2.4 275s
2277 2331 1603.80298 508 89 1606.52991 1602.74693 0.24% 2.4 282s
H 2280 2331 1606.5297627 1602.74693 0.24% 2.4 282s
H 2294 2331 1606.5198520 1602.74693 0.23% 2.4 282s
2330 2440 1603.80323 522 88 1606.51985 1602.74693 0.23% 2.3 290s
H 2335 2440 1606.3674171 1602.74693 0.23% 2.3 290s
2439 2623 1603.88712 548 86 1606.36742 1602.74693 0.23% 2.4 298s
2622 2774 1603.96610 579 82 1606.36742 1602.74693 0.23% 2.4 305s
2773 2922 1603.98755 600 81 1606.36742 1602.74693 0.23% 2.4 313s
2921 3081 1604.00874 627 78 1606.36742 1602.74693 0.23% 2.5 322s
H 2958 3081 1606.2246328 1602.74693 0.22% 2.5 322s
3080 3221 1604.37179 657 73 1606.22463 1602.74693 0.22% 2.5 331s
H 3220 3368 1606.2151581 1602.74693 0.22% 2.6 340s
3367 3519 1604.31636 705 70 1606.21516 1602.74693 0.22% 2.6 350s
3518 3678 1604.34502 731 68 1606.21516 1602.74693 0.22% 2.7 360s
3677 3840 1604.44704 762 65 1606.21516 1602.74693 0.22% 2.7 370s
H 3693 3840 1606.1699375 1602.74693 0.21% 2.7 370s
H 3839 3844 1605.8834888 1602.74693 0.20% 2.7 450s
H 3842 3844 1605.8692771 1602.74693 0.19% 2.7 451s
Cutting planes:
Gomory: 120
Implied bound: 199
MIR: 314
Flow cover: 1824
Relax-and-lift: 115
Explored 3843 nodes (66342 simplex iterations) in 451.27 seconds (326.43 work units)
Thread count was 8 (of 8 available processors)
Solution count 10: 1605.87 1605.88 1606.17 ... 1606.54
Optimal solution found (tolerance 2.00e-03)
Best objective 1.605869277090e+03, best bound 1.602746932564e+03, gap 0.1944%
451.58 seconds required for solver
3.46 seconds required for postsolve
[+ 525.81] Wrote LP file and solved
[+ 0.08] Loaded results
[+ 84.37] Save model txt0 -
-
-
Gurobi Staff
Hi Fabian,
Are coefficeint statistics looking good enough/could they still have an impact on runtime? It would be possible to change the objective range to make it bigger than 1.
The coefficient statistics of your models look reasonable. As stated in Gurobi’s numerical guidelines, the general rule of thumb is to have matrix coefficients to be contained in 6 orders of magnitude and ideally be within \([10^{-3}, 10^6]\). Please note that bad numbers in the model is only one of the potential sources of numerical challenges. Having almost parallel constraints in the model (see this example) is another potential source of numerical challenges.
Further, I wonder why model runtime is heavily dependent on energy system component sizes.
How does changing the component size reflect on the model? From your logs, it seems that it changes the constraint coefficients and as a result changes the feasible region. Therefore, it is not unexpected that the runtime varies depending on the feasible region.
Do you have any recommendations how to handle this application of the MILP and how to deal with Gurbobi parameter setting in such a case?
To close the gap sooner based on your last log, it would make sense to reach a high quality solution faster. You can consider experimenting with the parameters below:
- NoRelHeurTime: Experiment with the NoRel heuristic which is effective in finding high-quality solutions before solving the root relaxation. You might start by setting this parameter to 100 seconds and then adjust based on its progress and effectiveness.
- Heuristics: Increase the percentage of the total runtime spent on heuristic algorithms. You can start by bumping the default value to 0.1 or 0.2.
- Cuts=1: Decrease the intensity of applying cutting plane generation
- MIPFocus=1: Change the focus of the optimizer to reaching a high quality solution faster
- Presolve=2: Intensify the utilization of the presolve procedure techniques in the hope of reducing the presolved model size and tightening it
Best regards,
Maliheh
0 -
-
Dear Maliheh,
thank you very much for your feedback and recommendations.
I will experiment with the parameters you mentioned, thanks for the support.
Fabian
0 -
-
Hi! I'm having right now the same kind of problem with the modelling of battery operations with resolution on the second...
Did you find any solution to speed up the process?
Andrea0 -
-
Hi Andrea,
I could identify some improvements with the changing of the recommended solver settings.
But nevertheless when the problem changes, identified parameters might be again not-optimal.
Best
Fabian0 -
Please sign in to leave a comment.
Comments
5 comments