I need to convince my boss to approve budget for a commercial MIP solver compared to a free one. I just started to do some modeling with GAMS so my models are still small. You have some bigger ones I can use to show some hopefully significant performance improvement?
Here are some simple but difficult to solve MIP models: http://www.amsterdamoptimization.com/benchmarkmodels.html.
That would be very interesting to see the difference between the academic research and commerical use of GAMS.
ReplyDeleteErwin, could you please tell us any insights on the real world side as from my prevoius experience in the research world, we are always trying to either mathematically find tighter bounds or solve very difficult problems to global optimality.
What is the main focus in the real commerical world and what companies are using GAMS for their commerical use?
Thanks a lot.
Erwin, I am not sure abou the computational time limit of GAMS or any other optimization tools when solving LP or QP problems with small sizes(no more than 50 continuous variables), is that possible to solve them within 5 milliseconds ? What is the critical point to solve an online optimization problem with very little time (around 1-2 milliseconds, more or less the same as adding two small matrix?) if I wrote somthing in C and all cplex or Gurobi, for an small LP, is that possible to speed it up?
ReplyDeletePure academic use often is blurred by consultancy services provided by academics. In the latter case, the use of GAMS is not that much different from commercial use. It also depends a bit on the nature of the research: an OR type of academic may care a lot about tight formulations and global solutions, but others may not care about that so much. In economics the focus is often not on the solver!
ReplyDeleteCplex and Gurobi are are designed to solve large models. They may employ algorithms that are not optimal for very small models. A simpler solver may well do better on these small models. Of course LP is more difficult than matrix addition (that is O(m*n) right?) so I would guess the goal is set too high here.
ReplyDelete