What is the difference between the nodes Explored time and the cpu time?
AnsweredHi,
Actually I have three questions, I will describe them one by one.
1.The first question: As I asked in the title,what is the difference between the nodes Explored time and the cpu time?
2.The second question: If there is difference between the nodes Explored time and the cpu time, which function should I use to get the cpu time?
3.The third question:
Actually I’m using gurobi to calculate a Unit Commitment model( referred to as UC model).
At first, I used gurobi to solve an initial UC model(I call it model A) and the result was acceptable.
Then I simplified the model by deactivating some redundant constraints(I call it model B). Before that, I have proved that these redundant constraints have no effect on the feasible region of the problem. So I think model B is the same with model A and as I have deactivated some redundant constraints, the problem solving speed should be faster. But the situation is exactly the opposite of what I expected. The speed of solving the problem has not increased but decreased. I can't understand this phenomenon, I mean, even if reducing the constraints does not improve the speed, at least it should be able to maintain the original solution speed.
Can anyone help me？
Thanks a lot.

Hi Zhu,1.The first question: As I asked in the title, what is the difference between the nodes Explored time and the cpu time?The message "Explored X nodes (Y simplex iterations) in Z seconds" shows the wallclock time of the recent call to the optimizer. Note that all times reported by the Gurobi optimizer are wallclock times.2. The second question: If there is difference between the nodes Explored time and the cpu time, which function should I use to get the cpu time?The Gurobi optimizer does not have any function to query the CPU time. You can rely on the standard libraries in the Gurobi API of your choice to access the CPU time.I can't understand this phenomenon, I mean, even if reducing the constraints does not improve the speed, at least it should be able to maintain the original solution speed.If model A and model B represent different formulations of the same model, we can only expect that both models result in the same optimal solution. Although model B has less constraints, it does not necessarily mean that it is an easier model to solve. The extra constraints in model A, despite being redundant, can be used by Gurobi to derive cuts to strengthen the linear relaxations, and hence the bounds at each node of the search tree, improving the runtime.Best regards,Maliheh0

Thank you so much Maliheh！！！
I've been confused by this question for a long time, but it can't torture me anymore hahaha.
Thanks for the help again~
Best wishes，
Zhu
0
Please sign in to leave a comment.
Comments
2 comments