Hello,
I tried to solve an MIP model with mip gap setting as 0%. The problem is same as the article I posted several days ago (
Decision Optimization ). I solved .sav model written from my C# via Interactive.
This time, I found that my best bound kept the same for a long time from the node log, so I refer to the parameter tuning in the slides (page9 & 10:
Mining the CPLEX Node Log for Faster MIP Performance ) and set emphasis mip to 3 (E
mphasize moving best bound). Then, it only took me 225 seconds to find an optimal.
I did the experiment for all "emphasis mip" setting. The results were 3(225s) < 2(270s) < 4(284s) < 1(301s) < 0(more than 1hr).
I would like to know the emphasis mip setting more clearly and understand the magic effects on solving time. I can't perfectly understand the description from IBM Knowledge Center (
https://www.ibm.com/support/knowledgecenter/SSSA5P_12.9.0/ilog.odms.cplex.help/CPLEX/Parameters/topics/MIPEmphasis.html ). Is there anyone who can give me more details and examples related to "Emphasis MIP Setting(0~4)"? Or tell me why "set emphasis mip 3" works on my model?
Thanks!
------------------------------
Chia-Yuan Wu
------------------------------
#DecisionOptimization