integer - CPLEX Error 3019: Failure to solve MIP subproblem -
i have written mixed-ineteger model (mip) in gams , trying solve cplex. encountered error:
cplex error 3019: failure solve mip subproblem
i have 4 cores , 16 gb ram on pc. optimality gap set 1%.
after lot of searching, found suggestions (no solution), , tried them all, yet no luck. every time takes 30min-60min until optimization terminates error. these combinations tried:
- i increased iteration limits 1 million. error 3019.
- some relaxation "feasoptmode 3". error 3019.
- give priority feasibility on optimality using "mipemphasis 1". error 3019.
- change algorithm solving subproblems using "subalg 3". error 3019.
then adding relaxations:
- "feasopt 1", "feastoptmode 3", "mipemphasis 1". out-of-memory
adding heuristics well:
- "feasopt 1", "feastoptmode 3", "mipemphasis 1", "fpheur 2" , "userheurfreq 20". out-of-memory
- "feasopt 1", "feastoptmode 3", "mipemphasis 1", "fpheur 2" , "userheurfreq 30", "subalg 4". out-of-memory
[update] also,
"mipemphasis 1" , "subalg 4". error 3019.
"mipemphasis 4", "subalg 5" , "preind 0". error 3019.
"fpheur -1", "heurfreq -1", "rinsheur -1", "preind 0", "mipemphasis 1", "mipsearch 1". error 3019. (all heuristics off)
so far, had 2.5 million variables. changed model (study period) , reduced number of variables half in 2 different ways, again, same error.
i'm running out of ideas now.
here of links have used (since don't have enough credits in website, can't post more 2 links. put important ones):
i. link1 (ibm): cplex strategies finding feasible solutions when solving mips
ii. link2 (ibm): cplex performance tuning mixed integer programs
finally found solution:
"big m"
the big ms i'm using in model (for linearization , other constraint) responsible. big m large; removed 4 of nines , viola. took 3 days figure out.
i increased iteration limit 2 million , optimality gap 2%. (iteration limit wasn't necessary, optimality gap helped). returned solution after 5 hours.
Comments
Post a Comment