Originally posted by: Brent_Thomas

I'm running a sizeable MIP using Cplex 12 via GAMS. Every now and then, Cplex will report in the iteration log, "Warning: form length is zero." Despite the warning, the branch & bound merrily goes its way and wends towards the optimum. I'm not sure what Cplex is trying to tell me, and I was wondering if other users might be familiar with the message. Is this an indicator of something that needs to be addressed in my formulation, or can I simply take heart that I'm converging where I ought.

Thanks for any help!
Brent
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

This is interesting. I have a guess where this comes from, but I need to check. Is it possible for you to dump the model into a *.sav file and send it to me via email?
If the file would be too large as email attachment, I can tell you a different way to send it.

My address is achterberg (at) de (dot) ibm (dot) com
Thanks,

Tobias
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Hi Brent,

thanks for sending me your data. I was able to reproduce the behavior and analyze the issue. The underlying cause is actually your setting of the numerical tolerances to such large values like epint = eprhs = 1e-2. This was confusing our code at one place which lead to the warning message. But this is really just a cosmetic issue; the warning itself is not critical in this situation.

The issue will be fixed in the next technical release.
Tobias
#CPLEXOptimizers
#DecisionOptimization