Terminated by solver but optimal

GAMS
1

Hi,

I am using CPLEX for my LP problem. It was terminated by the solver before iteration and resource limits are reached and it reports the optimal solution is found with unscaled infeasibilities. What could cause it to terminate the problem?

Thanks a lot,

Hailey

**** SOLVER STATUS 4 Terminated By Solver
**** MODEL STATUS 1 Optimal
**** OBJECTIVE VALUE 22312642985522.3400

RESOURCE USAGE, LIMIT 244.235 9999999.000
ITERATION COUNT, LIMIT 216673 9999999

IBM ILOG CPLEX 24.2.2 r44857 Released Mar 4, 2014 WEI x86_64/MS Windows
— GAMS/Cplex licensed for continuous and discrete problems.
Cplex 12.6.0.0

Reading parameter(s) from “C:\cplex.opt”

qpmethod 1
numericalemphasis 1
epopt 0.000000001
eprhs 0.000000001
epmrk 0.25
threads 0
Finished reading from “C:\cplex.opt”
Space for names approximately 1020.26 Mb
Use option ‘names no’ to turn use of names off
LP status(5): optimal with unscaled infeasibilities
Cplex Time: 182.08sec (det. 62191.80 ticks)
Optimal solution found, but with infeasibilities after unscaling.
Objective : 22312642985522.340000



**** REPORT SUMMARY : 0 NONOPT
4 INFEASIBLE (INFES)
SUM .
MAX .
MEAN .
0 UNBOUNDED

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/d/optout.

2

So CPLEX is scaling the problem, solving it, then unscaling the solution and checking it, but when it checks the unscaled solution in the original problem some constraints are violated. I see that your objective value is on quite a massive scale, so probably a lot of scaling down occurred. Let’s walk through a toy example and discuss what happens in scaling. Say we’ve got this problem:

(P0)
maximize 3000x + 1125000y
s.t. 1000x + 1000000y > qpmethod 1

numericalemphasis 1
epopt 0.000000001
eprhs 0.000000001
epmrk 0.25
threads 0
Finished reading from “C:\cplex.opt”
Space for names approximately 1020.26 Mb
Use option ‘names no’ to turn use of names off
LP status(5): optimal with unscaled infeasibilities
Cplex Time: 182.08sec (det. 62191.80 ticks)
Optimal solution found, but with infeasibilities after unscaling.
Objective : 22312642985522.340000



**** REPORT SUMMARY : 0 NONOPT
4 INFEASIBLE (INFES)
SUM .
MAX .
MEAN .
0 UNBOUNDED

–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/d/optout.

3

Dear Friends

It’s my great pleasure to contact with you.
While I solve LP models iteratively in a loop in GAMS, I see the following error:

LP status(5): optimal with unscaled infeasibilities

Cplex Time: 0.06sec (det. 1.07 ticks)

Optimal solution found, but with infeasibilities after unscaling.

However the optimal solution achieved but it is not feasible!
I want to set scaling to “OFF”. Since the size of my problem is not very high, I hope setting the scaling to “off” the scaling could help me.

4

You can find the option scanind in the documentation on CPLEX (https://www.gams.com/latest/docs/S_CPLEX.html). If you don’t want matrix scaling the value should be -1.
Cheers
Renger