Veda2.0 Released!


Unscaled Infeasibilities
#1
Hello Everyone,


I have a problem with my model that CPLEX solution is "Optimal with Unscaled Infeasibilities". 

Finished reading from "C:\Users\Veda\Desktop\Veda\GAMS_WrkTIMES\2050_T32_allscenarios\cplex.opt"



--- GMO memory 106.40 Mb (peak 108.27 Mb)

--- Dictionary memory 0.00 Mb

--- Cplex 12.10.0.0 link memory 7.26 Mb (peak 26.16 Mb)

--- Starting Cplex





--- LP status (5): optimal with unscaled infeasibilities.

--- Cplex Time: 29.66sec (det. 26501.65 ticks)





Optimal solution found, but with infeasibilities after unscaling

Simplex iterations after crossover: 2368

Objective:     26006226.220374







**** REPORT SUMMARY :      194     NONOPT ( NOPT)

                             0 INFEASIBLE

                             0  UNBOUNDED
                          1280  PROJECTED



Any advice how should I look up what is making my solution infeasable ? The .lst file does not show any more details about it. 
Reply
#2
The solver reported that it found an optimal solution, and the solution became infeasible while unscaling it, meaning that some small infeasibilities, which practically always exist in model solutions, then became larger than the infeasibility tolerance. This is not too serious (you can usually quite well accept such a solution), but it indicates that your model has some numerical instability.

See the following thread on the ETSAP Forum for some good advice: Scaling issue
Reply
#3
(12-05-2021, 02:16 AM)Antti-L Wrote: The solver reported that it found an optimal solution, and the solution became infeasible while unscaling it, meaning that some small infeasibilities, which practically always exist in model solutions, then became larger than the infeasibility tolerance. This is not too serious (you can usually quite well accept such a solution), but it indicates that your model has some numerical instability.

See the following thread on the ETSAP Forum for some good advice: Scaling issue

Thank you very much Antti! Indeed, it like you said, and I found out were the numerical instability were. 

Best Regards
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)