Identifying infeasible runs

Problems with modeling
Post Reply
teo21r
User
User
Posts: 10
Joined: 3 years ago

Identifying infeasible runs

Post by teo21r »

Hello,

I am running a GAMS model once for each observation/firm in my dataset. The total number of observations is 121 (set i /1*121/). I get a locally infeasible solution (i.e., Model Status 5 Locally Infeasible). The picture below shows the summary report I get for the variable that defines the objective value to be optimized (i.e., z2)
Image

If I am right, number 8 next to INFEASIBLE denotes the rows/columns in my model that are infeasible. This number does not necessarily reflect which of the 121 runs resulted in an infeasible solution, right (I am saying this because I have run models where this number exceeded the total number of observations in my dataset)?

Is there a way to see which of the 121 performed runs resulted in an infeasible solution? If yes, then is there a command that will allow me to export the related observation IDs (i.e., the IDs of those observations (i, i=1,...,121) for which the model yielded an infeasible solution)?

Thanks,
Teo
Attachments
pic1.png
pic1.png (4.93 KiB) Viewed 1190 times
Fred
Posts: 372
Joined: 7 years ago

Re: Identifying infeasible runs

Post by Fred »

Hi Teo,

If I understand correctly, you are are solving 121 models. The Report summary belongs to a particular solve, so you should get 121 report summaries in total. So the 8 infeasibilities refer to a particular solve where e.g. 8 constraints may be violated in the returned infeasible "solution" (see documentation for further details).

You can access the model status via model attribute modelstat, e.g.

Code: Select all

set i / i1*i121/;
parameter status(i)
[...]
loop(i,
  solve myModel ...;
  status(i) = myModel. modelstat;
  [...]
);  
I hope this helps!

Fred
Post Reply