Page 1 of 1
Binary variables
Posted: Wed Aug 05, 2020 2:29 pm
by Alexanre.ito
Hi I have no more ideas to correct my model.
I don't know why my binary variables is not a binary.
The binary result is not a binary, even declaring that as "Binary Variable"
could some one help me with this?
Re: Binary variables
Posted: Wed Aug 05, 2020 3:42 pm
by Fred
Hi, you should always check the model and solver status.
Code: Select all
S O L V E S U M M A R Y
MODEL Rede OBJECTIVE cost
TYPE MINLP DIRECTION MINIMIZE
SOLVER SBB FROM LINE 122
**** SOLVER STATUS 4 Terminated By Solver
**** MODEL STATUS 9 Intermediate Non-Integer
**** OBJECTIVE VALUE 7.0711
Intermediate non-integer means that SBB did not find a feasible solution
Fred
Re: Binary variables
Posted: Wed Aug 05, 2020 7:34 pm
by Alexanre.ito
Yes, thanks.
Fred wrote: ↑3 years ago
Hi, you should always check the model and solver status.
Code: Select all
S O L V E S U M M A R Y
MODEL Rede OBJECTIVE cost
TYPE MINLP DIRECTION MINIMIZE
SOLVER SBB FROM LINE 122
**** SOLVER STATUS 4 Terminated By Solver
**** MODEL STATUS 9 Intermediate Non-Integer
**** OBJECTIVE VALUE 7.0711
Intermediate non-integer means that SBB did not find a feasible solution
Fred
Re: Binary variables
Posted: Fri Mar 04, 2022 5:56 pm
by Francisco.Patrocinio
Hi,
I seem to be having a similar problem: SBB reports an INTEGER SOLUTION (Fig.1), but my binary variable y has values like 1e-6 (Fig.2).
However if I run ANTIGONE(or SBB again) afterwards, it accepts the objective value of SBB, but the binaries are now all 0 and 1.
Re: Binary variables
Posted: Sat Mar 05, 2022 7:51 am
by bussieck
As all solvers SBB has a tolerance for what it considers to be integer, see
https://www.gams.com/38/docs/S_SBB.html#SBBepint. The default for epint is at 1e-5. So if abs(round(x)-x)<epint x does not need to be branched on. You can change the tolerance within a solver option file.
-Michael
Re: Binary variables
Posted: Wed Mar 09, 2022 6:03 pm
by Francisco.Patrocinio
Thanks Michael!