SOS1 Modelling

[Nikou]

I would like to model the following in GAMS

B(i1,j,t) + B(i2,j,t) + B(i3,j,t) =l= V , B(.,.,.) nonnegative variables and V a constant.
B(i1,j,t) * B(i2,j,t) * B(i3,j,t) =e= 0

I would like to solve the problem using SOS1 sets. The triplet B(i1,j,t), B(i2,j,t), B(i3,j,t) forms a SOS1.

How can I translate this logic in GAMS code?

[bussieck]

Read https://www.gams.com/latest/docs/UG_Lan ... Type1-SOS1. Move your i index to the last position in variable, i.e. B(j,t,i) and declare the variable as SOS1 variables. But the SOS1 constraint is not equivalent with B(i1,j,t)*B(i2,j,t)*B(i3,j,t)=e=0;.

-Michael

[Nikou]

Thank you very much!

I suppose you mean that

B(i1,j,t) * B(i2,j,t) * B(i3,j,t) =e= 0

implies that at least one of the Bs has to be zero while SOS1 means that at most one of the Bs is nonzero.
This is ok for my model.