Extreme Ray of Sub Problem(Benders)

I am implementing benders decomposition in GAMS. I need to find the extreme ray of sub problem when it is infeasible. I don’t know how to generate extreme ray in gams directly.

I know that there exists implementation of Erwin K. for benders in gams. He directly modified the sub-problem by forcing its objective to be one. Moreover, he also modifies the constraints of sub-problem. But why?? Any help will be appreciated. Thanks in advance.

http://amsterdamoptimization.com/pdf/benders.pdf