search

ssriram1992 / EPECsolve

Code to compute mixed-equilibrium in linear EPECs
https://ssriram1992.github.io/EPECsolve/html/index.html
GNU General Public License v3.0
6 stars 2 forks source link

Best deviation could be unbounded #10

Closed ssriram1992 closed 5 years ago

ssriram1992 commented 5 years ago

Given some feasible point during iterativeNash, the best deviation for a player could be unbounded! In that case, should change the code to get a finite feasible deviation.

gdragotto commented 5 years ago

88dc203 (a8933da) tries to recover a feasible solution from Game::EPEC::RespondSol. The GRB_UNBOUNDED proves just the existence of an extreme ray wrt the direction of the objective function. Hence, up to now, we try to recover a feasible solution by setting the objective to zero. If one exists, the deviation is given accordingly.

TO-DO Get the extreme ray and multiply the feasible point for it

gdragotto commented 5 years ago

d7c1e6b manages unbounded responses by getting monotonically improving deviations (using the extreme ray).

ssriram1992 commented 5 years ago

Awesome!