Marc Lanctot
Marc Lanctot
Hi @95A31, We don't have too many single-agent games in OpenSpiel but there are a few (morpion solitaire, some of the gridworld domains, klondike solitaire) but they're all more complex...
Closing due to inactivity, please re-open if you want to follow up.
Hi @bwr125, Our sequence-form LP implementations don't support any form of subgame perfection. That would be a welcome contribution, though! What I don't understand from the page you linked from...
Ok, quick follow-up: On their Gambit page for the LP method they say it's based on Koller, Megiddo, and von Stengel '94.. which doesn't have any form of subgame perfection...
> Is this open I would love to work on this. It is, I guess? You are welcome to work on it.. but I remain confused about what "it" is...
Can you describe what you would intend to do, because I still don't understand the request. Subgame perfection in general EFGs (i.e. with imperfect information) is non-trivial and there have...
There are some papers I can point you to that contain LP methods for computing various forms of perfect equilibria using (sequence-form) LPs. But they do tend to get complicated....
> Now i Don't know if this what the creator of this issue wanted to solve, but this is what I could grasp. I couldn't clearly understand even after following...
Sounds great, thanks!!
Hi @demoncoder-crypto just checking in.. are you still planning to contribute your implementation of backward induction?