#11
|
|||
|
|||
Re: We here at 2+2 can solve poker!
[ QUOTE ]
[ QUOTE ] If you look at my starting post, you will see that it can easily be solved in under two years if all my information is accurate, and I'm pretty sure it is. [/ QUOTE ] Let's say the search space has roughly 7,000,000,000,000 states. Then, in order to find an optimal startegy we need to do roughly 49,000,000,000,000,000,000,000,000 operations. Now, let's assume, for a moment that each operation is one flop (in practice the operations are a bit larger but that's not important here.) Let's say we have 1 terraflop = 1,000,000,000,000 operations per cpu per second, and 1000 cpu's. Then the process will still take 49,000,000,000,000 seconds. That's roughly 1,500,000 years. [/ QUOTE ] Well, I'm free for the next 78,000,000 Saturdays... [img]/images/graemlins/tongue.gif[/img] |
#13
|
|||
|
|||
Re: We here at 2+2 can solve poker!
It might be possible to solve flop poker -- that is, solve hold-em that doesn't allow bets after the flop round.
I come up with about 3.1 million hands for each player. So roughly 58 million variables in a complete post-flop strategy -- that's small enough to fit into memory. Optimizing that is still roughly order 2^45, but that's computable, if large. It's worth noting that if the vast majority of hands end up being terminal i.e. folded by one of the players, then this will be a good approximation of the optimal strategy. |
#14
|
|||
|
|||
Re: We here at 2+2 can solve poker!
The OP had the right idea... distributed computing would be the way to go for any problem of this nature... set it up, and i'll let you have all my idle time!
|
|
|