I am trying to solve a large dynamic programming problem, say with 1 million states and 10 actions.
Currently, I have a bottleneck in adding constraints with the following code:
+ gamma*quicksum(Pns(snext, s, a)*stateval[snext] for snext in stateset)
for s in stateset for a in actionset[s]
Is there a way to add constraints in parallel (using multiple CPUs)? Thank you.
Please sign in to leave a comment.