Open richa-varma opened 2 years ago
You should just specify the smaller action space as the action space for the POMDP that POMCP is solving.
Note that this action space can be belief-dependent, since BasicPOMCP always uses actions(m::YourPOMDP, b)
to get the list of actions that it will consider at each node.
Note that the POMDP that you are simulating does not have to be the same as the one you are using with the POMCP solver, so you can create one version of your POMDP with a full action space, and one version with a limited action space to give to POMCP.
@zsunberg While looking at ways to reduce the size of the action space in my model during computation, I came across the
POMCP.sparse_actions
function in the documentation of POMDPs.jl, which I understand has been deprecated. I was wondering if there was some rationale behind not providing this function in BasicPOMCP.jl, and if there is another way to sample a subset of actions to be considered at the current node during planning.