Add components that allow description of MCTS algorithms in THTS.

Issue #21 resolved
tkeller
repo owner created an issue

MCTS algorithms (most importantly, UCT) can be modelled as THTS algorithms, and similar versions can be run in PROST, but they always differ in some details. Adding a random walk search engine and an initialization component that allows to add only a single successor to the tree should suffice to model UCT exactly in the version that is described by Kocsis and Szepesvari. (issue from 26.11.2015)

Comments (1)

  1. tkeller reporter

    The addition of a random walk search engine and a initializer that adds only a single node to the explicated tree allows to use the "original" UCT within PROST. Even though it would be nice to have a more general MCTS class that allows the specification of algorithms in terms of a tree policy (which is already possible by definition of a THTS action selection ingredient) and a default policy (which is not supported yet) I regard this issue as fixed for now since I am not aware of a default policy other than the random walk policy that has been implemented with this issue. (comment from 11.12.2015)

  2. Log in to comment