cotengra.experimental.path_compressed_mcts¶
Compressed contraction tree search using monte carlo tree search.
Classes¶
Module Contents¶
- class cotengra.experimental.path_compressed_mcts.Node(hg, nid_path, size, local_score, forward_score)¶
- __slots__ = ('hg', 'n', 'graph_key', 'nid_path', 'size', 'local_score', 'forward_score', 'mean', 'count',...¶
- hg¶
- n¶
- graph_key¶
- nid_path¶
- size¶
- local_score¶
- forward_score¶
- count = 0¶
- mean¶
- leaf_score = None¶
- update(x)¶
Report the score
x, presumably from a child node, updating this nodes score.
- __hash__()¶
- __lt__(other)¶
- __repr__()¶
- class cotengra.experimental.path_compressed_mcts.MCTS(chi, T=0.1, prune=True, optimize=None, optimize_factory=False, seed=None)¶
- chi¶
- T = 0.1¶
- prune = True¶
- optimize = None¶
- optimize_factory = False¶
- best_score¶
- best_nid_path = None¶
- children¶
- parents¶
- seen¶
- to_delete¶
- leaves = None¶
- root = None¶
- N = None¶
- gmblgen¶
- __repr__()¶
- setup(inputs, output, size_dict)¶
- get_ssa_path()¶
Convert unique node identifiers to ssa.
- check_node(node)¶
- delete_node(node)¶
- backprop(node)¶
- simulate_node(node)¶
- simulate_optimized(node)¶
- is_deadend(node)¶
- descend()¶
- property ssa_path¶
- property path¶
- run(inputs, output, size_dict)¶
- search(inputs, output, size_dict)¶
- __call__(inputs, output, size_dict)¶