2 * Manual page - full usage documentation
3 * GTP interface documentation
6 * Further optimize board implementation, oprofile fun
7 * Clean up GTP interface, allow custom GTP commands for modules
8 * Implement parameter setup over GTP
9 * gogui-friendly GTP interface
14 Further use in search. Experiment with crit_min_playouts adaptively
15 adjusting based on number of simulations so far.
17 * MM local-based patterns
18 * Balanced local-based patterns?
21 * Reverse status learning
22 Run on game corpus. Start at final position, watch development
23 of status of all stones. The moment the final status and expected
24 status changes, analyze, especially if move choice differs. Use
25 learnt status-fixing moves in simulations somehow.
28 * Split playout aspects to custom-stackable pieces
29 * Possibly replay top-rated unplayed moves from the tree
33 * Fix RAVE for good (still noticeably weaker than fuego?)
34 * Separate constants for prior and amaf elements?