Retire eval margin and gains

1/ eval margin and gains removed:
 - gains removed by Value(128): search() and qsearch() now behave consistently!

2/ futility_margin()
 - testing showed that there is no added value in this weird (log(depth), movecount)
   formula, and a much simpler linear formula is just as good. In fact, it is most
   likely better, as it is not yet optimally tuned.
 - the new simplified formula also means we get rid of FutilityMargins[], its
   initialization code, and more importantly ss->futilityMoveCount, and the hacky
   code that updates it throughout the search().
 - the current formula gives negative futility margins, and there is a hidden interaction
   between the move coutn pruning formula and the futility margin one: what happens is
   that MCP is supposed to be triggered before we use the non-sensical negative futility
   margins.

3/ unify pre & post futility pruning
 - pre futility pruning (what SF calls value based pruning) used depth < 7 plies,
   while post futility pruning (what SF calls static null move pruning) used depth < 4 plies.
 - also the condition depth < 7 in pre futility pruning was not obvious, and it seemd
   to be depth < 16 (futility_margin() returns an infinite value when depth >= 7).

Tested with fixed number of games both at short TC:
ELO: 0.82 +-2.1 (95%) LOS: 77.3%
Total: 40000 W: 7939 L: 7845 D: 24216

And long TC
ELO: 0.59 +-2.0 (95%) LOS: 71.9%
Total: 40000 W: 6876 L: 6808 D: 26316

bench: 10206576
This commit is contained in:
Lucas Braesch
2013-10-30 11:22:42 +08:00
committed by Marco Costalba
parent 52ae0efccf
commit ecd07e51d0
7 changed files with 47 additions and 112 deletions

View File

@@ -30,14 +30,13 @@
/// The Stats struct stores moves statistics. According to the template parameter
/// the class can store History, Gains and Countermoves. History records how often
/// the class can store History and Countermoves. History records how often
/// different moves have been successful or unsuccessful during the current search
/// and is used for reduction and move ordering decisions. Gains records the move's
/// best evaluation gain from one ply to the next and is used for pruning decisions.
/// and is used for reduction and move ordering decisions.
/// Countermoves store the move that refute a previous one. Entries are stored
/// according only to moving piece and destination square, hence two moves with
/// different origin but same destination and piece will be considered identical.
template<bool Gain, typename T>
template<typename T>
struct Stats {
static const Value Max = Value(2000);
@@ -56,10 +55,7 @@ struct Stats {
void update(Piece p, Square to, Value v) {
if (Gain)
table[p][to] = std::max(v, table[p][to] - 1);
else if (abs(table[p][to] + v) < Max)
if (abs(table[p][to] + v) < Max)
table[p][to] += v;
}
@@ -67,9 +63,8 @@ private:
T table[PIECE_NB][SQUARE_NB];
};
typedef Stats< true, Value> GainsStats;
typedef Stats<false, Value> HistoryStats;
typedef Stats<false, std::pair<Move, Move> > CountermovesStats;
typedef Stats<Value> HistoryStats;
typedef Stats<std::pair<Move, Move> > CountermovesStats;
/// MovePicker class is used to pick one pseudo legal move at a time from the