uct/walk.c

   1 #include <assert.h>
   2 #include <pthread.h>
   3 #include <signal.h>
   4 #include <stdio.h>
   5 #include <stdlib.h>
   6 #include <string.h>
   7
   8 #define DEBUG
   9
  10 #include "debug.h"
  11 #include "board.h"
  12 #include "move.h"
  13 #include "playout.h"
  14 #include "probdist.h"
  15 #include "random.h"
  16 #include "tactics.h"
  17 #include "uct/internal.h"
  18 #include "uct/tree.h"
  19 #include "uct/uct.h"
  20 #include "uct/walk.h"
  21
  22 float
  23 uct_get_extra_komi(struct uct *u, struct board *b)
  24 {
  25         float extra_komi = board_effective_handicap(b) * (u->dynkomi - b->moves) / u->dynkomi;
  26         return extra_komi;
  27 }
  28
  29 void
  30 uct_progress_status(struct uct *u, struct tree *t, enum stone color, int playouts)
  31 {
  32         if (!UDEBUGL(0))
  33                 return;
  34
  35         /* Best move */
  36         struct tree_node *best = u->policy->choose(u->policy, t->root, t->board, color);
  37         if (!best) {
  38                 fprintf(stderr, "... No moves left\n");
  39                 return;
  40         }
  41         fprintf(stderr, "[%d] ", playouts);
  42         fprintf(stderr, "best %f ", tree_node_get_value(t, 1, best->u.value));
  43
  44         /* Max depth */
  45         fprintf(stderr, "deepest % 2d ", t->max_depth - t->root->depth);
  46
  47         /* Best sequence */
  48         fprintf(stderr, "| seq ");
  49         for (int depth = 0; depth < 6; depth++) {
  50                 if (best && best->u.playouts >= 25) {
  51                         fprintf(stderr, "%3s ", coord2sstr(best->coord, t->board));
  52                         best = u->policy->choose(u->policy, best, t->board, color);
  53                 } else {
  54                         fprintf(stderr, "    ");
  55                 }
  56         }
  57
  58         /* Best candidates */
  59         fprintf(stderr, "| can ");
  60         int cans = 4;
  61         struct tree_node *can[cans];
  62         memset(can, 0, sizeof(can));
  63         best = t->root->children;
  64         while (best) {
  65                 int c = 0;
  66                 while ((!can[c] || best->u.playouts > can[c]->u.playouts) && ++c < cans);
  67                 for (int d = 0; d < c; d++) can[d] = can[d + 1];
  68                 if (c > 0) can[c - 1] = best;
  69                 best = best->sibling;
  70         }
  71         while (--cans >= 0) {
  72                 if (can[cans]) {
  73                         fprintf(stderr, "%3s(%.3f) ",
  74                                 coord2sstr(can[cans]->coord, t->board),
  75                                 tree_node_get_value(t, 1, can[cans]->u.value));
  76                 } else {
  77                         fprintf(stderr, "           ");
  78                 }
  79         }
  80
  81         fprintf(stderr, "\n");
  82 }
  83
  84
  85 static int
  86 uct_leaf_node(struct uct *u, struct board *b, enum stone player_color,
  87               struct playout_amafmap *amaf,
  88               struct tree *t, struct tree_node *n, enum stone node_color,
  89               char *spaces)
  90 {
  91         enum stone next_color = stone_other(node_color);
  92         int parity = (next_color == player_color ? 1 : -1);
  93
  94         /* If we don't anticipate well the opponent move during pondering
  95          * (the played move has few playouts) we still need more memory
  96          * during genmove to explore the tree actually played.
  97          * For fast_alloc, the tree compaction will free enough memory
  98          * immediately. */
  99         unsigned long max_tree_size = u->max_tree_size;
 100         if (u->pondering && !u->fast_alloc)
 101                 max_tree_size = (max_tree_size * (100 - MIN_FREE_MEM_PERCENT)) / 100;
 102
 103         /* We need to make sure only one thread expands the node. If
 104          * we are unlucky enough for two threads to meet in the same
 105          * node, the latter one will simply do another simulation from
 106          * the node itself, no big deal. t->nodes_size may exceed
 107          * the maximum in multi-threaded case but not by much so it's ok.
 108          * The size test must be before the test&set not after, to allow
 109          * expansion of the node later if enough nodes have been freed. */
 110         if (n->u.playouts >= u->expand_p && t->nodes_size < max_tree_size
 111             && !__sync_lock_test_and_set(&n->is_expanded, 1)) {
 112                 tree_expand_node(t, n, b, next_color, u, parity);
 113         }
 114         if (UDEBUGL(7))
 115                 fprintf(stderr, "%s*-- UCT playout #%d start [%s] %f\n",
 116                         spaces, n->u.playouts, coord2sstr(n->coord, t->board),
 117                         tree_node_get_value(t, parity, n->u.value));
 118
 119         struct playout_setup ps = { .gamelen = u->gamelen, .mercymin = u->mercymin };
 120         int result = play_random_game(&ps, b, next_color,
 121                                       u->playout_amaf ? amaf : NULL,
 122                                       &u->ownermap, u->playout);
 123         if (next_color == S_WHITE) {
 124                 /* We need the result from black's perspective. */
 125                 result = - result;
 126         }
 127         if (UDEBUGL(7))
 128                 fprintf(stderr, "%s -- [%d..%d] %s random playout result %d\n",
 129                         spaces, player_color, next_color, coord2sstr(n->coord, t->board), result);
 130
 131         return result;
 132 }
 133
 134 static float
 135 scale_value(struct uct *u, struct board *b, int result)
 136 {
 137         float rval = result > 0;
 138         if (u->val_scale) {
 139                 int vp = u->val_points;
 140                 if (!vp) {
 141                         vp = board_size(b) - 1; vp *= vp; vp *= 2;
 142                 }
 143
 144                 float sval = (float) abs(result) / vp;
 145                 sval = sval > 1 ? 1 : sval;
 146                 if (result < 0) sval = 1 - sval;
 147                 if (u->val_extra)
 148                         rval += u->val_scale * sval;
 149                 else
 150                         rval = (1 - u->val_scale) * rval + u->val_scale * sval;
 151                 // fprintf(stderr, "score %d => sval %f, rval %f\n", result, sval, rval);
 152         }
 153         return rval;
 154 }
 155
 156
 157 int
 158 uct_playout(struct uct *u, struct board *b, enum stone player_color, struct tree *t)
 159 {
 160         struct board b2;
 161         board_copy(&b2, b);
 162
 163         struct playout_amafmap *amaf = NULL;
 164         if (u->policy->wants_amaf) {
 165                 amaf = calloc(1, sizeof(*amaf));
 166                 amaf->map = calloc(board_size2(&b2) + 1, sizeof(*amaf->map));
 167                 amaf->map++; // -1 is pass
 168         }
 169
 170         /* Walk the tree until we find a leaf, then expand it and do
 171          * a random playout. */
 172         struct tree_node *n = t->root;
 173         enum stone node_color = stone_other(player_color);
 174         assert(node_color == t->root_color);
 175
 176         void *dstate = NULL, *dstater = NULL;
 177
 178         int result;
 179         int pass_limit = (board_size(&b2) - 2) * (board_size(&b2) - 2) / 2;
 180         int passes = is_pass(b->last_move.coord) && b->moves > 0;
 181
 182         /* debug */
 183         int depth = 0;
 184         static char spaces[] = "\0                                                      ";
 185         /* /debug */
 186         if (UDEBUGL(8))
 187                 fprintf(stderr, "--- UCT walk with color %d\n", player_color);
 188
 189         while (!tree_leaf_node(n) && passes < 2) {
 190                 spaces[depth++] = ' '; spaces[depth] = 0;
 191
 192                 /* Parity is chosen already according to the child color, since
 193                  * it is applied to children. */
 194                 node_color = stone_other(node_color);
 195                 int parity = (node_color == player_color ? 1 : -1);
 196                 n = (!u->random_policy_chance || fast_random(u->random_policy_chance))
 197                         ? u->policy->descend(u->policy, &dstate, t, n, parity, b2.moves > pass_limit)
 198                         : u->random_policy->descend(u->random_policy, &dstater, t, n, parity, b2.moves > pass_limit);
 199
 200                 assert(n == t->root || n->parent);
 201                 if (UDEBUGL(7))
 202                         fprintf(stderr, "%s+-- UCT sent us to [%s:%d] %f\n",
 203                                 spaces, coord2sstr(n->coord, t->board), n->coord,
 204                                 tree_node_get_value(t, parity, n->u.value));
 205
 206                 /* Add virtual loss if we need to; this is used to discourage
 207                  * other threads from visiting this node in case of multiple
 208                  * threads doing the tree search. */
 209                 if (u->virtual_loss)
 210                         stats_add_result(&n->u, tree_parity(t, parity) > 0 ? 0 : 1, 1);
 211
 212                 assert(n->coord >= -1);
 213                 if (amaf && !is_pass(n->coord)) {
 214                         if (amaf->map[n->coord] == S_NONE || amaf->map[n->coord] == node_color) {
 215                                 amaf->map[n->coord] = node_color;
 216                         } else { // XXX: Respect amaf->record_nakade
 217                                 amaf_op(amaf->map[n->coord], +);
 218                         }
 219                         amaf->game[amaf->gamelen].coord = n->coord;
 220                         amaf->game[amaf->gamelen].color = node_color;
 221                         amaf->gamelen++;
 222                         assert(amaf->gamelen < sizeof(amaf->game) / sizeof(amaf->game[0]));
 223                 }
 224
 225                 struct move m = { n->coord, node_color };
 226                 int res = board_play(&b2, &m);
 227
 228                 if (res < 0 || (!is_pass(m.coord) && !group_at(&b2, m.coord)) /* suicide */
 229                     || b2.superko_violation) {
 230                         if (UDEBUGL(3)) {
 231                                 for (struct tree_node *ni = n; ni; ni = ni->parent)
 232                                         fprintf(stderr, "%s<%"PRIhash"> ", coord2sstr(ni->coord, t->board), ni->hash);
 233                                 fprintf(stderr, "marking invalid %s node %d,%d res %d group %d spk %d\n",
 234                                         stone2str(node_color), coord_x(n->coord,b), coord_y(n->coord,b),
 235                                         res, group_at(&b2, m.coord), b2.superko_violation);
 236                         }
 237                         n->hints |= TREE_HINT_INVALID;
 238                         result = 0;
 239                         goto end;
 240                 }
 241
 242                 if (is_pass(n->coord))
 243                         passes++;
 244                 else
 245                         passes = 0;
 246         }
 247
 248         if (amaf) {
 249                 amaf->game_baselen = amaf->gamelen;
 250                 amaf->record_nakade = u->playout_amaf_nakade;
 251         }
 252
 253         if (u->dynkomi > b2.moves && (player_color & u->dynkomi_mask))
 254                 b2.komi += uct_get_extra_komi(u, &b2);
 255
 256         if (passes >= 2) {
 257                 /* XXX: No dead groups support. */
 258                 float score = board_official_score(&b2, NULL);
 259                 /* Result from black's perspective (no matter who
 260                  * the player; black's perspective is always
 261                  * what the tree stores. */
 262                 result = - (score * 2);
 263
 264                 if (UDEBUGL(5))
 265                         fprintf(stderr, "[%d..%d] %s p-p scoring playout result %d (W %f)\n",
 266                                 player_color, node_color, coord2sstr(n->coord, t->board), result, score);
 267                 if (UDEBUGL(6))
 268                         board_print(&b2, stderr);
 269
 270                 board_ownermap_fill(&u->ownermap, &b2);
 271
 272         } else { assert(u->parallel_tree || tree_leaf_node(n));
 273                 /* In case of parallel tree search, the assertion might
 274                  * not hold if two threads chew on the same node. */
 275                 result = uct_leaf_node(u, &b2, player_color, amaf, t, n, node_color, spaces);
 276         }
 277
 278         if (amaf && u->playout_amaf_cutoff) {
 279                 int cutoff = amaf->game_baselen;
 280                 cutoff += (amaf->gamelen - amaf->game_baselen) * u->playout_amaf_cutoff / 100;
 281                 /* Now, reconstruct the amaf map. */
 282                 memset(amaf->map, 0, board_size2(&b2) * sizeof(*amaf->map));
 283                 for (int i = 0; i < cutoff; i++) {
 284                         coord_t coord = amaf->game[i].coord;
 285                         enum stone color = amaf->game[i].color;
 286                         if (amaf->map[coord] == S_NONE || amaf->map[coord] == color) {
 287                                 amaf->map[coord] = color;
 288                         /* Nakade always recorded for in-tree part */
 289                         } else if (amaf->record_nakade || i <= amaf->game_baselen) {
 290                                 amaf_op(amaf->map[n->coord], +);
 291                         }
 292                 }
 293         }
 294
 295         assert(n == t->root || n->parent);
 296         if (result != 0) {
 297                 float rval = scale_value(u, b, result);
 298
 299                 u->policy->update(u->policy, t, n, node_color, player_color, amaf, rval);
 300
 301                 if (u->root_heuristic && n->parent) {
 302                         if (!t->chvals) {
 303                                 t->chvals = calloc(board_size2(b), sizeof(t->chvals[0]));
 304                                 t->chchvals = calloc(board_size2(b), sizeof(t->chchvals[0]));
 305                         }
 306
 307                         /* Possibly transform the rval appropriately. */
 308                         rval = stats_temper_value(rval, n->parent->u.value, u->root_heuristic);
 309
 310                         struct tree_node *ni = n;
 311                         while (ni->parent->parent && ni->parent->parent->parent)
 312                                 ni = ni->parent;
 313                         if (ni->parent->parent) {
 314                                 if (likely(!is_pass(ni->coord)))
 315                                         stats_add_result(&t->chchvals[ni->coord], rval, 1);
 316                                 ni = ni->parent;
 317                         }
 318                         assert(ni->parent && !ni->parent->parent);
 319
 320                         if (likely(!is_pass(ni->coord)))
 321                                 stats_add_result(&t->chvals[ni->coord], rval, 1);
 322                 }
 323         }
 324
 325 end:
 326         /* We need to undo the virtual loss we added during descend. */
 327         if (u->virtual_loss) {
 328                 int parity = (node_color == player_color ? 1 : -1);
 329                 for (; n->parent; n = n->parent) {
 330                         stats_rm_result(&n->u, tree_parity(t, parity) > 0 ? 0 : 1, 1);
 331                         parity = -parity;
 332                 }
 333         }
 334
 335         if (dstater) free(dstater);
 336         if (dstate) free(dstate);
 337         if (amaf) {
 338                 free(amaf->map - 1);
 339                 free(amaf);
 340         }
 341         board_done_noalloc(&b2);
 342         return result;
 343 }
 344
 345 int
 346 uct_playouts(struct uct *u, struct board *b, enum stone color, struct tree *t)
 347 {
 348         int i;
 349         for (i = 0; !uct_halt; i++)
 350                 uct_playout(u, b, color, t);
 351         return i;
 352 }