uct/walk.c

   1 #include <assert.h>
   2 #include <math.h>
   3 #include <pthread.h>
   4 #include <signal.h>
   5 #include <stdio.h>
   6 #include <stdlib.h>
   7 #include <string.h>
   8
   9 #define DEBUG
  10
  11 #include "debug.h"
  12 #include "board.h"
  13 #include "move.h"
  14 #include "playout.h"
  15 #include "probdist.h"
  16 #include "random.h"
  17 #include "uct/dynkomi.h"
  18 #include "uct/internal.h"
  19 #include "uct/search.h"
  20 #include "uct/tree.h"
  21 #include "uct/uct.h"
  22 #include "uct/walk.h"
  23
  24 #define DESCENT_DLEN 512
  25
  26 void
  27 uct_progress_status(struct uct *u, struct tree *t, enum stone color, int playouts)
  28 {
  29         if (!UDEBUGL(0))
  30                 return;
  31
  32         /* Best move */
  33         struct tree_node *best = u->policy->choose(u->policy, t->root, t->board, color, resign);
  34         if (!best) {
  35                 fprintf(stderr, "... No moves left\n");
  36                 return;
  37         }
  38         fprintf(stderr, "[%d] ", playouts);
  39         fprintf(stderr, "best %f ", tree_node_get_value(t, 1, best->u.value));
  40
  41         /* Dynamic komi */
  42         if (t->use_extra_komi)
  43                 fprintf(stderr, "komi %.1f ", t->extra_komi);
  44
  45         /* Best sequence */
  46         fprintf(stderr, "| seq ");
  47         for (int depth = 0; depth < 4; depth++) {
  48                 if (best && best->u.playouts >= 25) {
  49                         fprintf(stderr, "%3s ", coord2sstr(best->coord, t->board));
  50                         best = u->policy->choose(u->policy, best, t->board, color, resign);
  51                 } else {
  52                         fprintf(stderr, "    ");
  53                 }
  54         }
  55
  56         /* Best candidates */
  57         fprintf(stderr, "| can ");
  58         int cans = 4;
  59         struct tree_node *can[cans];
  60         memset(can, 0, sizeof(can));
  61         best = t->root->children;
  62         while (best) {
  63                 int c = 0;
  64                 while ((!can[c] || best->u.playouts > can[c]->u.playouts) && ++c < cans);
  65                 for (int d = 0; d < c; d++) can[d] = can[d + 1];
  66                 if (c > 0) can[c - 1] = best;
  67                 best = best->sibling;
  68         }
  69         while (--cans >= 0) {
  70                 if (can[cans]) {
  71                         fprintf(stderr, "%3s(%.3f) ",
  72                                 coord2sstr(can[cans]->coord, t->board),
  73                                 tree_node_get_value(t, 1, can[cans]->u.value));
  74                 } else {
  75                         fprintf(stderr, "           ");
  76                 }
  77         }
  78
  79         fprintf(stderr, "\n");
  80 }
  81
  82
  83 static void
  84 record_amaf_move(struct playout_amafmap *amaf, coord_t coord, enum stone color)
  85 {
  86         if (amaf->map[coord] == S_NONE || amaf->map[coord] == color) {
  87                 amaf->map[coord] = color;
  88         } else { // XXX: Respect amaf->record_nakade
  89                 amaf_op(amaf->map[coord], +);
  90         }
  91         amaf->game[amaf->gamelen].coord = coord;
  92         amaf->game[amaf->gamelen].color = color;
  93         amaf->gamelen++;
  94         assert(amaf->gamelen < sizeof(amaf->game) / sizeof(amaf->game[0]));
  95 }
  96
  97
  98 struct uct_playout_callback {
  99         struct uct *uct;
 100         struct tree *tree;
 101         struct tree_node *lnode;
 102 };
 103
 104
 105 static coord_t
 106 uct_playout_hook(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color, int mode)
 107 {
 108         /* XXX: This is used in some non-master branches. */
 109         return pass;
 110 }
 111
 112 static coord_t
 113 uct_playout_prepolicy(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color)
 114 {
 115         return uct_playout_hook(playout, setup, b, color, 0);
 116 }
 117
 118 static coord_t
 119 uct_playout_postpolicy(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color)
 120 {
 121         return uct_playout_hook(playout, setup, b, color, 1);
 122 }
 123
 124
 125 static int
 126 uct_leaf_node(struct uct *u, struct board *b, enum stone player_color,
 127               struct playout_amafmap *amaf,
 128               struct uct_descent *descent, int *dlen,
 129               struct tree_node *significant[2],
 130               struct tree *t, struct tree_node *n, enum stone node_color,
 131               char *spaces)
 132 {
 133         enum stone next_color = stone_other(node_color);
 134         int parity = (next_color == player_color ? 1 : -1);
 135
 136         /* We need to make sure only one thread expands the node. If
 137          * we are unlucky enough for two threads to meet in the same
 138          * node, the latter one will simply do another simulation from
 139          * the node itself, no big deal. t->nodes_size may exceed
 140          * the maximum in multi-threaded case but not by much so it's ok.
 141          * The size test must be before the test&set not after, to allow
 142          * expansion of the node later if enough nodes have been freed. */
 143         if (n->u.playouts >= u->expand_p && t->nodes_size < u->max_tree_size
 144             && !__sync_lock_test_and_set(&n->is_expanded, 1)) {
 145                 tree_expand_node(t, n, b, next_color, u, parity);
 146         }
 147         if (UDEBUGL(7))
 148                 fprintf(stderr, "%s*-- UCT playout #%d start [%s] %f\n",
 149                         spaces, n->u.playouts, coord2sstr(n->coord, t->board),
 150                         tree_node_get_value(t, parity, n->u.value));
 151
 152         struct uct_playout_callback upc = {
 153                 .uct = u,
 154                 .tree = t,
 155                 /* TODO: Don't necessarily restart the sequence walk when
 156                  * entering playout. */
 157                 .lnode = NULL,
 158         };
 159
 160         struct playout_setup ps = {
 161                 .gamelen = u->gamelen,
 162                 .mercymin = u->mercymin,
 163                 .prepolicy_hook = uct_playout_prepolicy,
 164                 .postpolicy_hook = uct_playout_postpolicy,
 165                 .hook_data = &upc,
 166         };
 167         int result = play_random_game(&ps, b, next_color,
 168                                       u->playout_amaf ? amaf : NULL,
 169                                       &u->ownermap, u->playout);
 170         if (next_color == S_WHITE) {
 171                 /* We need the result from black's perspective. */
 172                 result = - result;
 173         }
 174         if (UDEBUGL(7))
 175                 fprintf(stderr, "%s -- [%d..%d] %s random playout result %d\n",
 176                         spaces, player_color, next_color, coord2sstr(n->coord, t->board), result);
 177
 178         return result;
 179 }
 180
 181 static floating_t
 182 scale_value(struct uct *u, struct board *b, int result)
 183 {
 184         if (result == 0) return 0.5;
 185         floating_t rval = result > 0 ? 1.0 : 0.0;
 186
 187         floating_t scale = u->val_scale;
 188         /* Give more weight to territory when winning big (maximize win). This reduces
 189          * the number of silly moves and makes the game more enjoyable for humans. */
 190         if (u->t->root->u.playouts > GJ_MINGAMES &&
 191             tree_node_get_value(u->t, -1, u->t->root->u.value) >= u->sure_win_threshold) {
 192                 scale = u->val_scale_max;
 193         }
 194         if (scale == 0) return rval;
 195
 196         int vp = u->val_points;
 197         /* By default do not try to win by more than 44 points on 19x19,
 198          * 12 points on 9x9. Remember that result here is twice the score. */
 199         if (!vp) vp = board_size2(b) / 5;
 200
 201         floating_t sval = (floating_t) abs(result) / vp;
 202         sval = sval > 1 ? 1 : sval;
 203         if (result < 0) sval = 1 - sval;
 204         if (u->val_extra)
 205                 rval += scale * sval;
 206         else
 207                 rval = (1 - scale) * rval + scale * sval;
 208         // fprintf(stderr, "score %d => sval %f, rval %f\n", result, sval, rval);
 209         return rval;
 210 }
 211
 212 static double
 213 local_value(struct uct *u, struct board *b, coord_t coord, enum stone color)
 214 {
 215         /* Tactical evaluation of move @coord by color @color, given
 216          * simulation end position @b. I.e., a move is tactically good
 217          * if the resulting group stays on board until the game end. */
 218         /* We can also take into account surrounding stones, e.g. to
 219          * encourage taking off external liberties during a semeai. */
 220         double val;
 221         if (u->local_tree_neival) {
 222                 int friends = neighbor_count_at(b, coord, color) + neighbor_count_at(b, coord, S_OFFBOARD);
 223                 if (immediate_liberty_count(b, coord) > 0) {
 224                         foreach_neighbor(b, coord, {
 225                                 friends += board_is_one_point_eye(b, coord, color);
 226                         });
 227                 }
 228                 val = (double) (2 * (board_at(b, coord) == color) + friends) / 6.f;
 229         } else {
 230                 val = (board_at(b, coord) == color) ? 1.f : 0.f;
 231         }
 232         return (color == S_WHITE) ? 1.f - val : val;
 233 }
 234
 235 static void
 236 record_local_sequence(struct uct *u, struct tree *t, struct board *endb,
 237                       struct uct_descent *descent, int dlen, int di,
 238                       enum stone seq_color)
 239 {
 240 #define LTREE_DEBUG if (UDEBUGL(6))
 241
 242         /* Ignore pass sequences. */
 243         if (is_pass(descent[di].node->coord))
 244                 return;
 245
 246         LTREE_DEBUG board_print(endb, stderr);
 247         LTREE_DEBUG fprintf(stderr, "recording local %s sequence: ",
 248                 stone2str(seq_color));
 249
 250         /* Sequences starting deeper are less relevant in general. */
 251         int pval = LTREE_PLAYOUTS_MULTIPLIER;
 252         if (u->local_tree && u->local_tree_depth_decay > 0)
 253                 pval = ((floating_t) pval) / pow(u->local_tree_depth_decay, di - 1);
 254         if (!pval) {
 255                 LTREE_DEBUG fprintf(stderr, "too deep @%d\n", di);
 256                 return;
 257         }
 258
 259         /* Pick the right local tree root... */
 260         struct tree_node *lnode = seq_color == S_BLACK ? t->ltree_black : t->ltree_white;
 261         lnode->u.playouts++;
 262
 263         double sval = 0.5;
 264         if (u->local_tree_rootgoal) {
 265                 sval = local_value(u, endb, descent[di].node->coord, seq_color);
 266                 LTREE_DEBUG fprintf(stderr, "(goal %s[%s %1.3f][%d]) ",
 267                         coord2sstr(descent[di].node->coord, t->board),
 268                         stone2str(seq_color), sval, descent[di].node->d);
 269         }
 270
 271         /* ...and record the sequence. */
 272         int di0 = di;
 273         while (di < dlen && (di == di0 || descent[di].node->d < u->tenuki_d)) {
 274                 enum stone color = (di - di0) % 2 ? stone_other(seq_color) : seq_color;
 275                 double rval;
 276                 if (u->local_tree_rootgoal)
 277                         rval = sval;
 278                 else
 279                         rval = local_value(u, endb, descent[di].node->coord, color);
 280                 LTREE_DEBUG fprintf(stderr, "%s[%s %1.3f][%d] ",
 281                         coord2sstr(descent[di].node->coord, t->board),
 282                         stone2str(color), rval, descent[di].node->d);
 283                 lnode = tree_get_node(t, lnode, descent[di++].node->coord, true);
 284                 assert(lnode);
 285                 stats_add_result(&lnode->u, rval, pval);
 286         }
 287
 288         /* Add lnode for tenuki (pass) if we descended further. */
 289         if (di < dlen) {
 290                 double rval = u->local_tree_rootgoal ? sval : 0.5;
 291                 LTREE_DEBUG fprintf(stderr, "pass ");
 292                 lnode = tree_get_node(t, lnode, pass, true);
 293                 assert(lnode);
 294                 stats_add_result(&lnode->u, rval, pval);
 295         }
 296
 297         LTREE_DEBUG fprintf(stderr, "\n");
 298 }
 299
 300
 301 int
 302 uct_playout(struct uct *u, struct board *b, enum stone player_color, struct tree *t)
 303 {
 304         struct board b2;
 305         board_copy(&b2, b);
 306
 307         struct playout_amafmap *amaf = NULL;
 308         if (u->policy->wants_amaf) {
 309                 amaf = calloc2(1, sizeof(*amaf));
 310                 amaf->map = calloc2(board_size2(&b2) + 1, sizeof(*amaf->map));
 311                 amaf->map++; // -1 is pass
 312         }
 313
 314         /* Walk the tree until we find a leaf, then expand it and do
 315          * a random playout. */
 316         struct tree_node *n = t->root;
 317         enum stone node_color = stone_other(player_color);
 318         assert(node_color == t->root_color);
 319
 320         /* Tree descent history. */
 321         /* XXX: This is somewhat messy since @n and descent[dlen-1].node are
 322          * redundant. */
 323         struct uct_descent descent[DESCENT_DLEN];
 324         descent[0].node = n; descent[0].lnode = NULL;
 325         int dlen = 1;
 326         /* Total value of the sequence. */
 327         struct move_stats seq_value = { .playouts = 0 };
 328         /* The last "significant" node along the descent (i.e. node
 329          * with higher than configured number of playouts). For black
 330          * and white. */
 331         struct tree_node *significant[2] = { NULL, NULL };
 332         if (n->u.playouts >= u->significant_threshold)
 333                 significant[node_color - 1] = n;
 334
 335         int result;
 336         int pass_limit = (board_size(&b2) - 2) * (board_size(&b2) - 2) / 2;
 337         int passes = is_pass(b->last_move.coord) && b->moves > 0;
 338
 339         /* debug */
 340         static char spaces[] = "\0                                                      ";
 341         /* /debug */
 342         if (UDEBUGL(8))
 343                 fprintf(stderr, "--- UCT walk with color %d\n", player_color);
 344
 345         while (!tree_leaf_node(n) && passes < 2) {
 346                 spaces[dlen - 1] = ' '; spaces[dlen] = 0;
 347
 348
 349                 /*** Choose a node to descend to: */
 350
 351                 /* Parity is chosen already according to the child color, since
 352                  * it is applied to children. */
 353                 node_color = stone_other(node_color);
 354                 int parity = (node_color == player_color ? 1 : -1);
 355
 356                 assert(dlen < DESCENT_DLEN);
 357                 descent[dlen] = descent[dlen - 1];
 358                 if (u->local_tree && (!descent[dlen].lnode || descent[dlen].node->d >= u->tenuki_d)) {
 359                         /* Start new local sequence. */
 360                         /* Remember that node_color already holds color of the
 361                          * to-be-found child. */
 362                         descent[dlen].lnode = node_color == S_BLACK ? t->ltree_black : t->ltree_white;
 363                 }
 364
 365                 if (!u->random_policy_chance || fast_random(u->random_policy_chance))
 366                         u->policy->descend(u->policy, t, &descent[dlen], parity, b2.moves > pass_limit);
 367                 else
 368                         u->random_policy->descend(u->random_policy, t, &descent[dlen], parity, b2.moves > pass_limit);
 369
 370
 371                 /*** Perform the descent: */
 372
 373                 if (descent[dlen].node->u.playouts >= u->significant_threshold) {
 374                         significant[node_color - 1] = descent[dlen].node;
 375                 }
 376
 377                 seq_value.playouts += descent[dlen].value.playouts;
 378                 seq_value.value += descent[dlen].value.value * descent[dlen].value.playouts;
 379                 n = descent[dlen++].node;
 380                 assert(n == t->root || n->parent);
 381                 if (UDEBUGL(7))
 382                         fprintf(stderr, "%s+-- UCT sent us to [%s:%d] %d,%f\n",
 383                                 spaces, coord2sstr(n->coord, t->board),
 384                                 n->coord, n->u.playouts,
 385                                 tree_node_get_value(t, parity, n->u.value));
 386
 387                 /* Add virtual loss if we need to; this is used to discourage
 388                  * other threads from visiting this node in case of multiple
 389                  * threads doing the tree search. */
 390                 if (u->virtual_loss)
 391                         stats_add_result(&n->u, node_color == S_BLACK ? 0.0 : 1.0, u->virtual_loss);
 392
 393                 assert(n->coord >= -1);
 394                 if (amaf && !is_pass(n->coord))
 395                         record_amaf_move(amaf, n->coord, node_color);
 396
 397                 struct move m = { n->coord, node_color };
 398                 int res = board_play(&b2, &m);
 399
 400                 if (res < 0 || (!is_pass(m.coord) && !group_at(&b2, m.coord)) /* suicide */
 401                     || b2.superko_violation) {
 402                         if (UDEBUGL(4)) {
 403                                 for (struct tree_node *ni = n; ni; ni = ni->parent)
 404                                         fprintf(stderr, "%s<%"PRIhash"> ", coord2sstr(ni->coord, t->board), ni->hash);
 405                                 fprintf(stderr, "marking invalid %s node %d,%d res %d group %d spk %d\n",
 406                                         stone2str(node_color), coord_x(n->coord,b), coord_y(n->coord,b),
 407                                         res, group_at(&b2, m.coord), b2.superko_violation);
 408                         }
 409                         n->hints |= TREE_HINT_INVALID;
 410                         result = 0;
 411                         goto end;
 412                 }
 413
 414                 if (is_pass(n->coord))
 415                         passes++;
 416                 else
 417                         passes = 0;
 418         }
 419
 420         if (amaf) {
 421                 amaf->game_baselen = amaf->gamelen;
 422                 amaf->record_nakade = u->playout_amaf_nakade;
 423         }
 424
 425         if (t->use_extra_komi && u->dynkomi->persim) {
 426                 b2.komi += round(u->dynkomi->persim(u->dynkomi, &b2, t, n));
 427         }
 428
 429         if (passes >= 2) {
 430                 /* XXX: No dead groups support. */
 431                 floating_t score = board_official_score(&b2, NULL);
 432                 /* Result from black's perspective (no matter who
 433                  * the player; black's perspective is always
 434                  * what the tree stores. */
 435                 result = - (score * 2);
 436
 437                 if (UDEBUGL(5))
 438                         fprintf(stderr, "[%d..%d] %s p-p scoring playout result %d (W %f)\n",
 439                                 player_color, node_color, coord2sstr(n->coord, t->board), result, score);
 440                 if (UDEBUGL(6))
 441                         board_print(&b2, stderr);
 442
 443                 board_ownermap_fill(&u->ownermap, &b2);
 444
 445         } else { // assert(tree_leaf_node(n));
 446                 /* In case of parallel tree search, the assertion might
 447                  * not hold if two threads chew on the same node. */
 448                 result = uct_leaf_node(u, &b2, player_color, amaf, descent, &dlen, significant, t, n, node_color, spaces);
 449         }
 450
 451         if (amaf && u->playout_amaf_cutoff) {
 452                 unsigned int cutoff = amaf->game_baselen;
 453                 cutoff += (amaf->gamelen - amaf->game_baselen) * u->playout_amaf_cutoff / 100;
 454                 /* Now, reconstruct the amaf map. */
 455                 memset(amaf->map, 0, board_size2(&b2) * sizeof(*amaf->map));
 456                 for (unsigned int i = 0; i < cutoff; i++) {
 457                         coord_t coord = amaf->game[i].coord;
 458                         enum stone color = amaf->game[i].color;
 459                         if (amaf->map[coord] == S_NONE || amaf->map[coord] == color) {
 460                                 amaf->map[coord] = color;
 461                         /* Nakade always recorded for in-tree part */
 462                         } else if (amaf->record_nakade || i <= amaf->game_baselen) {
 463                                 amaf_op(amaf->map[n->coord], +);
 464                         }
 465                 }
 466         }
 467
 468         /* Record the result. */
 469
 470         assert(n == t->root || n->parent);
 471         floating_t rval = scale_value(u, b, result);
 472         u->policy->update(u->policy, t, n, node_color, player_color, amaf, &b2, rval);
 473
 474         if (t->use_extra_komi) {
 475                 stats_add_result(&u->dynkomi->score, result / 2, 1);
 476                 stats_add_result(&u->dynkomi->value, rval, 1);
 477         }
 478
 479         if (u->local_tree && n->parent && !is_pass(n->coord) && dlen > 0) {
 480                 /* Get the local sequences and record them in ltree. */
 481                 /* We will look for sequence starts in our descent
 482                  * history, then run record_local_sequence() for each
 483                  * found sequence start; record_local_sequence() may
 484                  * pick longer sequences from descent history then,
 485                  * which is expected as it will create new lnodes. */
 486                 enum stone seq_color = player_color;
 487                 /* First move always starts a sequence. */
 488                 record_local_sequence(u, t, &b2, descent, dlen, 1, seq_color);
 489                 seq_color = stone_other(seq_color);
 490                 for (int dseqi = 2; dseqi < dlen; dseqi++, seq_color = stone_other(seq_color)) {
 491                         if (u->local_tree_allseq) {
 492                                 /* We are configured to record all subsequences. */
 493                                 record_local_sequence(u, t, &b2, descent, dlen, dseqi, seq_color);
 494                                 continue;
 495                         }
 496                         if (descent[dseqi].node->d >= u->tenuki_d) {
 497                                 /* Tenuki! Record the fresh sequence. */
 498                                 record_local_sequence(u, t, &b2, descent, dlen, dseqi, seq_color);
 499                                 continue;
 500                         }
 501                         if (descent[dseqi].lnode && !descent[dseqi].lnode) {
 502                                 /* Record result for in-descent picked sequence. */
 503                                 record_local_sequence(u, t, &b2, descent, dlen, dseqi, seq_color);
 504                                 continue;
 505                         }
 506                 }
 507         }
 508
 509 end:
 510         /* We need to undo the virtual loss we added during descend. */
 511         if (u->virtual_loss) {
 512                 floating_t loss = node_color == S_BLACK ? 0.0 : 1.0;
 513                 for (; n->parent; n = n->parent) {
 514                         stats_rm_result(&n->u, loss, u->virtual_loss);
 515                         loss = 1.0 - loss;
 516                 }
 517         }
 518
 519         if (amaf) {
 520                 free(amaf->map - 1);
 521                 free(amaf);
 522         }
 523         board_done_noalloc(&b2);
 524         return result;
 525 }
 526
 527 int
 528 uct_playouts(struct uct *u, struct board *b, enum stone color, struct tree *t, struct time_info *ti)
 529 {
 530         int i;
 531         if (ti && ti->dim == TD_GAMES) {
 532                 for (i = 0; t->root->u.playouts <= ti->len.games; i++)
 533                         uct_playout(u, b, color, t);
 534         } else {
 535                 for (i = 0; !uct_halt; i++)
 536                         uct_playout(u, b, color, t);
 537         }
 538         return i;
 539 }