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