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 "playout/elo.h"
  16 #include "probdist.h"
  17 #include "random.h"
  18 #include "uct/dynkomi.h"
  19 #include "uct/internal.h"
  20 #include "uct/search.h"
  21 #include "uct/tree.h"
  22 #include "uct/uct.h"
  23 #include "uct/walk.h"
  24
  25 void
  26 uct_progress_status(struct uct *u, struct tree *t, enum stone color, int playouts)
  27 {
  28         if (!UDEBUGL(0))
  29                 return;
  30
  31         /* Best move */
  32         struct tree_node *best = u->policy->choose(u->policy, t->root, t->board, color, resign);
  33         if (!best) {
  34                 fprintf(stderr, "... No moves left\n");
  35                 return;
  36         }
  37         fprintf(stderr, "[%d] ", playouts);
  38         fprintf(stderr, "best %f ", tree_node_get_value(t, 1, best->u.value));
  39
  40         /* Dynamic komi */
  41         if (t->use_extra_komi)
  42                 fprintf(stderr, "komi %.1f ", t->extra_komi);
  43
  44         /* Best sequence */
  45         fprintf(stderr, "| seq ");
  46         for (int depth = 0; depth < 4; depth++) {
  47                 if (best && best->u.playouts >= 25) {
  48                         fprintf(stderr, "%3s ", coord2sstr(best->coord, t->board));
  49                         best = u->policy->choose(u->policy, best, t->board, color, resign);
  50                 } else {
  51                         fprintf(stderr, "    ");
  52                 }
  53         }
  54
  55         /* Best candidates */
  56         fprintf(stderr, "| can ");
  57         int cans = 4;
  58         struct tree_node *can[cans];
  59         memset(can, 0, sizeof(can));
  60         best = t->root->children;
  61         while (best) {
  62                 int c = 0;
  63                 while ((!can[c] || best->u.playouts > can[c]->u.playouts) && ++c < cans);
  64                 for (int d = 0; d < c; d++) can[d] = can[d + 1];
  65                 if (c > 0) can[c - 1] = best;
  66                 best = best->sibling;
  67         }
  68         while (--cans >= 0) {
  69                 if (can[cans]) {
  70                         fprintf(stderr, "%3s(%.3f) ",
  71                                 coord2sstr(can[cans]->coord, t->board),
  72                                 tree_node_get_value(t, 1, can[cans]->u.value));
  73                 } else {
  74                         fprintf(stderr, "           ");
  75                 }
  76         }
  77
  78         fprintf(stderr, "\n");
  79 }
  80
  81
  82 static void
  83 record_amaf_move(struct playout_amafmap *amaf, coord_t coord, enum stone color)
  84 {
  85         if (amaf->map[coord] == S_NONE || amaf->map[coord] == color) {
  86                 amaf->map[coord] = color;
  87         } else { // XXX: Respect amaf->record_nakade
  88                 amaf_op(amaf->map[coord], +);
  89         }
  90         amaf->game[amaf->gamelen].coord = coord;
  91         amaf->game[amaf->gamelen].color = color;
  92         amaf->gamelen++;
  93         assert(amaf->gamelen < sizeof(amaf->game) / sizeof(amaf->game[0]));
  94 }
  95
  96 static double
  97 ltree_node_gamma(struct tree_node *li, enum stone color)
  98 {
  99         /* TODO: How to do this? */
 100         #define li_value(color, li) (li->u.playouts * (color == S_BLACK ? li->u.value : (1 - li->u.value)))
 101         return 0.5 + li_value(color, li);
 102 }
 103
 104
 105 struct uct_playout_callback {
 106         struct uct *uct;
 107         struct tree *tree;
 108         struct tree_node *lnode;
 109 };
 110
 111 static void
 112 uct_playout_probdist(void *data, struct board *b, enum stone to_play, struct probdist *pd)
 113 {
 114         /* Create probability distribution according to found local tree
 115          * sequence. */
 116         struct uct_playout_callback *upc = data;
 117         assert(upc && upc->tree && pd && b);
 118         coord_t c = b->last_move.coord;
 119         enum stone color = b->last_move.color;
 120
 121         if (is_pass(c)) {
 122                 /* Break local sequence. */
 123                 upc->lnode = NULL;
 124         } else if (upc->lnode) {
 125                 /* Try to follow local sequence. */
 126                 upc->lnode = tree_get_node(upc->tree, upc->lnode, c, false);
 127         }
 128
 129         if (!upc->lnode || !upc->lnode->children) {
 130                 /* There's no local sequence, start new one! */
 131                 upc->lnode = color == S_BLACK ? upc->tree->ltree_black : upc->tree->ltree_white;
 132                 upc->lnode = tree_get_node(upc->tree, upc->lnode, c, false);
 133         }
 134
 135         if (!upc->lnode || !upc->lnode->children) {
 136                 /* We have no local sequence and we cannot find any starting
 137                  * by node corresponding to last move. */
 138                 if (!upc->uct->local_tree_pseqroot) {
 139                         /* Give up then, we have nothing to contribute. */
 140                         return;
 141                 }
 142                 /* Construct probability distribution from possible first
 143                  * sequence move. Remember that @color is color of the
 144                  * *last* move. */
 145                 upc->lnode = color == S_BLACK ? upc->tree->ltree_white : upc->tree->ltree_black;
 146                 if (!upc->lnode->children) {
 147                         /* We don't even have anything in our tree yet. */
 148                         return;
 149                 }
 150         }
 151
 152         /* The probdist has the right structure only if BOARD_GAMMA is defined. */
 153 #ifndef BOARD_GAMMA
 154         assert(0);
 155 #endif
 156
 157         /* Construct probability distribution from lnode children. */
 158         struct tree_node *li = upc->lnode->children;
 159         assert(li);
 160         if (is_pass(li->coord)) {
 161                 /* Tenuki. */
 162                 /* TODO: Spread tenuki gamma over all moves we don't touch. */
 163                 li = li->sibling;
 164         }
 165         for (; li; li = li->sibling) {
 166                 if (board_at(b, li->coord) != S_NONE)
 167                         continue;
 168                 double gamma = fixp_to_double(pd->items[li->coord]) * ltree_node_gamma(li, to_play);
 169                 probdist_set(pd, li->coord, double_to_fixp(gamma));
 170         }
 171 }
 172
 173
 174 static coord_t
 175 uct_playout_hook(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color, int mode)
 176 {
 177         struct uct_playout_callback *upc = setup->hook_data;
 178
 179         /* TODO: Fill me. */
 180         return pass;
 181 }
 182
 183 static coord_t
 184 uct_playout_prepolicy(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color)
 185 {
 186         return uct_playout_hook(playout, setup, b, color, 0);
 187 }
 188
 189 static coord_t
 190 uct_playout_postpolicy(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color)
 191 {
 192         return uct_playout_hook(playout, setup, b, color, 1);
 193 }
 194
 195
 196 static int
 197 uct_leaf_node(struct uct *u, struct board *b, enum stone player_color,
 198               struct playout_amafmap *amaf,
 199               struct tree *t, struct tree_node *n, enum stone node_color,
 200               char *spaces)
 201 {
 202         enum stone next_color = stone_other(node_color);
 203         int parity = (next_color == player_color ? 1 : -1);
 204
 205         /* We need to make sure only one thread expands the node. If
 206          * we are unlucky enough for two threads to meet in the same
 207          * node, the latter one will simply do another simulation from
 208          * the node itself, no big deal. t->nodes_size may exceed
 209          * the maximum in multi-threaded case but not by much so it's ok.
 210          * The size test must be before the test&set not after, to allow
 211          * expansion of the node later if enough nodes have been freed. */
 212         if (n->u.playouts >= u->expand_p && t->nodes_size < u->max_tree_size
 213             && !__sync_lock_test_and_set(&n->is_expanded, 1)) {
 214                 tree_expand_node(t, n, b, next_color, u, parity);
 215         }
 216         if (UDEBUGL(7))
 217                 fprintf(stderr, "%s*-- UCT playout #%d start [%s] %f\n",
 218                         spaces, n->u.playouts, coord2sstr(n->coord, t->board),
 219                         tree_node_get_value(t, parity, n->u.value));
 220
 221         /* TODO: Don't necessarily restart the sequence walk when entering
 222          * playout. */
 223         struct uct_playout_callback upc = { .uct = u, .tree = t, .lnode = NULL };
 224         if (u->local_tree_playout) {
 225                 /* N.B.: We know this is ELO playout. */
 226                 playout_elo_callback(u->playout, uct_playout_probdist, &upc);
 227         }
 228
 229         struct playout_setup ps = {
 230                 .gamelen = u->gamelen,
 231                 .mercymin = u->mercymin,
 232                 .prepolicy_hook = uct_playout_prepolicy,
 233                 .postpolicy_hook = uct_playout_postpolicy,
 234                 .hook_data = &upc,
 235         };
 236         int result = play_random_game(&ps, b, next_color,
 237                                       u->playout_amaf ? amaf : NULL,
 238                                       &u->ownermap, u->playout);
 239         if (next_color == S_WHITE) {
 240                 /* We need the result from black's perspective. */
 241                 result = - result;
 242         }
 243         if (UDEBUGL(7))
 244                 fprintf(stderr, "%s -- [%d..%d] %s random playout result %d\n",
 245                         spaces, player_color, next_color, coord2sstr(n->coord, t->board), result);
 246
 247         return result;
 248 }
 249
 250 static float
 251 scale_value(struct uct *u, struct board *b, int result)
 252 {
 253         float rval = result > 0;
 254         if (u->val_scale) {
 255                 int vp = u->val_points;
 256                 if (!vp) {
 257                         vp = board_size(b) - 1; vp *= vp; vp *= 2;
 258                 }
 259
 260                 float sval = (float) abs(result) / vp;
 261                 sval = sval > 1 ? 1 : sval;
 262                 if (result < 0) sval = 1 - sval;
 263                 if (u->val_extra)
 264                         rval += u->val_scale * sval;
 265                 else
 266                         rval = (1 - u->val_scale) * rval + u->val_scale * sval;
 267                 // fprintf(stderr, "score %d => sval %f, rval %f\n", result, sval, rval);
 268         }
 269         return rval;
 270 }
 271
 272 static void
 273 record_local_sequence(struct uct *u, struct tree *t,
 274                       struct uct_descent *descent, int dlen, int di,
 275                       enum stone seq_color, float rval)
 276 {
 277         /* Ignore pass sequences. */
 278         if (is_pass(descent[di].node->coord))
 279                 return;
 280
 281 #define LTREE_DEBUG if (UDEBUGL(6))
 282         LTREE_DEBUG fprintf(stderr, "recording result %f in local %s sequence: ",
 283                 rval, stone2str(seq_color));
 284         int di0 = di;
 285
 286         /* Pick the right local tree root... */
 287         struct tree_node *lnode = seq_color == S_BLACK ? t->ltree_black : t->ltree_white;
 288         lnode->u.playouts++;
 289
 290         /* ...and record the sequence. */
 291         while (di < dlen && (di == di0 || descent[di].node->d < u->tenuki_d)) {
 292                 LTREE_DEBUG fprintf(stderr, "%s[%d] ",
 293                         coord2sstr(descent[di].node->coord, t->board),
 294                         descent[di].node->d);
 295                 lnode = tree_get_node(t, lnode, descent[di++].node->coord, true);
 296                 assert(lnode);
 297                 stats_add_result(&lnode->u, rval, 1);
 298         }
 299
 300         /* Add lnode for tenuki (pass) if we descended further. */
 301         if (di < dlen) {
 302                 LTREE_DEBUG fprintf(stderr, "pass ");
 303                 lnode = tree_get_node(t, lnode, pass, true);
 304                 assert(lnode);
 305                 stats_add_result(&lnode->u, rval, 1);
 306         }
 307
 308         LTREE_DEBUG fprintf(stderr, "\n");
 309 }
 310
 311
 312 int
 313 uct_playout(struct uct *u, struct board *b, enum stone player_color, struct tree *t)
 314 {
 315         struct board b2;
 316         board_copy(&b2, b);
 317
 318         struct playout_amafmap *amaf = NULL;
 319         if (u->policy->wants_amaf) {
 320                 amaf = calloc2(1, sizeof(*amaf));
 321                 amaf->map = calloc2(board_size2(&b2) + 1, sizeof(*amaf->map));
 322                 amaf->map++; // -1 is pass
 323         }
 324
 325         /* Walk the tree until we find a leaf, then expand it and do
 326          * a random playout. */
 327         struct tree_node *n = t->root;
 328         enum stone node_color = stone_other(player_color);
 329         assert(node_color == t->root_color);
 330
 331         /* Tree descent history. */
 332         /* XXX: This is somewhat messy since @n and descent[dlen-1].node are
 333          * redundant. */
 334         #define DLEN 512
 335         struct uct_descent descent[DLEN];
 336         descent[0].node = n; descent[0].lnode = NULL;
 337         descent[0].significant[0] = descent[0].significant[1] = NULL;
 338         int dlen = 1;
 339         /* Total value of the sequence. */
 340         struct move_stats seq_value = { .playouts = 0 };
 341
 342         int result;
 343         int pass_limit = (board_size(&b2) - 2) * (board_size(&b2) - 2) / 2;
 344         int passes = is_pass(b->last_move.coord) && b->moves > 0;
 345
 346         /* debug */
 347         int depth = 0;
 348         static char spaces[] = "\0                                                      ";
 349         /* /debug */
 350         if (UDEBUGL(8))
 351                 fprintf(stderr, "--- UCT walk with color %d\n", player_color);
 352
 353         while (!tree_leaf_node(n) && passes < 2) {
 354                 spaces[depth++] = ' '; spaces[depth] = 0;
 355
 356
 357                 /*** Choose a node to descend to: */
 358
 359                 /* Parity is chosen already according to the child color, since
 360                  * it is applied to children. */
 361                 node_color = stone_other(node_color);
 362                 int parity = (node_color == player_color ? 1 : -1);
 363
 364                 assert(dlen < DLEN);
 365                 descent[dlen] = descent[dlen - 1];
 366                 if (u->local_tree && (!descent[dlen].lnode || descent[dlen].node->d >= u->tenuki_d)) {
 367                         /* Start new local sequence. */
 368                         /* Remember that node_color already holds color of the
 369                          * to-be-found child. */
 370                         descent[dlen].lnode = node_color == S_BLACK ? t->ltree_black : t->ltree_white;
 371                 }
 372
 373                 if (!u->random_policy_chance || fast_random(u->random_policy_chance))
 374                         u->policy->descend(u->policy, t, &descent[dlen], parity, b2.moves > pass_limit);
 375                 else
 376                         u->random_policy->descend(u->random_policy, t, &descent[dlen], parity, b2.moves > pass_limit);
 377
 378
 379                 /*** Perform the descent: */
 380
 381                 if (descent[dlen].node->u.playouts >= u->significant_threshold) {
 382                         descent[dlen].significant[node_color - 1] = n;
 383                 }
 384
 385                 seq_value.playouts += descent[dlen].value.playouts;
 386                 seq_value.value += descent[dlen].value.value * descent[dlen].value.playouts;
 387                 n = descent[dlen++].node;
 388                 assert(n == t->root || n->parent);
 389                 if (UDEBUGL(7))
 390                         fprintf(stderr, "%s+-- UCT sent us to [%s:%d] %f\n",
 391                                 spaces, coord2sstr(n->coord, t->board), n->coord,
 392                                 tree_node_get_value(t, parity, n->u.value));
 393
 394                 /* Add virtual loss if we need to; this is used to discourage
 395                  * other threads from visiting this node in case of multiple
 396                  * threads doing the tree search. */
 397                 if (u->virtual_loss)
 398                         stats_add_result(&n->u, tree_parity(t, parity) > 0 ? 0 : 1, 1);
 399
 400                 assert(n->coord >= -1);
 401                 if (amaf && !is_pass(n->coord))
 402                         record_amaf_move(amaf, n->coord, node_color);
 403
 404                 struct move m = { n->coord, node_color };
 405                 int res = board_play(&b2, &m);
 406
 407                 if (res < 0 || (!is_pass(m.coord) && !group_at(&b2, m.coord)) /* suicide */
 408                     || b2.superko_violation) {
 409                         if (UDEBUGL(4)) {
 410                                 for (struct tree_node *ni = n; ni; ni = ni->parent)
 411                                         fprintf(stderr, "%s<%"PRIhash"> ", coord2sstr(ni->coord, t->board), ni->hash);
 412                                 fprintf(stderr, "marking invalid %s node %d,%d res %d group %d spk %d\n",
 413                                         stone2str(node_color), coord_x(n->coord,b), coord_y(n->coord,b),
 414                                         res, group_at(&b2, m.coord), b2.superko_violation);
 415                         }
 416                         n->hints |= TREE_HINT_INVALID;
 417                         result = 0;
 418                         goto end;
 419                 }
 420
 421                 if (is_pass(n->coord))
 422                         passes++;
 423                 else
 424                         passes = 0;
 425         }
 426
 427         if (amaf) {
 428                 amaf->game_baselen = amaf->gamelen;
 429                 amaf->record_nakade = u->playout_amaf_nakade;
 430         }
 431
 432         if (t->use_extra_komi && u->dynkomi->persim) {
 433                 b2.komi += round(u->dynkomi->persim(u->dynkomi, &b2, t, n));
 434         }
 435
 436         if (passes >= 2) {
 437                 /* XXX: No dead groups support. */
 438                 float score = board_official_score(&b2, NULL);
 439                 /* Result from black's perspective (no matter who
 440                  * the player; black's perspective is always
 441                  * what the tree stores. */
 442                 result = - (score * 2);
 443
 444                 if (UDEBUGL(5))
 445                         fprintf(stderr, "[%d..%d] %s p-p scoring playout result %d (W %f)\n",
 446                                 player_color, node_color, coord2sstr(n->coord, t->board), result, score);
 447                 if (UDEBUGL(6))
 448                         board_print(&b2, stderr);
 449
 450                 board_ownermap_fill(&u->ownermap, &b2);
 451
 452         } else { // assert(tree_leaf_node(n));
 453                 /* In case of parallel tree search, the assertion might
 454                  * not hold if two threads chew on the same node. */
 455                 result = uct_leaf_node(u, &b2, player_color, amaf, &descent[dlen - 1], t, n, node_color, spaces);
 456         }
 457
 458         if (amaf && u->playout_amaf_cutoff) {
 459                 unsigned int cutoff = amaf->game_baselen;
 460                 cutoff += (amaf->gamelen - amaf->game_baselen) * u->playout_amaf_cutoff / 100;
 461                 /* Now, reconstruct the amaf map. */
 462                 memset(amaf->map, 0, board_size2(&b2) * sizeof(*amaf->map));
 463                 for (unsigned int i = 0; i < cutoff; i++) {
 464                         coord_t coord = amaf->game[i].coord;
 465                         enum stone color = amaf->game[i].color;
 466                         if (amaf->map[coord] == S_NONE || amaf->map[coord] == color) {
 467                                 amaf->map[coord] = color;
 468                         /* Nakade always recorded for in-tree part */
 469                         } else if (amaf->record_nakade || i <= amaf->game_baselen) {
 470                                 amaf_op(amaf->map[n->coord], +);
 471                         }
 472                 }
 473         }
 474
 475         assert(n == t->root || n->parent);
 476         if (result != 0) {
 477                 float rval = scale_value(u, b, result);
 478                 u->policy->update(u->policy, t, n, node_color, player_color, amaf, rval);
 479
 480                 if (t->use_extra_komi) {
 481                         stats_add_result(&u->dynkomi->score, result / 2, 1);
 482                         stats_add_result(&u->dynkomi->value, rval, 1);
 483                 }
 484
 485                 if (u->local_tree && n->parent && !is_pass(n->coord) && dlen > 0) {
 486                         /* Possibly transform the rval appropriately. */
 487                         float expval = seq_value.value / seq_value.playouts;
 488                         rval = stats_temper_value(rval, expval, u->local_tree);
 489
 490                         /* Get the local sequences and record them in ltree. */
 491                         /* We will look for sequence starts in our descent
 492                          * history, then run record_local_sequence() for each
 493                          * found sequence start; record_local_sequence() may
 494                          * pick longer sequences from descent history then,
 495                          * which is expected as it will create new lnodes. */
 496                         enum stone seq_color = player_color;
 497                         /* First move always starts a sequence. */
 498                         record_local_sequence(u, t, descent, dlen, 1, seq_color, rval);
 499                         seq_color = stone_other(seq_color);
 500                         for (int dseqi = 2; dseqi < dlen; dseqi++, seq_color = stone_other(seq_color)) {
 501                                 if (u->local_tree_allseq) {
 502                                         /* We are configured to record all subsequences. */
 503                                         record_local_sequence(u, t, descent, dlen, dseqi, seq_color, rval);
 504                                         continue;
 505                                 }
 506                                 if (descent[dseqi].node->d >= u->tenuki_d) {
 507                                         /* Tenuki! Record the fresh sequence. */
 508                                         record_local_sequence(u, t, descent, dlen, dseqi, seq_color, rval);
 509                                         continue;
 510                                 }
 511                                 if (descent[dseqi].lnode && !descent[dseqi].lnode) {
 512                                         /* Record result for in-descent picked sequence. */
 513                                         record_local_sequence(u, t, descent, dlen, dseqi, seq_color, rval);
 514                                         continue;
 515                                 }
 516                         }
 517                 }
 518         }
 519
 520 end:
 521         /* We need to undo the virtual loss we added during descend. */
 522         if (u->virtual_loss) {
 523                 int parity = (node_color == player_color ? 1 : -1);
 524                 for (; n->parent; n = n->parent) {
 525                         stats_rm_result(&n->u, tree_parity(t, parity) > 0 ? 0 : 1, 1);
 526                         parity = -parity;
 527                 }
 528         }
 529
 530         if (amaf) {
 531                 free(amaf->map - 1);
 532                 free(amaf);
 533         }
 534         board_done_noalloc(&b2);
 535         return result;
 536 }
 537
 538 int
 539 uct_playouts(struct uct *u, struct board *b, enum stone color, struct tree *t, struct time_info *ti)
 540 {
 541         int i;
 542         if (ti->dim == TD_GAMES) {
 543                 for (i = 0; t->root->u.playouts <= ti->len.games; i++)
 544                         uct_playout(u, b, color, t);
 545         } else {
 546                 for (i = 0; !uct_halt; i++)
 547                         uct_playout(u, b, color, t);
 548         }
 549         return i;
 550 }