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