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