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