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