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 "probdist.h"
  16 #include "random.h"
  17 #include "tactics/util.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 #define DESCENT_DLEN 512
  26
  27 void
  28 uct_progress_status(struct uct *u, struct tree *t, enum stone color, int playouts)
  29 {
  30         if (!UDEBUGL(0))
  31                 return;
  32
  33         /* Best move */
  34         struct tree_node *best = u->policy->choose(u->policy, t->root, t->board, color, resign);
  35         if (!best) {
  36                 fprintf(stderr, "... No moves left\n");
  37                 return;
  38         }
  39         fprintf(stderr, "[%d] ", playouts);
  40         fprintf(stderr, "best %f ", tree_node_get_value(t, 1, best->u.value));
  41
  42         /* Dynamic komi */
  43         if (t->use_extra_komi)
  44                 fprintf(stderr, "komi %.1f ", t->extra_komi);
  45
  46         /* Best sequence */
  47         fprintf(stderr, "| seq ");
  48         for (int depth = 0; depth < 4; depth++) {
  49                 if (best && best->u.playouts >= 25) {
  50                         fprintf(stderr, "%3s ", coord2sstr(node_coord(best), 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(node_coord(can[cans]), 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 static inline void
  85 record_amaf_move(struct playout_amafmap *amaf, coord_t coord)
  86 {
  87         assert(amaf->gamelen < MAX_GAMELEN);
  88         amaf->game[amaf->gamelen++] = coord;
  89 }
  90
  91
  92 struct uct_playout_callback {
  93         struct uct *uct;
  94         struct tree *tree;
  95         struct tree_node *lnode;
  96 };
  97
  98
  99 static coord_t
 100 uct_playout_hook(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color, int mode)
 101 {
 102         /* XXX: This is used in some non-master branches. */
 103         return pass;
 104 }
 105
 106 static coord_t
 107 uct_playout_prepolicy(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color)
 108 {
 109         return uct_playout_hook(playout, setup, b, color, 0);
 110 }
 111
 112 static coord_t
 113 uct_playout_postpolicy(struct playout_policy *playout, struct playout_setup *setup, struct board *b, enum stone color)
 114 {
 115         return uct_playout_hook(playout, setup, b, color, 1);
 116 }
 117
 118
 119 static int
 120 uct_leaf_node(struct uct *u, struct board *b, enum stone player_color,
 121               struct playout_amafmap *amaf,
 122               struct uct_descent *descent, int *dlen,
 123               struct tree_node *significant[2],
 124               struct tree *t, struct tree_node *n, enum stone node_color,
 125               char *spaces)
 126 {
 127         enum stone next_color = stone_other(node_color);
 128         int parity = (next_color == player_color ? 1 : -1);
 129
 130         if (UDEBUGL(7))
 131                 fprintf(stderr, "%s*-- UCT playout #%d start [%s] %f\n",
 132                         spaces, n->u.playouts, coord2sstr(node_coord(n), t->board),
 133                         tree_node_get_value(t, parity, n->u.value));
 134
 135         struct uct_playout_callback upc = {
 136                 .uct = u,
 137                 .tree = t,
 138                 /* TODO: Don't necessarily restart the sequence walk when
 139                  * entering playout. */
 140                 .lnode = NULL,
 141         };
 142
 143         struct playout_setup ps = {
 144                 .gamelen = u->gamelen,
 145                 .mercymin = u->mercymin,
 146                 .prepolicy_hook = uct_playout_prepolicy,
 147                 .postpolicy_hook = uct_playout_postpolicy,
 148                 .hook_data = &upc,
 149         };
 150         int result = play_random_game(&ps, b, next_color,
 151                                       u->playout_amaf ? amaf : NULL,
 152                                       &u->ownermap, u->playout);
 153         if (next_color == S_WHITE) {
 154                 /* We need the result from black's perspective. */
 155                 result = - result;
 156         }
 157         if (UDEBUGL(7))
 158                 fprintf(stderr, "%s -- [%d..%d] %s random playout result %d\n",
 159                         spaces, player_color, next_color, coord2sstr(node_coord(n), t->board), result);
 160
 161         return result;
 162 }
 163
 164 static floating_t
 165 scale_value(struct uct *u, struct board *b, int result)
 166 {
 167         floating_t rval = result > 0 ? 1.0 : result < 0 ? 0.0 : 0.5;
 168         if (u->val_scale && result != 0) {
 169                 int vp = u->val_points;
 170                 if (!vp) {
 171                         vp = board_size(b) - 1; vp *= vp; vp *= 2;
 172                 }
 173
 174                 floating_t sval = (floating_t) abs(result) / vp;
 175                 sval = sval > 1 ? 1 : sval;
 176                 if (result < 0) sval = 1 - sval;
 177                 if (u->val_extra)
 178                         rval += u->val_scale * sval;
 179                 else
 180                         rval = (1 - u->val_scale) * rval + u->val_scale * sval;
 181                 // fprintf(stderr, "score %d => sval %f, rval %f\n", result, sval, rval);
 182         }
 183         return rval;
 184 }
 185
 186 static double
 187 local_value(struct uct *u, struct board *b, coord_t coord, enum stone color)
 188 {
 189         /* Tactical evaluation of move @coord by color @color, given
 190          * simulation end position @b. I.e., a move is tactically good
 191          * if the resulting group stays on board until the game end. */
 192         /* We can also take into account surrounding stones, e.g. to
 193          * encourage taking off external liberties during a semeai. */
 194         double val = board_local_value(u->local_tree_neival, b, coord, color);
 195         return (color == S_WHITE) ? 1.f - val : val;
 196 }
 197
 198 static void
 199 record_local_sequence(struct uct *u, struct tree *t, struct board *endb,
 200                       struct uct_descent *descent, int dlen, int di,
 201                       enum stone seq_color)
 202 {
 203 #define LTREE_DEBUG if (UDEBUGL(6))
 204
 205         /* Ignore pass sequences. */
 206         if (is_pass(node_coord(descent[di].node)))
 207                 return;
 208
 209         LTREE_DEBUG board_print(endb, stderr);
 210         LTREE_DEBUG fprintf(stderr, "recording local %s sequence: ",
 211                 stone2str(seq_color));
 212
 213         /* Sequences starting deeper are less relevant in general. */
 214         int pval = LTREE_PLAYOUTS_MULTIPLIER;
 215         if (u->local_tree && u->local_tree_depth_decay > 0)
 216                 pval = ((floating_t) pval) / pow(u->local_tree_depth_decay, di - 1);
 217         if (!pval) {
 218                 LTREE_DEBUG fprintf(stderr, "too deep @%d\n", di);
 219                 return;
 220         }
 221
 222         /* Pick the right local tree root... */
 223         struct tree_node *lnode = seq_color == S_BLACK ? t->ltree_black : t->ltree_white;
 224         lnode->u.playouts++;
 225
 226         double sval = 0.5;
 227         if (u->local_tree_rootgoal) {
 228                 sval = local_value(u, endb, node_coord(descent[di].node), seq_color);
 229                 LTREE_DEBUG fprintf(stderr, "(goal %s[%s %1.3f][%d]) ",
 230                         coord2sstr(node_coord(descent[di].node), t->board),
 231                         stone2str(seq_color), sval, descent[di].node->d);
 232         }
 233
 234         /* ...and record the sequence. */
 235         int di0 = di;
 236         while (di < dlen && (di == di0 || descent[di].node->d < u->tenuki_d)) {
 237                 enum stone color = (di - di0) % 2 ? stone_other(seq_color) : seq_color;
 238                 double rval;
 239                 if (u->local_tree_rootgoal)
 240                         rval = sval;
 241                 else
 242                         rval = local_value(u, endb, node_coord(descent[di].node), color);
 243                 LTREE_DEBUG fprintf(stderr, "%s[%s %1.3f][%d] ",
 244                         coord2sstr(node_coord(descent[di].node), t->board),
 245                         stone2str(color), rval, descent[di].node->d);
 246                 lnode = tree_get_node(t, lnode, node_coord(descent[di++].node), true);
 247                 assert(lnode);
 248                 stats_add_result(&lnode->u, rval, pval);
 249         }
 250
 251         /* Add lnode for tenuki (pass) if we descended further. */
 252         if (di < dlen) {
 253                 double rval = u->local_tree_rootgoal ? sval : 0.5;
 254                 LTREE_DEBUG fprintf(stderr, "pass ");
 255                 lnode = tree_get_node(t, lnode, pass, true);
 256                 assert(lnode);
 257                 stats_add_result(&lnode->u, rval, pval);
 258         }
 259
 260         LTREE_DEBUG fprintf(stderr, "\n");
 261 }
 262
 263
 264 int
 265 uct_playout(struct uct *u, struct board *b, enum stone player_color, struct tree *t)
 266 {
 267         struct board b2;
 268         board_copy(&b2, b);
 269
 270         struct playout_amafmap amaf;
 271         amaf.gamelen = amaf.game_baselen = 0;
 272
 273         /* Walk the tree until we find a leaf, then expand it and do
 274          * a random playout. */
 275         struct tree_node *n = t->root;
 276         enum stone node_color = stone_other(player_color);
 277         assert(node_color == t->root_color);
 278
 279         /* Make sure the root node is expanded. */
 280         if (tree_leaf_node(n) && !__sync_lock_test_and_set(&n->is_expanded, 1))
 281                 tree_expand_node(t, n, &b2, player_color, u, 1);
 282
 283         /* Tree descent history. */
 284         /* XXX: This is somewhat messy since @n and descent[dlen-1].node are
 285          * redundant. */
 286         struct uct_descent descent[DESCENT_DLEN];
 287         descent[0].node = n; descent[0].lnode = NULL;
 288         int dlen = 1;
 289         /* Total value of the sequence. */
 290         struct move_stats seq_value = { .playouts = 0 };
 291         /* The last "significant" node along the descent (i.e. node
 292          * with higher than configured number of playouts). For black
 293          * and white. */
 294         struct tree_node *significant[2] = { NULL, NULL };
 295         if (n->u.playouts >= u->significant_threshold)
 296                 significant[node_color - 1] = n;
 297
 298         int result;
 299         int pass_limit = (board_size(&b2) - 2) * (board_size(&b2) - 2) / 2;
 300         int passes = is_pass(b->last_move.coord) && b->moves > 0;
 301
 302         /* debug */
 303         static char spaces[] = "\0                                                      ";
 304         /* /debug */
 305         if (UDEBUGL(8))
 306                 fprintf(stderr, "--- UCT walk with color %d\n", player_color);
 307
 308         while (!tree_leaf_node(n) && passes < 2) {
 309                 spaces[dlen - 1] = ' '; spaces[dlen] = 0;
 310
 311
 312                 /*** Choose a node to descend to: */
 313
 314                 /* Parity is chosen already according to the child color, since
 315                  * it is applied to children. */
 316                 node_color = stone_other(node_color);
 317                 int parity = (node_color == player_color ? 1 : -1);
 318
 319                 assert(dlen < DESCENT_DLEN);
 320                 descent[dlen] = descent[dlen - 1];
 321                 if (u->local_tree && (!descent[dlen].lnode || descent[dlen].node->d >= u->tenuki_d)) {
 322                         /* Start new local sequence. */
 323                         /* Remember that node_color already holds color of the
 324                          * to-be-found child. */
 325                         descent[dlen].lnode = node_color == S_BLACK ? t->ltree_black : t->ltree_white;
 326                 }
 327
 328                 if (!u->random_policy_chance || fast_random(u->random_policy_chance))
 329                         u->policy->descend(u->policy, t, &descent[dlen], parity, b2.moves > pass_limit);
 330                 else
 331                         u->random_policy->descend(u->random_policy, t, &descent[dlen], parity, b2.moves > pass_limit);
 332
 333
 334                 /*** Perform the descent: */
 335
 336                 if (descent[dlen].node->u.playouts >= u->significant_threshold) {
 337                         significant[node_color - 1] = descent[dlen].node;
 338                 }
 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] %d,%f\n",
 346                                 spaces, coord2sstr(node_coord(n), t->board),
 347                                 node_coord(n), n->u.playouts,
 348                                 tree_node_get_value(t, parity, n->u.value));
 349
 350                 /* Add virtual loss if we need to; this is used to discourage
 351                  * other threads from visiting this node in case of multiple
 352                  * threads doing the tree search. */
 353                 if (u->virtual_loss)
 354                         stats_add_result(&n->u, node_color == S_BLACK ? 0.0 : 1.0, u->virtual_loss);
 355
 356                 assert(node_coord(n) >= -1);
 357                 record_amaf_move(&amaf, node_coord(n));
 358
 359                 struct move m = { node_coord(n), node_color };
 360                 int res = board_play(&b2, &m);
 361
 362                 if (res < 0 || (!is_pass(m.coord) && !group_at(&b2, m.coord)) /* suicide */
 363                     || b2.superko_violation) {
 364                         if (UDEBUGL(4)) {
 365                                 for (struct tree_node *ni = n; ni; ni = ni->parent)
 366                                         fprintf(stderr, "%s<%"PRIhash"> ", coord2sstr(node_coord(ni), t->board), ni->hash);
 367                                 fprintf(stderr, "marking invalid %s node %d,%d res %d group %d spk %d\n",
 368                                         stone2str(node_color), coord_x(node_coord(n),b), coord_y(node_coord(n),b),
 369                                         res, group_at(&b2, m.coord), b2.superko_violation);
 370                         }
 371                         n->hints |= TREE_HINT_INVALID;
 372                         result = 0;
 373                         goto end;
 374                 }
 375
 376                 if (is_pass(node_coord(n)))
 377                         passes++;
 378                 else
 379                         passes = 0;
 380
 381                 enum stone next_color = stone_other(node_color);
 382                 /* We need to make sure only one thread expands the node. If
 383                  * we are unlucky enough for two threads to meet in the same
 384                  * node, the latter one will simply do another simulation from
 385                  * the node itself, no big deal. t->nodes_size may exceed
 386                  * the maximum in multi-threaded case but not by much so it's ok.
 387                  * The size test must be before the test&set not after, to allow
 388                  * expansion of the node later if enough nodes have been freed. */
 389                 if (tree_leaf_node(n)
 390                     && n->u.playouts - u->virtual_loss >= u->expand_p && t->nodes_size < u->max_tree_size
 391                     && !__sync_lock_test_and_set(&n->is_expanded, 1))
 392                         tree_expand_node(t, n, &b2, next_color, u, -parity);
 393         }
 394
 395         amaf.game_baselen = amaf.gamelen;
 396
 397         if (t->use_extra_komi && u->dynkomi->persim) {
 398                 b2.komi += round(u->dynkomi->persim(u->dynkomi, &b2, t, n));
 399         }
 400
 401         if (passes >= 2) {
 402                 /* XXX: No dead groups support. */
 403                 floating_t score = board_official_score(&b2, NULL);
 404                 /* Result from black's perspective (no matter who
 405                  * the player; black's perspective is always
 406                  * what the tree stores. */
 407                 result = - (score * 2);
 408
 409                 if (UDEBUGL(5))
 410                         fprintf(stderr, "[%d..%d] %s p-p scoring playout result %d (W %f)\n",
 411                                 player_color, node_color, coord2sstr(node_coord(n), t->board), result, score);
 412                 if (UDEBUGL(6))
 413                         board_print(&b2, stderr);
 414
 415                 board_ownermap_fill(&u->ownermap, &b2);
 416
 417         } else { // assert(tree_leaf_node(n));
 418                 /* In case of parallel tree search, the assertion might
 419                  * not hold if two threads chew on the same node. */
 420                 result = uct_leaf_node(u, &b2, player_color, &amaf, descent, &dlen, significant, t, n, node_color, spaces);
 421         }
 422
 423         if (u->policy->wants_amaf && u->playout_amaf_cutoff) {
 424                 unsigned int cutoff = amaf.game_baselen;
 425                 cutoff += (amaf.gamelen - amaf.game_baselen) * u->playout_amaf_cutoff / 100;
 426                 amaf.gamelen = cutoff;
 427         }
 428
 429         /* Record the result. */
 430
 431         assert(n == t->root || n->parent);
 432         floating_t rval = scale_value(u, b, result);
 433         u->policy->update(u->policy, t, n, node_color, player_color, &amaf, &b2, rval);
 434
 435         if (t->use_extra_komi) {
 436                 stats_add_result(&u->dynkomi->score, result / 2, 1);
 437                 stats_add_result(&u->dynkomi->value, rval, 1);
 438         }
 439
 440         if (u->local_tree && n->parent && !is_pass(node_coord(n)) && dlen > 0) {
 441                 /* Get the local sequences and record them in ltree. */
 442                 /* We will look for sequence starts in our descent
 443                  * history, then run record_local_sequence() for each
 444                  * found sequence start; record_local_sequence() may
 445                  * pick longer sequences from descent history then,
 446                  * which is expected as it will create new lnodes. */
 447                 enum stone seq_color = player_color;
 448                 /* First move always starts a sequence. */
 449                 record_local_sequence(u, t, &b2, descent, dlen, 1, seq_color);
 450                 seq_color = stone_other(seq_color);
 451                 for (int dseqi = 2; dseqi < dlen; dseqi++, seq_color = stone_other(seq_color)) {
 452                         if (u->local_tree_allseq) {
 453                                 /* We are configured to record all subsequences. */
 454                                 record_local_sequence(u, t, &b2, descent, dlen, dseqi, seq_color);
 455                                 continue;
 456                         }
 457                         if (descent[dseqi].node->d >= u->tenuki_d) {
 458                                 /* Tenuki! Record the fresh sequence. */
 459                                 record_local_sequence(u, t, &b2, descent, dlen, dseqi, seq_color);
 460                                 continue;
 461                         }
 462                         if (descent[dseqi].lnode && !descent[dseqi].lnode) {
 463                                 /* Record result for in-descent picked sequence. */
 464                                 record_local_sequence(u, t, &b2, descent, dlen, dseqi, seq_color);
 465                                 continue;
 466                         }
 467                 }
 468         }
 469
 470 end:
 471         /* We need to undo the virtual loss we added during descend. */
 472         if (u->virtual_loss) {
 473                 floating_t loss = node_color == S_BLACK ? 0.0 : 1.0;
 474                 for (; n->parent; n = n->parent) {
 475                         stats_rm_result(&n->u, loss, u->virtual_loss);
 476                         loss = 1.0 - loss;
 477                 }
 478         }
 479
 480         board_done_noalloc(&b2);
 481         return result;
 482 }
 483
 484 int
 485 uct_playouts(struct uct *u, struct board *b, enum stone color, struct tree *t, struct time_info *ti)
 486 {
 487         int i;
 488         if (ti && ti->dim == TD_GAMES) {
 489                 for (i = 0; t->root->u.playouts <= ti->len.games && !uct_halt; i++)
 490                         uct_playout(u, b, color, t);
 491         } else {
 492                 for (i = 0; !uct_halt; i++)
 493                         uct_playout(u, b, color, t);
 494         }
 495         return i;
 496 }