uct/uct.c

   1 #include <assert.h>
   2 #include <math.h>
   3 #include <stdio.h>
   4 #include <stdlib.h>
   5 #include <string.h>
   6 #include <time.h>
   7
   8 #define DEBUG
   9
  10 #include "debug.h"
  11 #include "board.h"
  12 #include "gtp.h"
  13 #include "move.h"
  14 #include "mq.h"
  15 #include "joseki/base.h"
  16 #include "playout.h"
  17 #include "playout/elo.h"
  18 #include "playout/moggy.h"
  19 #include "playout/light.h"
  20 #include "tactics/util.h"
  21 #include "timeinfo.h"
  22 #include "uct/dynkomi.h"
  23 #include "uct/internal.h"
  24 #include "uct/plugins.h"
  25 #include "uct/prior.h"
  26 #include "uct/search.h"
  27 #include "uct/slave.h"
  28 #include "uct/tree.h"
  29 #include "uct/uct.h"
  30 #include "uct/walk.h"
  31
  32 struct uct_policy *policy_ucb1_init(struct uct *u, char *arg);
  33 struct uct_policy *policy_ucb1amaf_init(struct uct *u, char *arg);
  34 static void uct_pondering_start(struct uct *u, struct board *b0, struct tree *t, enum stone color);
  35
  36 /* Maximal simulation length. */
  37 #define MC_GAMELEN      MAX_GAMELEN
  38
  39
  40 static void
  41 setup_state(struct uct *u, struct board *b, enum stone color)
  42 {
  43         u->t = tree_init(b, color, u->fast_alloc ? u->max_tree_size : 0,
  44                          u->max_pruned_size, u->pruning_threshold, u->local_tree_aging, u->stats_hbits);
  45         if (u->force_seed)
  46                 fast_srandom(u->force_seed);
  47         if (UDEBUGL(0))
  48                 fprintf(stderr, "Fresh board with random seed %lu\n", fast_getseed());
  49         //board_print(b, stderr);
  50         if (!u->no_book && b->moves == 0) {
  51                 assert(color == S_BLACK);
  52                 tree_load(u->t, b);
  53         }
  54 }
  55
  56 static void
  57 reset_state(struct uct *u)
  58 {
  59         assert(u->t);
  60         tree_done(u->t); u->t = NULL;
  61 }
  62
  63 static void
  64 setup_dynkomi(struct uct *u, struct board *b, enum stone to_play)
  65 {
  66         if (u->t->use_extra_komi && !u->pondering && u->dynkomi->permove)
  67                 u->t->extra_komi = u->dynkomi->permove(u->dynkomi, b, u->t);
  68         else if (!u->t->use_extra_komi)
  69                 u->t->extra_komi = 0;
  70 }
  71
  72 void
  73 uct_prepare_move(struct uct *u, struct board *b, enum stone color)
  74 {
  75         if (u->t) {
  76                 /* Verify that we have sane state. */
  77                 assert(b->es == u);
  78                 assert(u->t && b->moves);
  79                 if (color != stone_other(u->t->root_color)) {
  80                         fprintf(stderr, "Fatal: Non-alternating play detected %d %d\n",
  81                                 color, u->t->root_color);
  82                         exit(1);
  83                 }
  84                 uct_htable_reset(u->t);
  85
  86         } else {
  87                 /* We need fresh state. */
  88                 b->es = u;
  89                 setup_state(u, b, color);
  90         }
  91
  92         u->ownermap.playouts = 0;
  93         memset(u->ownermap.map, 0, board_size2(b) * sizeof(u->ownermap.map[0]));
  94         u->played_own = u->played_all = 0;
  95 }
  96
  97 static void
  98 dead_group_list(struct uct *u, struct board *b, struct move_queue *mq)
  99 {
 100         struct group_judgement gj;
 101         gj.thres = GJ_THRES;
 102         gj.gs = alloca(board_size2(b) * sizeof(gj.gs[0]));
 103         board_ownermap_judge_group(b, &u->ownermap, &gj);
 104         groups_of_status(b, &gj, GS_DEAD, mq);
 105 }
 106
 107 bool
 108 uct_pass_is_safe(struct uct *u, struct board *b, enum stone color, bool pass_all_alive)
 109 {
 110         if (u->ownermap.playouts < GJ_MINGAMES)
 111                 return false;
 112
 113         struct move_queue mq = { .moves = 0 };
 114         dead_group_list(u, b, &mq);
 115         if (pass_all_alive && mq.moves > 0)
 116                 return false; // We need to remove some dead groups first.
 117         return pass_is_safe(b, color, &mq);
 118 }
 119
 120 static char *
 121 uct_printhook_ownermap(struct board *board, coord_t c, char *s, char *end)
 122 {
 123         struct uct *u = board->es;
 124         assert(u);
 125         const char chr[] = ":XO,"; // dame, black, white, unclear
 126         const char chm[] = ":xo,";
 127         char ch = chr[board_ownermap_judge_point(&u->ownermap, c, GJ_THRES)];
 128         if (ch == ',') { // less precise estimate then?
 129                 ch = chm[board_ownermap_judge_point(&u->ownermap, c, 0.67)];
 130         }
 131         s += snprintf(s, end - s, "%c ", ch);
 132         return s;
 133 }
 134
 135 static char *
 136 uct_notify_play(struct engine *e, struct board *b, struct move *m)
 137 {
 138         struct uct *u = e->data;
 139         if (!u->t) {
 140                 /* No state, create one - this is probably game beginning
 141                  * and we need to load the opening book right now. */
 142                 uct_prepare_move(u, b, m->color);
 143                 assert(u->t);
 144         }
 145
 146         /* Stop pondering, required by tree_promote_at() */
 147         uct_pondering_stop(u);
 148         if (UDEBUGL(2) && u->slave)
 149                 tree_dump(u->t, u->dumpthres);
 150
 151         if (is_resign(m->coord)) {
 152                 /* Reset state. */
 153                 reset_state(u);
 154                 return NULL;
 155         }
 156
 157         /* Promote node of the appropriate move to the tree root. */
 158         assert(u->t->root);
 159         if (!tree_promote_at(u->t, b, m->coord)) {
 160                 if (UDEBUGL(0))
 161                         fprintf(stderr, "Warning: Cannot promote move node! Several play commands in row?\n");
 162                 reset_state(u);
 163                 return NULL;
 164         }
 165
 166         /* If we are a slave in a distributed engine, start pondering once
 167          * we know which move we actually played. See uct_genmove() about
 168          * the check for pass. */
 169         if (u->pondering_opt && u->slave && m->color == u->my_color && !is_pass(m->coord))
 170                 uct_pondering_start(u, b, u->t, stone_other(m->color));
 171
 172         return NULL;
 173 }
 174
 175 static char *
 176 uct_result(struct engine *e, struct board *b)
 177 {
 178         struct uct *u = e->data;
 179         static char reply[1024];
 180
 181         if (!u->t)
 182                 return NULL;
 183         enum stone color = u->t->root_color;
 184         struct tree_node *n = u->t->root;
 185         snprintf(reply, 1024, "%s %s %d %.2f %.1f",
 186                  stone2str(color), coord2sstr(n->coord, b),
 187                  n->u.playouts, tree_node_get_value(u->t, -1, n->u.value),
 188                  u->t->use_extra_komi ? u->t->extra_komi : 0);
 189         return reply;
 190 }
 191
 192 static char *
 193 uct_chat(struct engine *e, struct board *b, char *cmd)
 194 {
 195         struct uct *u = e->data;
 196         static char reply[1024];
 197
 198         cmd += strspn(cmd, " \n\t");
 199         if (!strncasecmp(cmd, "winrate", 7)) {
 200                 if (!u->t)
 201                         return "no game context (yet?)";
 202                 enum stone color = u->t->root_color;
 203                 struct tree_node *n = u->t->root;
 204                 snprintf(reply, 1024, "In %d playouts at %d threads, %s %s can win with %.2f%% probability",
 205                          n->u.playouts, u->threads, stone2str(color), coord2sstr(n->coord, b),
 206                          tree_node_get_value(u->t, -1, n->u.value) * 100);
 207                 if (u->t->use_extra_komi && abs(u->t->extra_komi) >= 0.5) {
 208                         sprintf(reply + strlen(reply), ", while self-imposing extra komi %.1f",
 209                                 u->t->extra_komi);
 210                 }
 211                 strcat(reply, ".");
 212                 return reply;
 213         }
 214         return NULL;
 215 }
 216
 217 static void
 218 uct_dead_group_list(struct engine *e, struct board *b, struct move_queue *mq)
 219 {
 220         struct uct *u = e->data;
 221
 222         /* This means the game is probably over, no use pondering on. */
 223         uct_pondering_stop(u);
 224
 225         if (u->pass_all_alive)
 226                 return; // no dead groups
 227
 228         bool mock_state = false;
 229
 230         if (!u->t) {
 231                 /* No state, but we cannot just back out - we might
 232                  * have passed earlier, only assuming some stones are
 233                  * dead, and then re-connected, only to lose counting
 234                  * when all stones are assumed alive. */
 235                 uct_prepare_move(u, b, S_BLACK); assert(u->t);
 236                 mock_state = true;
 237         }
 238         /* Make sure the ownermap is well-seeded. */
 239         while (u->ownermap.playouts < GJ_MINGAMES)
 240                 uct_playout(u, b, S_BLACK, u->t);
 241         /* Show the ownermap: */
 242         if (DEBUGL(2))
 243                 board_print_custom(b, stderr, uct_printhook_ownermap);
 244
 245         dead_group_list(u, b, mq);
 246
 247         if (mock_state) {
 248                 /* Clean up the mock state in case we will receive
 249                  * a genmove; we could get a non-alternating-move
 250                  * error from uct_prepare_move() in that case otherwise. */
 251                 reset_state(u);
 252         }
 253 }
 254
 255 static void
 256 playout_policy_done(struct playout_policy *p)
 257 {
 258         if (p->done) p->done(p);
 259         if (p->data) free(p->data);
 260         free(p);
 261 }
 262
 263 static void
 264 uct_done(struct engine *e)
 265 {
 266         /* This is called on engine reset, especially when clear_board
 267          * is received and new game should begin. */
 268         struct uct *u = e->data;
 269         uct_pondering_stop(u);
 270         if (u->t) reset_state(u);
 271         free(u->ownermap.map);
 272
 273         free(u->policy);
 274         free(u->random_policy);
 275         playout_policy_done(u->playout);
 276         uct_prior_done(u->prior);
 277         joseki_done(u->jdict);
 278         pluginset_done(u->plugins);
 279 }
 280
 281
 282
 283 /* Run time-limited MCTS search on foreground. */
 284 static int
 285 uct_search(struct uct *u, struct board *b, struct time_info *ti, enum stone color, struct tree *t)
 286 {
 287         struct uct_search_state s;
 288         uct_search_start(u, b, color, t, ti, &s);
 289         if (UDEBUGL(2) && s.base_playouts > 0)
 290                 fprintf(stderr, "<pre-simulated %d games>\n", s.base_playouts);
 291
 292         /* The search tree is ctx->t. This is currently == . It is important
 293          * to reference ctx->t directly since the
 294          * thread manager will swap the tree pointer asynchronously. */
 295
 296         /* Now, just periodically poll the search tree. */
 297         while (1) {
 298                 time_sleep(TREE_BUSYWAIT_INTERVAL);
 299                 /* TREE_BUSYWAIT_INTERVAL should never be less than desired time, or the
 300                  * time control is broken. But if it happens to be less, we still search
 301                  * at least 100ms otherwise the move is completely random. */
 302
 303                 int i = uct_search_games(&s);
 304                 /* Print notifications etc. */
 305                 uct_search_progress(u, b, color, t, ti, &s, i);
 306                 /* Check if we should stop the search. */
 307                 if (uct_search_check_stop(u, b, color, t, ti, &s, i))
 308                         break;
 309         }
 310
 311         struct uct_thread_ctx *ctx = uct_search_stop();
 312         if (UDEBUGL(2)) tree_dump(t, u->dumpthres);
 313         if (UDEBUGL(2))
 314                 fprintf(stderr, "(avg score %f/%d value %f/%d)\n",
 315                         u->dynkomi->score.value, u->dynkomi->score.playouts,
 316                         u->dynkomi->value.value, u->dynkomi->value.playouts);
 317         if (UDEBUGL(0))
 318                 uct_progress_status(u, t, color, ctx->games);
 319
 320         u->played_own += ctx->games;
 321         return ctx->games;
 322 }
 323
 324 /* Start pondering background with @color to play. */
 325 static void
 326 uct_pondering_start(struct uct *u, struct board *b0, struct tree *t, enum stone color)
 327 {
 328         if (UDEBUGL(1))
 329                 fprintf(stderr, "Starting to ponder with color %s\n", stone2str(stone_other(color)));
 330         u->pondering = true;
 331
 332         /* We need a local board copy to ponder upon. */
 333         struct board *b = malloc2(sizeof(*b)); board_copy(b, b0);
 334
 335         /* *b0 did not have the genmove'd move played yet. */
 336         struct move m = { t->root->coord, t->root_color };
 337         int res = board_play(b, &m);
 338         assert(res >= 0);
 339         setup_dynkomi(u, b, stone_other(m.color));
 340
 341         /* Start MCTS manager thread "headless". */
 342         static struct uct_search_state s;
 343         uct_search_start(u, b, color, t, NULL, &s);
 344 }
 345
 346 /* uct_search_stop() frontend for the pondering (non-genmove) mode, and
 347  * to stop the background search for a slave in the distributed engine. */
 348 void
 349 uct_pondering_stop(struct uct *u)
 350 {
 351         if (!thread_manager_running)
 352                 return;
 353
 354         /* Stop the thread manager. */
 355         struct uct_thread_ctx *ctx = uct_search_stop();
 356         if (UDEBUGL(1)) {
 357                 if (u->pondering) fprintf(stderr, "(pondering) ");
 358                 uct_progress_status(u, ctx->t, ctx->color, ctx->games);
 359         }
 360         if (u->pondering) {
 361                 free(ctx->b);
 362                 u->pondering = false;
 363         }
 364 }
 365
 366
 367 void
 368 uct_genmove_setup(struct uct *u, struct board *b, enum stone color)
 369 {
 370         if (b->superko_violation) {
 371                 fprintf(stderr, "!!! WARNING: SUPERKO VIOLATION OCCURED BEFORE THIS MOVE\n");
 372                 fprintf(stderr, "Maybe you play with situational instead of positional superko?\n");
 373                 fprintf(stderr, "I'm going to ignore the violation, but note that I may miss\n");
 374                 fprintf(stderr, "some moves valid under this ruleset because of this.\n");
 375                 b->superko_violation = false;
 376         }
 377
 378         uct_prepare_move(u, b, color);
 379
 380         assert(u->t);
 381         u->my_color = color;
 382
 383         /* How to decide whether to use dynkomi in this game? Since we use
 384          * pondering, it's not simple "who-to-play" matter. Decide based on
 385          * the last genmove issued. */
 386         u->t->use_extra_komi = !!(u->dynkomi_mask & color);
 387         /* Moreover, we do not use extra komi at the game end - we are not
 388          * to fool ourselves at this point. */
 389         if (board_estimated_moves_left(b) <= MIN_MOVES_LEFT)
 390                 u->t->use_extra_komi = false;
 391         setup_dynkomi(u, b, color);
 392
 393         if (b->rules == RULES_JAPANESE)
 394                 u->territory_scoring = true;
 395
 396         /* Make pessimistic assumption about komi for Japanese rules to
 397          * avoid losing by 0.5 when winning by 0.5 with Chinese rules.
 398          * The rules usually give the same winner if the integer part of komi
 399          * is odd so we adjust the komi only if it is even (for a board of
 400          * odd size). We are not trying  to get an exact evaluation for rare
 401          * cases of seki. For details see http://home.snafu.de/jasiek/parity.html */
 402         if (u->territory_scoring && (((int)floor(b->komi) + board_size(b)) & 1)) {
 403                 b->komi += (color == S_BLACK ? 1.0 : -1.0);
 404                 if (UDEBUGL(0))
 405                         fprintf(stderr, "Setting komi to %.1f assuming Japanese rules\n",
 406                                 b->komi);
 407         }
 408 }
 409
 410 static coord_t *
 411 uct_genmove(struct engine *e, struct board *b, struct time_info *ti, enum stone color, bool pass_all_alive)
 412 {
 413         double start_time = time_now();
 414         struct uct *u = e->data;
 415         uct_pondering_stop(u);
 416         uct_genmove_setup(u, b, color);
 417
 418         /* Start the Monte Carlo Tree Search! */
 419         int base_playouts = u->t->root->u.playouts;
 420         int played_games = uct_search(u, b, ti, color, u->t);
 421
 422         coord_t best_coord;
 423         struct tree_node *best;
 424         best = uct_search_result(u, b, color, pass_all_alive, played_games, base_playouts, &best_coord);
 425
 426         if (UDEBUGL(2)) {
 427                 double time = time_now() - start_time + 0.000001; /* avoid divide by zero */
 428                 fprintf(stderr, "genmove in %0.2fs (%d games/s, %d games/s/thread)\n",
 429                         time, (int)(played_games/time), (int)(played_games/time/u->threads));
 430         }
 431
 432         if (!best) {
 433                 /* Pass or resign. */
 434                 reset_state(u);
 435                 return coord_copy(best_coord);
 436         }
 437         tree_promote_node(u->t, &best);
 438
 439         /* After a pass, pondering is harmful for two reasons:
 440          * (i) We might keep pondering even when the game is over.
 441          * Of course this is the case for opponent resign as well.
 442          * (ii) More importantly, the ownermap will get skewed since
 443          * the UCT will start cutting off any playouts. */
 444         if (u->pondering_opt && !is_pass(best->coord)) {
 445                 uct_pondering_start(u, b, u->t, stone_other(color));
 446         }
 447         return coord_copy(best_coord);
 448 }
 449
 450
 451 bool
 452 uct_genbook(struct engine *e, struct board *b, struct time_info *ti, enum stone color)
 453 {
 454         struct uct *u = e->data;
 455         if (!u->t) uct_prepare_move(u, b, color);
 456         assert(u->t);
 457
 458         if (ti->dim == TD_GAMES) {
 459                 /* Don't count in games that already went into the book. */
 460                 ti->len.games += u->t->root->u.playouts;
 461         }
 462         uct_search(u, b, ti, color, u->t);
 463
 464         assert(ti->dim == TD_GAMES);
 465         tree_save(u->t, b, ti->len.games / 100);
 466
 467         return true;
 468 }
 469
 470 void
 471 uct_dumpbook(struct engine *e, struct board *b, enum stone color)
 472 {
 473         struct uct *u = e->data;
 474         struct tree *t = tree_init(b, color, u->fast_alloc ? u->max_tree_size : 0,
 475                          u->max_pruned_size, u->pruning_threshold, u->local_tree_aging, 0);
 476         tree_load(t, b);
 477         tree_dump(t, 0);
 478         tree_done(t);
 479 }
 480
 481
 482 float
 483 uct_evaluate(struct engine *e, struct board *b, struct time_info *ti, coord_t c, enum stone color)
 484 {
 485         struct uct *u = e->data;
 486
 487         struct board b2;
 488         board_copy(&b2, b);
 489         struct move m = { c, color };
 490         int res = board_play(&b2, &m);
 491         if (res < 0)
 492                 return NAN;
 493         color = stone_other(color);
 494
 495         if (u->t) reset_state(u);
 496         uct_prepare_move(u, &b2, color);
 497         assert(u->t);
 498
 499         float bestval;
 500         uct_search(u, &b2, ti, color, u->t);
 501         struct tree_node *best = u->policy->choose(u->policy, u->t->root, &b2, color, resign);
 502         if (!best) {
 503                 bestval = NAN; // the opponent has no reply!
 504         } else {
 505                 bestval = tree_node_get_value(u->t, 1, best->u.value);
 506         }
 507
 508         reset_state(u); // clean our junk
 509
 510         return isnan(bestval) ? NAN : 1.0f - bestval;
 511 }
 512
 513
 514 struct uct *
 515 uct_state_init(char *arg, struct board *b)
 516 {
 517         struct uct *u = calloc2(1, sizeof(struct uct));
 518         bool using_elo = false;
 519
 520         u->debug_level = debug_level;
 521         u->gamelen = MC_GAMELEN;
 522         u->resign_threshold = 0.2;
 523         u->sure_win_threshold = 0.85;
 524         u->mercymin = 0;
 525         u->expand_p = 2;
 526         u->dumpthres = 1000;
 527         u->playout_amaf = true;
 528         u->playout_amaf_nakade = false;
 529         u->amaf_prior = false;
 530         u->max_tree_size = 3072ULL * 1048576;
 531         u->pruning_threshold = 0;
 532
 533         u->threads = 1;
 534         u->thread_model = TM_TREEVL;
 535         u->virtual_loss = true;
 536
 537         u->fuseki_end = 20; // max time at 361*20% = 72 moves (our 36th move, still 99 to play)
 538         u->yose_start = 40; // (100-40-25)*361/100/2 = 63 moves still to play by us then
 539         u->bestr_ratio = 0.02;
 540         // 2.5 is clearly too much, but seems to compensate well for overly stern time allocations.
 541         // TODO: Further tuning and experiments with better time allocation schemes.
 542         u->best2_ratio = 2.5;
 543
 544         u->val_scale = 0.04; u->val_points = 40;
 545         u->dynkomi_interval = 1000;
 546         u->dynkomi_mask = S_BLACK | S_WHITE;
 547
 548         u->tenuki_d = 4;
 549         u->local_tree_aging = 2;
 550
 551         u->plugins = pluginset_init(b);
 552
 553         u->jdict = joseki_load(b->size);
 554
 555         if (arg) {
 556                 char *optspec, *next = arg;
 557                 while (*next) {
 558                         optspec = next;
 559                         next += strcspn(next, ",");
 560                         if (*next) { *next++ = 0; } else { *next = 0; }
 561
 562                         char *optname = optspec;
 563                         char *optval = strchr(optspec, '=');
 564                         if (optval) *optval++ = 0;
 565
 566                         /** Basic options */
 567
 568                         if (!strcasecmp(optname, "debug")) {
 569                                 if (optval)
 570                                         u->debug_level = atoi(optval);
 571                                 else
 572                                         u->debug_level++;
 573                         } else if (!strcasecmp(optname, "dumpthres") && optval) {
 574                                 /* When dumping the UCT tree on output, include
 575                                  * nodes with at least this many playouts.
 576                                  * (This value is re-scaled "intelligently"
 577                                  * in case of very large trees.) */
 578                                 u->dumpthres = atoi(optval);
 579                         } else if (!strcasecmp(optname, "resign_threshold") && optval) {
 580                                 /* Resign when this ratio of games is lost
 581                                  * after GJ_MINGAMES sample is taken. */
 582                                 u->resign_threshold = atof(optval);
 583                         } else if (!strcasecmp(optname, "sure_win_threshold") && optval) {
 584                                 /* Stop reading when this ratio of games is won
 585                                  * after PLAYOUT_EARLY_BREAK_MIN sample is
 586                                  * taken. (Prevents stupid time losses,
 587                                  * friendly to human opponents.) */
 588                                 u->sure_win_threshold = atof(optval);
 589                         } else if (!strcasecmp(optname, "force_seed") && optval) {
 590                                 /* Set RNG seed at the tree setup. */
 591                                 u->force_seed = atoi(optval);
 592                         } else if (!strcasecmp(optname, "no_book")) {
 593                                 /* Disable UCT opening book. */
 594                                 u->no_book = true;
 595                         } else if (!strcasecmp(optname, "pass_all_alive")) {
 596                                 /* Whether to consider passing only after all
 597                                  * dead groups were removed from the board;
 598                                  * this is like all genmoves are in fact
 599                                  * kgs-genmove_cleanup. */
 600                                 u->pass_all_alive = !optval || atoi(optval);
 601                         } else if (!strcasecmp(optname, "territory_scoring")) {
 602                                 /* Use territory scoring (default is area scoring).
 603                                  * An explicit kgs-rules command overrides this. */
 604                                 u->territory_scoring = !optval || atoi(optval);
 605                         } else if (!strcasecmp(optname, "banner") && optval) {
 606                                 /* Additional banner string. This must come as the
 607                                  * last engine parameter. */
 608                                 if (*next) *--next = ',';
 609                                 u->banner = strdup(optval);
 610                                 break;
 611                         } else if (!strcasecmp(optname, "plugin") && optval) {
 612                                 /* Load an external plugin; filename goes before the colon,
 613                                  * extra arguments after the colon. */
 614                                 char *pluginarg = strchr(optval, ':');
 615                                 if (pluginarg)
 616                                         *pluginarg++ = 0;
 617                                 plugin_load(u->plugins, optval, pluginarg);
 618
 619                         /** UCT behavior and policies */
 620
 621                         } else if ((!strcasecmp(optname, "policy")
 622                                 /* Node selection policy. ucb1amaf is the
 623                                  * default policy implementing RAVE, while
 624                                  * ucb1 is the simple exploration/exploitation
 625                                  * policy. Policies can take further extra
 626                                  * options. */
 627                                     || !strcasecmp(optname, "random_policy")) && optval) {
 628                                 /* A policy to be used randomly with small
 629                                  * chance instead of the default policy. */
 630                                 char *policyarg = strchr(optval, ':');
 631                                 struct uct_policy **p = !strcasecmp(optname, "policy") ? &u->policy : &u->random_policy;
 632                                 if (policyarg)
 633                                         *policyarg++ = 0;
 634                                 if (!strcasecmp(optval, "ucb1")) {
 635                                         *p = policy_ucb1_init(u, policyarg);
 636                                 } else if (!strcasecmp(optval, "ucb1amaf")) {
 637                                         *p = policy_ucb1amaf_init(u, policyarg);
 638                                 } else {
 639                                         fprintf(stderr, "UCT: Invalid tree policy %s\n", optval);
 640                                         exit(1);
 641                                 }
 642                         } else if (!strcasecmp(optname, "playout") && optval) {
 643                                 /* Random simulation (playout) policy.
 644                                  * moggy is the default policy with large
 645                                  * amount of domain-specific knowledge and
 646                                  * heuristics. light is a simple uniformly
 647                                  * random move selection policy. */
 648                                 char *playoutarg = strchr(optval, ':');
 649                                 if (playoutarg)
 650                                         *playoutarg++ = 0;
 651                                 if (!strcasecmp(optval, "moggy")) {
 652                                         u->playout = playout_moggy_init(playoutarg, b, u->jdict);
 653                                 } else if (!strcasecmp(optval, "light")) {
 654                                         u->playout = playout_light_init(playoutarg, b);
 655                                 } else if (!strcasecmp(optval, "elo")) {
 656                                         u->playout = playout_elo_init(playoutarg, b);
 657                                         using_elo = true;
 658                                 } else {
 659                                         fprintf(stderr, "UCT: Invalid playout policy %s\n", optval);
 660                                         exit(1);
 661                                 }
 662                         } else if (!strcasecmp(optname, "prior") && optval) {
 663                                 /* Node priors policy. When expanding a node,
 664                                  * it will seed node values heuristically
 665                                  * (most importantly, based on playout policy
 666                                  * opinion, but also with regard to other
 667                                  * things). See uct/prior.c for details.
 668                                  * Use prior=eqex=0 to disable priors. */
 669                                 u->prior = uct_prior_init(optval, b);
 670                         } else if (!strcasecmp(optname, "mercy") && optval) {
 671                                 /* Minimal difference of black/white captures
 672                                  * to stop playout - "Mercy Rule". Speeds up
 673                                  * hopeless playouts at the expense of some
 674                                  * accuracy. */
 675                                 u->mercymin = atoi(optval);
 676                         } else if (!strcasecmp(optname, "gamelen") && optval) {
 677                                 /* Maximum length of single simulation
 678                                  * in moves. */
 679                                 u->gamelen = atoi(optval);
 680                         } else if (!strcasecmp(optname, "expand_p") && optval) {
 681                                 /* Expand UCT nodes after it has been
 682                                  * visited this many times. */
 683                                 u->expand_p = atoi(optval);
 684                         } else if (!strcasecmp(optname, "random_policy_chance") && optval) {
 685                                 /* If specified (N), with probability 1/N, random_policy policy
 686                                  * descend is used instead of main policy descend; useful
 687                                  * if specified policy (e.g. UCB1AMAF) can make unduly biased
 688                                  * choices sometimes, you can fall back to e.g.
 689                                  * random_policy=UCB1. */
 690                                 u->random_policy_chance = atoi(optval);
 691
 692                         /** General AMAF behavior */
 693                         /* (Only relevant if the policy supports AMAF.
 694                          * More variables can be tuned as policy
 695                          * parameters.) */
 696
 697                         } else if (!strcasecmp(optname, "playout_amaf")) {
 698                                 /* Whether to include random playout moves in
 699                                  * AMAF as well. (Otherwise, only tree moves
 700                                  * are included in AMAF. Of course makes sense
 701                                  * only in connection with an AMAF policy.) */
 702                                 /* with-without: 55.5% (+-4.1) */
 703                                 if (optval && *optval == '0')
 704                                         u->playout_amaf = false;
 705                                 else
 706                                         u->playout_amaf = true;
 707                         } else if (!strcasecmp(optname, "playout_amaf_nakade")) {
 708                                 /* Whether to include nakade moves from playouts
 709                                  * in the AMAF statistics; this tends to nullify
 710                                  * the playout_amaf effect by adding too much
 711                                  * noise. */
 712                                 if (optval && *optval == '0')
 713                                         u->playout_amaf_nakade = false;
 714                                 else
 715                                         u->playout_amaf_nakade = true;
 716                         } else if (!strcasecmp(optname, "playout_amaf_cutoff") && optval) {
 717                                 /* Keep only first N% of playout stage AMAF
 718                                  * information. */
 719                                 u->playout_amaf_cutoff = atoi(optval);
 720                         } else if (!strcasecmp(optname, "amaf_prior") && optval) {
 721                                 /* In node policy, consider prior values
 722                                  * part of the real result term or part
 723                                  * of the AMAF term? */
 724                                 u->amaf_prior = atoi(optval);
 725
 726                         /** Performance and memory management */
 727
 728                         } else if (!strcasecmp(optname, "threads") && optval) {
 729                                 /* By default, Pachi will run with only single
 730                                  * tree search thread! */
 731                                 u->threads = atoi(optval);
 732                         } else if (!strcasecmp(optname, "thread_model") && optval) {
 733                                 if (!strcasecmp(optval, "tree")) {
 734                                         /* Tree parallelization - all threads
 735                                          * grind on the same tree. */
 736                                         u->thread_model = TM_TREE;
 737                                         u->virtual_loss = false;
 738                                 } else if (!strcasecmp(optval, "treevl")) {
 739                                         /* Tree parallelization, but also
 740                                          * with virtual losses - this discou-
 741                                          * rages most threads choosing the
 742                                          * same tree branches to read. */
 743                                         u->thread_model = TM_TREEVL;
 744                                         u->virtual_loss = true;
 745                                 } else {
 746                                         fprintf(stderr, "UCT: Invalid thread model %s\n", optval);
 747                                         exit(1);
 748                                 }
 749                         } else if (!strcasecmp(optname, "pondering")) {
 750                                 /* Keep searching even during opponent's turn. */
 751                                 u->pondering_opt = !optval || atoi(optval);
 752                         } else if (!strcasecmp(optname, "max_tree_size") && optval) {
 753                                 /* Maximum amount of memory [MiB] consumed by the move tree.
 754                                  * For fast_alloc it includes the temp tree used for pruning.
 755                                  * Default is 3072 (3 GiB). */
 756                                 u->max_tree_size = atol(optval) * 1048576;
 757                         } else if (!strcasecmp(optname, "fast_alloc")) {
 758                                 u->fast_alloc = !optval || atoi(optval);
 759                         } else if (!strcasecmp(optname, "pruning_threshold") && optval) {
 760                                 /* Force pruning at beginning of a move if the tree consumes
 761                                  * more than this [MiB]. Default is 10% of max_tree_size.
 762                                  * Increase to reduce pruning time overhead if memory is plentiful.
 763                                  * This option is meaningful only for fast_alloc. */
 764                                 u->pruning_threshold = atol(optval) * 1048576;
 765
 766                         /** Time control */
 767
 768                         } else if (!strcasecmp(optname, "best2_ratio") && optval) {
 769                                 /* If set, prolong simulating while
 770                                  * first_best/second_best playouts ratio
 771                                  * is less than best2_ratio. */
 772                                 u->best2_ratio = atof(optval);
 773                         } else if (!strcasecmp(optname, "bestr_ratio") && optval) {
 774                                 /* If set, prolong simulating while
 775                                  * best,best_best_child values delta
 776                                  * is more than bestr_ratio. */
 777                                 u->bestr_ratio = atof(optval);
 778                         } else if (!strcasecmp(optname, "fuseki_end") && optval) {
 779                                 /* At the very beginning it's not worth thinking
 780                                  * too long because the playout evaluations are
 781                                  * very noisy. So gradually increase the thinking
 782                                  * time up to maximum when fuseki_end percent
 783                                  * of the board has been played.
 784                                  * This only applies if we are not in byoyomi. */
 785                                 u->fuseki_end = atoi(optval);
 786                         } else if (!strcasecmp(optname, "yose_start") && optval) {
 787                                 /* When yose_start percent of the board has been
 788                                  * played, or if we are in byoyomi, stop spending
 789                                  * more time and spread the remaining time
 790                                  * uniformly.
 791                                  * Between fuseki_end and yose_start, we spend
 792                                  * a constant proportion of the remaining time
 793                                  * on each move. (yose_start should actually
 794                                  * be much earlier than when real yose start,
 795                                  * but "yose" is a good short name to convey
 796                                  * the idea.) */
 797                                 u->yose_start = atoi(optval);
 798
 799                         /** Dynamic komi */
 800
 801                         } else if (!strcasecmp(optname, "dynkomi") && optval) {
 802                                 /* Dynamic komi approach; there are multiple
 803                                  * ways to adjust komi dynamically throughout
 804                                  * play. We currently support two: */
 805                                 char *dynkomiarg = strchr(optval, ':');
 806                                 if (dynkomiarg)
 807                                         *dynkomiarg++ = 0;
 808                                 if (!strcasecmp(optval, "none")) {
 809                                         u->dynkomi = uct_dynkomi_init_none(u, dynkomiarg, b);
 810                                 } else if (!strcasecmp(optval, "linear")) {
 811                                         /* You should set dynkomi_mask=1
 812                                          * since this doesn't work well
 813                                          * for white handicaps! */
 814                                         u->dynkomi = uct_dynkomi_init_linear(u, dynkomiarg, b);
 815                                 } else if (!strcasecmp(optval, "adaptive")) {
 816                                         /* There are many more knobs to
 817                                          * crank - see uct/dynkomi.c. */
 818                                         u->dynkomi = uct_dynkomi_init_adaptive(u, dynkomiarg, b);
 819                                 } else {
 820                                         fprintf(stderr, "UCT: Invalid dynkomi mode %s\n", optval);
 821                                         exit(1);
 822                                 }
 823                         } else if (!strcasecmp(optname, "dynkomi_mask") && optval) {
 824                                 /* Bitmask of colors the player must be
 825                                  * for dynkomi be applied; you may want
 826                                  * to use dynkomi_mask=3 to allow dynkomi
 827                                  * even in games where Pachi is white. */
 828                                 u->dynkomi_mask = atoi(optval);
 829                         } else if (!strcasecmp(optname, "dynkomi_interval") && optval) {
 830                                 /* If non-zero, re-adjust dynamic komi
 831                                  * throughout a single genmove reading,
 832                                  * roughly every N simulations. */
 833                                 /* XXX: Does not work with tree
 834                                  * parallelization. */
 835                                 u->dynkomi_interval = atoi(optval);
 836
 837                         /** Node value result scaling */
 838
 839                         } else if (!strcasecmp(optname, "val_scale") && optval) {
 840                                 /* How much of the game result value should be
 841                                  * influenced by win size. Zero means it isn't. */
 842                                 u->val_scale = atof(optval);
 843                         } else if (!strcasecmp(optname, "val_points") && optval) {
 844                                 /* Maximum size of win to be scaled into game
 845                                  * result value. Zero means boardsize^2. */
 846                                 u->val_points = atoi(optval) * 2; // result values are doubled
 847                         } else if (!strcasecmp(optname, "val_extra")) {
 848                                 /* If false, the score coefficient will be simply
 849                                  * added to the value, instead of scaling the result
 850                                  * coefficient because of it. */
 851                                 u->val_extra = !optval || atoi(optval);
 852
 853                         /** Local trees */
 854                         /* (Purely experimental. Does not work - yet!) */
 855
 856                         } else if (!strcasecmp(optname, "local_tree") && optval) {
 857                                 /* Whether to bias exploration by local tree values
 858                                  * (must be supported by the used policy).
 859                                  * 0: Don't.
 860                                  * 1: Do, value = result.
 861                                  * Try to temper the result:
 862                                  * 2: Do, value = 0.5+(result-expected)/2.
 863                                  * 3: Do, value = 0.5+bzz((result-expected)^2).
 864                                  * 4: Do, value = 0.5+sqrt(result-expected)/2. */
 865                                 u->local_tree = atoi(optval);
 866                         } else if (!strcasecmp(optname, "tenuki_d") && optval) {
 867                                 /* Tenuki distance at which to break the local tree. */
 868                                 u->tenuki_d = atoi(optval);
 869                                 if (u->tenuki_d > TREE_NODE_D_MAX + 1) {
 870                                         fprintf(stderr, "uct: tenuki_d must not be larger than TREE_NODE_D_MAX+1 %d\n", TREE_NODE_D_MAX + 1);
 871                                         exit(1);
 872                                 }
 873                         } else if (!strcasecmp(optname, "local_tree_aging") && optval) {
 874                                 /* How much to reduce local tree values between moves. */
 875                                 u->local_tree_aging = atof(optval);
 876                         } else if (!strcasecmp(optname, "local_tree_allseq")) {
 877                                 /* By default, only complete sequences are stored
 878                                  * in the local tree. If this is on, also
 879                                  * subsequences starting at each move are stored. */
 880                                 u->local_tree_allseq = !optval || atoi(optval);
 881                         } else if (!strcasecmp(optname, "local_tree_playout")) {
 882                                 /* Whether to adjust ELO playout probability
 883                                  * distributions according to matched localtree
 884                                  * information. */
 885                                 u->local_tree_playout = !optval || atoi(optval);
 886                         } else if (!strcasecmp(optname, "local_tree_pseqroot")) {
 887                                 /* By default, when we have no sequence move
 888                                  * to suggest in-playout, we give up. If this
 889                                  * is on, we make probability distribution from
 890                                  * sequences first moves instead. */
 891                                 u->local_tree_pseqroot = !optval || atoi(optval);
 892
 893                         /** Distributed engine slaves setup */
 894
 895                         } else if (!strcasecmp(optname, "slave")) {
 896                                 /* Act as slave for the distributed engine. */
 897                                 u->slave = !optval || atoi(optval);
 898                         } else if (!strcasecmp(optname, "shared_nodes") && optval) {
 899                                 /* Share at most shared_nodes between master and slave at each genmoves.
 900                                  * Must use the same value in master and slaves. */
 901                                 u->shared_nodes = atoi(optval);
 902                         } else if (!strcasecmp(optname, "shared_levels") && optval) {
 903                                 /* Share only nodes of level <= shared_levels. */
 904                                 u->shared_levels = atoi(optval);
 905                         } else if (!strcasecmp(optname, "stats_hbits") && optval) {
 906                                 /* Set hash table size to 2^stats_hbits for the shared stats. */
 907                                 u->stats_hbits = atoi(optval);
 908
 909                         } else {
 910                                 fprintf(stderr, "uct: Invalid engine argument %s or missing value\n", optname);
 911                                 exit(1);
 912                         }
 913                 }
 914         }
 915
 916         if (!u->policy)
 917                 u->policy = policy_ucb1amaf_init(u, NULL);
 918
 919         if (!!u->random_policy_chance ^ !!u->random_policy) {
 920                 fprintf(stderr, "uct: Only one of random_policy and random_policy_chance is set\n");
 921                 exit(1);
 922         }
 923
 924         if (!u->local_tree) {
 925                 /* No ltree aging. */
 926                 u->local_tree_aging = 1.0f;
 927         }
 928         if (!using_elo)
 929                 u->local_tree_playout = false;
 930
 931         if (u->fast_alloc) {
 932                 if (u->pruning_threshold < u->max_tree_size / 10)
 933                         u->pruning_threshold = u->max_tree_size / 10;
 934                 if (u->pruning_threshold > u->max_tree_size / 2)
 935                         u->pruning_threshold = u->max_tree_size / 2;
 936
 937                 /* Limit pruning temp space to 20% of memory. Beyond this we discard
 938                  * the nodes and recompute them at the next move if necessary. */
 939                 u->max_pruned_size = u->max_tree_size / 5;
 940                 u->max_tree_size -= u->max_pruned_size;
 941         } else {
 942                 /* Reserve 5% memory in case the background free() are slower
 943                  * than the concurrent allocations. */
 944                 u->max_tree_size -= u->max_tree_size / 20;
 945         }
 946
 947         if (!u->prior)
 948                 u->prior = uct_prior_init(NULL, b);
 949
 950         if (!u->playout)
 951                 u->playout = playout_moggy_init(NULL, b, u->jdict);
 952         if (!u->playout->debug_level)
 953                 u->playout->debug_level = u->debug_level;
 954
 955         u->ownermap.map = malloc2(board_size2(b) * sizeof(u->ownermap.map[0]));
 956
 957         if (u->slave) {
 958                 if (!u->stats_hbits) u->stats_hbits = DEFAULT_STATS_HBITS;
 959                 if (!u->shared_nodes) u->shared_nodes = DEFAULT_SHARED_NODES;
 960                 assert(u->shared_levels * board_bits2(b) <= 8 * (int)sizeof(path_t));
 961         }
 962
 963         if (!u->dynkomi)
 964                 u->dynkomi = uct_dynkomi_init_adaptive(u, NULL, b);
 965
 966         /* Some things remain uninitialized for now - the opening book
 967          * is not loaded and the tree not set up. */
 968         /* This will be initialized in setup_state() at the first move
 969          * received/requested. This is because right now we are not aware
 970          * about any komi or handicap setup and such. */
 971
 972         return u;
 973 }
 974
 975 struct engine *
 976 engine_uct_init(char *arg, struct board *b)
 977 {
 978         struct uct *u = uct_state_init(arg, b);
 979         struct engine *e = calloc2(1, sizeof(struct engine));
 980         e->name = "UCT Engine";
 981         e->printhook = uct_printhook_ownermap;
 982         e->notify_play = uct_notify_play;
 983         e->chat = uct_chat;
 984         e->result = uct_result;
 985         e->genmove = uct_genmove;
 986         e->genmoves = uct_genmoves;
 987         e->dead_group_list = uct_dead_group_list;
 988         e->done = uct_done;
 989         e->data = u;
 990         if (u->slave)
 991                 e->notify = uct_notify;
 992
 993         const char banner[] = "I'm playing UCT. When I'm losing, I will resign, "
 994                 "if I think I win, I play until you pass. "
 995                 "Anyone can send me 'winrate' in private chat to get my assessment of the position.";
 996         if (!u->banner) u->banner = "";
 997         e->comment = malloc2(sizeof(banner) + strlen(u->banner) + 1);
 998         sprintf(e->comment, "%s %s", banner, u->banner);
 999
1000         return e;
1001 }