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/moggy.h"
  18 #include "playout/light.h"
  19 #include "tactics/util.h"
  20 #include "timeinfo.h"
  21 #include "uct/dynkomi.h"
  22 #include "uct/internal.h"
  23 #include "uct/plugins.h"
  24 #include "uct/prior.h"
  25 #include "uct/search.h"
  26 #include "uct/slave.h"
  27 #include "uct/tree.h"
  28 #include "uct/uct.h"
  29 #include "uct/walk.h"
  30
  31 struct uct_policy *policy_ucb1_init(struct uct *u, char *arg);
  32 struct uct_policy *policy_ucb1amaf_init(struct uct *u, char *arg);
  33 static void uct_pondering_start(struct uct *u, struct board *b0, struct tree *t, enum stone color);
  34
  35 /* Maximal simulation length. */
  36 #define MC_GAMELEN      MAX_GAMELEN
  37
  38
  39 static void
  40 setup_state(struct uct *u, struct board *b, enum stone color)
  41 {
  42         u->t = tree_init(b, color, u->fast_alloc ? u->max_tree_size : 0,
  43                          u->max_pruned_size, u->pruning_threshold, u->local_tree_aging, u->stats_hbits);
  44         if (u->force_seed)
  45                 fast_srandom(u->force_seed);
  46         if (UDEBUGL(0))
  47                 fprintf(stderr, "Fresh board with random seed %lu\n", fast_getseed());
  48         //board_print(b, stderr);
  49         if (!u->no_tbook && b->moves == 0) {
  50                 assert(color == S_BLACK);
  51                 tree_load(u->t, b);
  52         }
  53 }
  54
  55 static void
  56 reset_state(struct uct *u)
  57 {
  58         assert(u->t);
  59         tree_done(u->t); u->t = NULL;
  60 }
  61
  62 static void
  63 setup_dynkomi(struct uct *u, struct board *b, enum stone to_play)
  64 {
  65         if (u->t->use_extra_komi && !u->pondering && u->dynkomi->permove)
  66                 u->t->extra_komi = u->dynkomi->permove(u->dynkomi, b, u->t);
  67         else if (!u->t->use_extra_komi)
  68                 u->t->extra_komi = 0;
  69 }
  70
  71 void
  72 uct_prepare_move(struct uct *u, struct board *b, enum stone color)
  73 {
  74         if (u->t) {
  75                 /* Verify that we have sane state. */
  76                 assert(b->es == u);
  77                 assert(u->t && b->moves);
  78                 if (color != stone_other(u->t->root_color)) {
  79                         fprintf(stderr, "Fatal: Non-alternating play detected %d %d\n",
  80                                 color, u->t->root_color);
  81                         exit(1);
  82                 }
  83                 uct_htable_reset(u->t);
  84
  85         } else {
  86                 /* We need fresh state. */
  87                 b->es = u;
  88                 setup_state(u, b, color);
  89         }
  90
  91         u->ownermap.playouts = 0;
  92         memset(u->ownermap.map, 0, board_size2(b) * sizeof(u->ownermap.map[0]));
  93         u->played_own = u->played_all = 0;
  94 }
  95
  96 static void
  97 dead_group_list(struct uct *u, struct board *b, struct move_queue *mq)
  98 {
  99         struct group_judgement gj;
 100         gj.thres = GJ_THRES;
 101         gj.gs = alloca(board_size2(b) * sizeof(gj.gs[0]));
 102         board_ownermap_judge_group(b, &u->ownermap, &gj);
 103         groups_of_status(b, &gj, GS_DEAD, mq);
 104 }
 105
 106 bool
 107 uct_pass_is_safe(struct uct *u, struct board *b, enum stone color, bool pass_all_alive)
 108 {
 109         if (u->ownermap.playouts < GJ_MINGAMES)
 110                 return false;
 111
 112         struct move_queue mq = { .moves = 0 };
 113         dead_group_list(u, b, &mq);
 114         if (pass_all_alive && mq.moves > 0)
 115                 return false; // We need to remove some dead groups first.
 116         return pass_is_safe(b, color, &mq);
 117 }
 118
 119 static char *
 120 uct_printhook_ownermap(struct board *board, coord_t c, char *s, char *end)
 121 {
 122         struct uct *u = board->es;
 123         if (!u) {
 124                 strcat(s, ". ");
 125                 return s + 2;
 126         }
 127         const char chr[] = ":XO,"; // dame, black, white, unclear
 128         const char chm[] = ":xo,";
 129         char ch = chr[board_ownermap_judge_point(&u->ownermap, c, GJ_THRES)];
 130         if (ch == ',') { // less precise estimate then?
 131                 ch = chm[board_ownermap_judge_point(&u->ownermap, c, 0.67)];
 132         }
 133         s += snprintf(s, end - s, "%c ", ch);
 134         return s;
 135 }
 136
 137 static char *
 138 uct_notify_play(struct engine *e, struct board *b, struct move *m)
 139 {
 140         struct uct *u = e->data;
 141         if (!u->t) {
 142                 /* No state, create one - this is probably game beginning
 143                  * and we need to load the opening tbook right now. */
 144                 uct_prepare_move(u, b, m->color);
 145                 assert(u->t);
 146         }
 147
 148         /* Stop pondering, required by tree_promote_at() */
 149         uct_pondering_stop(u);
 150         if (UDEBUGL(2) && u->slave)
 151                 tree_dump(u->t, u->dumpthres);
 152
 153         if (is_resign(m->coord)) {
 154                 /* Reset state. */
 155                 reset_state(u);
 156                 return NULL;
 157         }
 158
 159         /* Promote node of the appropriate move to the tree root. */
 160         assert(u->t->root);
 161         if (!tree_promote_at(u->t, b, m->coord)) {
 162                 if (UDEBUGL(0))
 163                         fprintf(stderr, "Warning: Cannot promote move node! Several play commands in row?\n");
 164                 reset_state(u);
 165                 return NULL;
 166         }
 167
 168         /* If we are a slave in a distributed engine, start pondering once
 169          * we know which move we actually played. See uct_genmove() about
 170          * the check for pass. */
 171         if (u->pondering_opt && u->slave && m->color == u->my_color && !is_pass(m->coord))
 172                 uct_pondering_start(u, b, u->t, stone_other(m->color));
 173
 174         return NULL;
 175 }
 176
 177 static char *
 178 uct_result(struct engine *e, struct board *b)
 179 {
 180         struct uct *u = e->data;
 181         static char reply[1024];
 182
 183         if (!u->t)
 184                 return NULL;
 185         enum stone color = u->t->root_color;
 186         struct tree_node *n = u->t->root;
 187         snprintf(reply, 1024, "%s %s %d %.2f %.1f",
 188                  stone2str(color), coord2sstr(n->coord, b),
 189                  n->u.playouts, tree_node_get_value(u->t, -1, n->u.value),
 190                  u->t->use_extra_komi ? u->t->extra_komi : 0);
 191         return reply;
 192 }
 193
 194 static char *
 195 uct_chat(struct engine *e, struct board *b, char *cmd)
 196 {
 197         struct uct *u = e->data;
 198         static char reply[1024];
 199
 200         cmd += strspn(cmd, " \n\t");
 201         if (!strncasecmp(cmd, "winrate", 7)) {
 202                 if (!u->t)
 203                         return "no game context (yet?)";
 204                 enum stone color = u->t->root_color;
 205                 struct tree_node *n = u->t->root;
 206                 snprintf(reply, 1024, "In %d playouts at %d threads, %s %s can win with %.2f%% probability",
 207                          n->u.playouts, u->threads, stone2str(color), coord2sstr(n->coord, b),
 208                          tree_node_get_value(u->t, -1, n->u.value) * 100);
 209                 if (u->t->use_extra_komi && abs(u->t->extra_komi) >= 0.5) {
 210                         sprintf(reply + strlen(reply), ", while self-imposing extra komi %.1f",
 211                                 u->t->extra_komi);
 212                 }
 213                 strcat(reply, ".");
 214                 return reply;
 215         }
 216         return NULL;
 217 }
 218
 219 static void
 220 uct_dead_group_list(struct engine *e, struct board *b, struct move_queue *mq)
 221 {
 222         struct uct *u = e->data;
 223
 224         /* This means the game is probably over, no use pondering on. */
 225         uct_pondering_stop(u);
 226
 227         if (u->pass_all_alive)
 228                 return; // no dead groups
 229
 230         bool mock_state = false;
 231
 232         if (!u->t) {
 233                 /* No state, but we cannot just back out - we might
 234                  * have passed earlier, only assuming some stones are
 235                  * dead, and then re-connected, only to lose counting
 236                  * when all stones are assumed alive. */
 237                 uct_prepare_move(u, b, S_BLACK); assert(u->t);
 238                 mock_state = true;
 239         }
 240         /* Make sure the ownermap is well-seeded. */
 241         while (u->ownermap.playouts < GJ_MINGAMES)
 242                 uct_playout(u, b, S_BLACK, u->t);
 243         /* Show the ownermap: */
 244         if (DEBUGL(2))
 245                 board_print_custom(b, stderr, uct_printhook_ownermap);
 246
 247         dead_group_list(u, b, mq);
 248
 249         if (mock_state) {
 250                 /* Clean up the mock state in case we will receive
 251                  * a genmove; we could get a non-alternating-move
 252                  * error from uct_prepare_move() in that case otherwise. */
 253                 reset_state(u);
 254         }
 255 }
 256
 257 static void
 258 playout_policy_done(struct playout_policy *p)
 259 {
 260         if (p->done) p->done(p);
 261         if (p->data) free(p->data);
 262         free(p);
 263 }
 264
 265 static void
 266 uct_done(struct engine *e)
 267 {
 268         /* This is called on engine reset, especially when clear_board
 269          * is received and new game should begin. */
 270         struct uct *u = e->data;
 271         uct_pondering_stop(u);
 272         if (u->t) reset_state(u);
 273         free(u->ownermap.map);
 274
 275         free(u->policy);
 276         free(u->random_policy);
 277         playout_policy_done(u->playout);
 278         uct_prior_done(u->prior);
 279         joseki_done(u->jdict);
 280         pluginset_done(u->plugins);
 281 }
 282
 283
 284
 285 /* Run time-limited MCTS search on foreground. */
 286 static int
 287 uct_search(struct uct *u, struct board *b, struct time_info *ti, enum stone color, struct tree *t)
 288 {
 289         struct uct_search_state s;
 290         uct_search_start(u, b, color, t, ti, &s);
 291         if (UDEBUGL(2) && s.base_playouts > 0)
 292                 fprintf(stderr, "<pre-simulated %d games>\n", s.base_playouts);
 293
 294         /* The search tree is ctx->t. This is currently == . It is important
 295          * to reference ctx->t directly since the
 296          * thread manager will swap the tree pointer asynchronously. */
 297
 298         /* Now, just periodically poll the search tree. */
 299         /* Note that in case of TD_GAMES, threads will terminate independently
 300          * of the uct_search_check_stop() signalization. */
 301         while (1) {
 302                 time_sleep(TREE_BUSYWAIT_INTERVAL);
 303                 /* TREE_BUSYWAIT_INTERVAL should never be less than desired time, or the
 304                  * time control is broken. But if it happens to be less, we still search
 305                  * at least 100ms otherwise the move is completely random. */
 306
 307                 int i = uct_search_games(&s);
 308                 /* Print notifications etc. */
 309                 uct_search_progress(u, b, color, t, ti, &s, i);
 310                 /* Check if we should stop the search. */
 311                 if (uct_search_check_stop(u, b, color, t, ti, &s, i))
 312                         break;
 313         }
 314
 315         struct uct_thread_ctx *ctx = uct_search_stop();
 316         if (UDEBUGL(2)) tree_dump(t, u->dumpthres);
 317         if (UDEBUGL(2))
 318                 fprintf(stderr, "(avg score %f/%d value %f/%d)\n",
 319                         u->dynkomi->score.value, u->dynkomi->score.playouts,
 320                         u->dynkomi->value.value, u->dynkomi->value.playouts);
 321         if (UDEBUGL(0))
 322                 uct_progress_status(u, t, color, ctx->games);
 323
 324         u->played_own += ctx->games;
 325         return ctx->games;
 326 }
 327
 328 /* Start pondering background with @color to play. */
 329 static void
 330 uct_pondering_start(struct uct *u, struct board *b0, struct tree *t, enum stone color)
 331 {
 332         if (UDEBUGL(1))
 333                 fprintf(stderr, "Starting to ponder with color %s\n", stone2str(stone_other(color)));
 334         u->pondering = true;
 335
 336         /* We need a local board copy to ponder upon. */
 337         struct board *b = malloc2(sizeof(*b)); board_copy(b, b0);
 338
 339         /* *b0 did not have the genmove'd move played yet. */
 340         struct move m = { t->root->coord, t->root_color };
 341         int res = board_play(b, &m);
 342         assert(res >= 0);
 343         setup_dynkomi(u, b, stone_other(m.color));
 344
 345         /* Start MCTS manager thread "headless". */
 346         static struct uct_search_state s;
 347         uct_search_start(u, b, color, t, NULL, &s);
 348 }
 349
 350 /* uct_search_stop() frontend for the pondering (non-genmove) mode, and
 351  * to stop the background search for a slave in the distributed engine. */
 352 void
 353 uct_pondering_stop(struct uct *u)
 354 {
 355         if (!thread_manager_running)
 356                 return;
 357
 358         /* Stop the thread manager. */
 359         struct uct_thread_ctx *ctx = uct_search_stop();
 360         if (UDEBUGL(1)) {
 361                 if (u->pondering) fprintf(stderr, "(pondering) ");
 362                 uct_progress_status(u, ctx->t, ctx->color, ctx->games);
 363         }
 364         if (u->pondering) {
 365                 free(ctx->b);
 366                 u->pondering = false;
 367         }
 368 }
 369
 370
 371 void
 372 uct_genmove_setup(struct uct *u, struct board *b, enum stone color)
 373 {
 374         if (b->superko_violation) {
 375                 fprintf(stderr, "!!! WARNING: SUPERKO VIOLATION OCCURED BEFORE THIS MOVE\n");
 376                 fprintf(stderr, "Maybe you play with situational instead of positional superko?\n");
 377                 fprintf(stderr, "I'm going to ignore the violation, but note that I may miss\n");
 378                 fprintf(stderr, "some moves valid under this ruleset because of this.\n");
 379                 b->superko_violation = false;
 380         }
 381
 382         uct_prepare_move(u, b, color);
 383
 384         assert(u->t);
 385         u->my_color = color;
 386
 387         /* How to decide whether to use dynkomi in this game? Since we use
 388          * pondering, it's not simple "who-to-play" matter. Decide based on
 389          * the last genmove issued. */
 390         u->t->use_extra_komi = !!(u->dynkomi_mask & color);
 391         /* Moreover, we do not use extra komi at the game end - we are not
 392          * to fool ourselves at this point. */
 393         if (board_estimated_moves_left(b) <= MIN_MOVES_LEFT)
 394                 u->t->use_extra_komi = false;
 395         setup_dynkomi(u, b, color);
 396
 397         if (b->rules == RULES_JAPANESE)
 398                 u->territory_scoring = true;
 399
 400         /* Make pessimistic assumption about komi for Japanese rules to
 401          * avoid losing by 0.5 when winning by 0.5 with Chinese rules.
 402          * The rules usually give the same winner if the integer part of komi
 403          * is odd so we adjust the komi only if it is even (for a board of
 404          * odd size). We are not trying  to get an exact evaluation for rare
 405          * cases of seki. For details see http://home.snafu.de/jasiek/parity.html */
 406         if (u->territory_scoring && (((int)floor(b->komi) + board_size(b)) & 1)) {
 407                 b->komi += (color == S_BLACK ? 1.0 : -1.0);
 408                 if (UDEBUGL(0))
 409                         fprintf(stderr, "Setting komi to %.1f assuming Japanese rules\n",
 410                                 b->komi);
 411         }
 412 }
 413
 414 static coord_t *
 415 uct_genmove(struct engine *e, struct board *b, struct time_info *ti, enum stone color, bool pass_all_alive)
 416 {
 417         double start_time = time_now();
 418         struct uct *u = e->data;
 419         uct_pondering_stop(u);
 420         uct_genmove_setup(u, b, color);
 421
 422         /* Start the Monte Carlo Tree Search! */
 423         int base_playouts = u->t->root->u.playouts;
 424         int played_games = uct_search(u, b, ti, color, u->t);
 425
 426         coord_t best_coord;
 427         struct tree_node *best;
 428         best = uct_search_result(u, b, color, pass_all_alive, played_games, base_playouts, &best_coord);
 429
 430         if (UDEBUGL(2)) {
 431                 double time = time_now() - start_time + 0.000001; /* avoid divide by zero */
 432                 fprintf(stderr, "genmove in %0.2fs (%d games/s, %d games/s/thread)\n",
 433                         time, (int)(played_games/time), (int)(played_games/time/u->threads));
 434         }
 435
 436         if (!best) {
 437                 /* Pass or resign. */
 438                 reset_state(u);
 439                 return coord_copy(best_coord);
 440         }
 441         tree_promote_node(u->t, &best);
 442
 443         /* After a pass, pondering is harmful for two reasons:
 444          * (i) We might keep pondering even when the game is over.
 445          * Of course this is the case for opponent resign as well.
 446          * (ii) More importantly, the ownermap will get skewed since
 447          * the UCT will start cutting off any playouts. */
 448         if (u->pondering_opt && !is_pass(best->coord)) {
 449                 uct_pondering_start(u, b, u->t, stone_other(color));
 450         }
 451         return coord_copy(best_coord);
 452 }
 453
 454
 455 bool
 456 uct_gentbook(struct engine *e, struct board *b, struct time_info *ti, enum stone color)
 457 {
 458         struct uct *u = e->data;
 459         if (!u->t) uct_prepare_move(u, b, color);
 460         assert(u->t);
 461
 462         if (ti->dim == TD_GAMES) {
 463                 /* Don't count in games that already went into the tbook. */
 464                 ti->len.games += u->t->root->u.playouts;
 465         }
 466         uct_search(u, b, ti, color, u->t);
 467
 468         assert(ti->dim == TD_GAMES);
 469         tree_save(u->t, b, ti->len.games / 100);
 470
 471         return true;
 472 }
 473
 474 void
 475 uct_dumptbook(struct engine *e, struct board *b, enum stone color)
 476 {
 477         struct uct *u = e->data;
 478         struct tree *t = tree_init(b, color, u->fast_alloc ? u->max_tree_size : 0,
 479                          u->max_pruned_size, u->pruning_threshold, u->local_tree_aging, 0);
 480         tree_load(t, b);
 481         tree_dump(t, 0);
 482         tree_done(t);
 483 }
 484
 485
 486 floating_t
 487 uct_evaluate(struct engine *e, struct board *b, struct time_info *ti, coord_t c, enum stone color)
 488 {
 489         struct uct *u = e->data;
 490
 491         struct board b2;
 492         board_copy(&b2, b);
 493         struct move m = { c, color };
 494         int res = board_play(&b2, &m);
 495         if (res < 0)
 496                 return NAN;
 497         color = stone_other(color);
 498
 499         if (u->t) reset_state(u);
 500         uct_prepare_move(u, &b2, color);
 501         assert(u->t);
 502
 503         floating_t bestval;
 504         uct_search(u, &b2, ti, color, u->t);
 505         struct tree_node *best = u->policy->choose(u->policy, u->t->root, &b2, color, resign);
 506         if (!best) {
 507                 bestval = NAN; // the opponent has no reply!
 508         } else {
 509                 bestval = tree_node_get_value(u->t, 1, best->u.value);
 510         }
 511
 512         reset_state(u); // clean our junk
 513
 514         return isnan(bestval) ? NAN : 1.0f - bestval;
 515 }
 516
 517
 518 struct uct *
 519 uct_state_init(char *arg, struct board *b)
 520 {
 521         struct uct *u = calloc2(1, sizeof(struct uct));
 522
 523         u->debug_level = debug_level;
 524         u->gamelen = MC_GAMELEN;
 525         u->resign_threshold = 0.2;
 526         u->sure_win_threshold = 0.85;
 527         u->mercymin = 0;
 528         u->significant_threshold = 50;
 529         u->expand_p = 2;
 530         u->dumpthres = 1000;
 531         u->playout_amaf = true;
 532         u->playout_amaf_nakade = false;
 533         u->amaf_prior = false;
 534         u->max_tree_size = 3072ULL * 1048576;
 535         u->pruning_threshold = 0;
 536
 537         u->threads = 1;
 538         u->thread_model = TM_TREEVL;
 539         u->virtual_loss = 1;
 540
 541         u->fuseki_end = 20; // max time at 361*20% = 72 moves (our 36th move, still 99 to play)
 542         u->yose_start = 40; // (100-40-25)*361/100/2 = 63 moves still to play by us then
 543         u->bestr_ratio = 0.02;
 544         // 2.5 is clearly too much, but seems to compensate well for overly stern time allocations.
 545         // TODO: Further tuning and experiments with better time allocation schemes.
 546         u->best2_ratio = 2.5;
 547         u->max_maintime_ratio = 8.0;
 548
 549         u->val_scale = 0.04; u->val_points = 40;
 550         u->dynkomi_interval = 1000;
 551         u->dynkomi_mask = S_BLACK | S_WHITE;
 552
 553         u->tenuki_d = 4;
 554         u->local_tree_aging = 80;
 555         u->local_tree_allseq = 1;
 556         u->local_tree_rootseqval = 1;
 557         u->local_tree_depth_decay = 1.5;
 558
 559         u->plugins = pluginset_init(b);
 560
 561         u->jdict = joseki_load(b->size);
 562
 563         if (arg) {
 564                 char *optspec, *next = arg;
 565                 while (*next) {
 566                         optspec = next;
 567                         next += strcspn(next, ",");
 568                         if (*next) { *next++ = 0; } else { *next = 0; }
 569
 570                         char *optname = optspec;
 571                         char *optval = strchr(optspec, '=');
 572                         if (optval) *optval++ = 0;
 573
 574                         /** Basic options */
 575
 576                         if (!strcasecmp(optname, "debug")) {
 577                                 if (optval)
 578                                         u->debug_level = atoi(optval);
 579                                 else
 580                                         u->debug_level++;
 581                         } else if (!strcasecmp(optname, "dumpthres") && optval) {
 582                                 /* When dumping the UCT tree on output, include
 583                                  * nodes with at least this many playouts.
 584                                  * (This value is re-scaled "intelligently"
 585                                  * in case of very large trees.) */
 586                                 u->dumpthres = atoi(optval);
 587                         } else if (!strcasecmp(optname, "resign_threshold") && optval) {
 588                                 /* Resign when this ratio of games is lost
 589                                  * after GJ_MINGAMES sample is taken. */
 590                                 u->resign_threshold = atof(optval);
 591                         } else if (!strcasecmp(optname, "sure_win_threshold") && optval) {
 592                                 /* Stop reading when this ratio of games is won
 593                                  * after PLAYOUT_EARLY_BREAK_MIN sample is
 594                                  * taken. (Prevents stupid time losses,
 595                                  * friendly to human opponents.) */
 596                                 u->sure_win_threshold = atof(optval);
 597                         } else if (!strcasecmp(optname, "force_seed") && optval) {
 598                                 /* Set RNG seed at the tree setup. */
 599                                 u->force_seed = atoi(optval);
 600                         } else if (!strcasecmp(optname, "no_tbook")) {
 601                                 /* Disable UCT opening tbook. */
 602                                 u->no_tbook = true;
 603                         } else if (!strcasecmp(optname, "pass_all_alive")) {
 604                                 /* Whether to consider passing only after all
 605                                  * dead groups were removed from the board;
 606                                  * this is like all genmoves are in fact
 607                                  * kgs-genmove_cleanup. */
 608                                 u->pass_all_alive = !optval || atoi(optval);
 609                         } else if (!strcasecmp(optname, "territory_scoring")) {
 610                                 /* Use territory scoring (default is area scoring).
 611                                  * An explicit kgs-rules command overrides this. */
 612                                 u->territory_scoring = !optval || atoi(optval);
 613                         } else if (!strcasecmp(optname, "banner") && optval) {
 614                                 /* Additional banner string. This must come as the
 615                                  * last engine parameter. */
 616                                 if (*next) *--next = ',';
 617                                 u->banner = strdup(optval);
 618                                 break;
 619                         } else if (!strcasecmp(optname, "plugin") && optval) {
 620                                 /* Load an external plugin; filename goes before the colon,
 621                                  * extra arguments after the colon. */
 622                                 char *pluginarg = strchr(optval, ':');
 623                                 if (pluginarg)
 624                                         *pluginarg++ = 0;
 625                                 plugin_load(u->plugins, optval, pluginarg);
 626
 627                         /** UCT behavior and policies */
 628
 629                         } else if ((!strcasecmp(optname, "policy")
 630                                 /* Node selection policy. ucb1amaf is the
 631                                  * default policy implementing RAVE, while
 632                                  * ucb1 is the simple exploration/exploitation
 633                                  * policy. Policies can take further extra
 634                                  * options. */
 635                                     || !strcasecmp(optname, "random_policy")) && optval) {
 636                                 /* A policy to be used randomly with small
 637                                  * chance instead of the default policy. */
 638                                 char *policyarg = strchr(optval, ':');
 639                                 struct uct_policy **p = !strcasecmp(optname, "policy") ? &u->policy : &u->random_policy;
 640                                 if (policyarg)
 641                                         *policyarg++ = 0;
 642                                 if (!strcasecmp(optval, "ucb1")) {
 643                                         *p = policy_ucb1_init(u, policyarg);
 644                                 } else if (!strcasecmp(optval, "ucb1amaf")) {
 645                                         *p = policy_ucb1amaf_init(u, policyarg);
 646                                 } else {
 647                                         fprintf(stderr, "UCT: Invalid tree policy %s\n", optval);
 648                                         exit(1);
 649                                 }
 650                         } else if (!strcasecmp(optname, "playout") && optval) {
 651                                 /* Random simulation (playout) policy.
 652                                  * moggy is the default policy with large
 653                                  * amount of domain-specific knowledge and
 654                                  * heuristics. light is a simple uniformly
 655                                  * random move selection policy. */
 656                                 char *playoutarg = strchr(optval, ':');
 657                                 if (playoutarg)
 658                                         *playoutarg++ = 0;
 659                                 if (!strcasecmp(optval, "moggy")) {
 660                                         u->playout = playout_moggy_init(playoutarg, b, u->jdict);
 661                                 } else if (!strcasecmp(optval, "light")) {
 662                                         u->playout = playout_light_init(playoutarg, b);
 663                                 } else {
 664                                         fprintf(stderr, "UCT: Invalid playout policy %s\n", optval);
 665                                         exit(1);
 666                                 }
 667                         } else if (!strcasecmp(optname, "prior") && optval) {
 668                                 /* Node priors policy. When expanding a node,
 669                                  * it will seed node values heuristically
 670                                  * (most importantly, based on playout policy
 671                                  * opinion, but also with regard to other
 672                                  * things). See uct/prior.c for details.
 673                                  * Use prior=eqex=0 to disable priors. */
 674                                 u->prior = uct_prior_init(optval, b);
 675                         } else if (!strcasecmp(optname, "mercy") && optval) {
 676                                 /* Minimal difference of black/white captures
 677                                  * to stop playout - "Mercy Rule". Speeds up
 678                                  * hopeless playouts at the expense of some
 679                                  * accuracy. */
 680                                 u->mercymin = atoi(optval);
 681                         } else if (!strcasecmp(optname, "gamelen") && optval) {
 682                                 /* Maximum length of single simulation
 683                                  * in moves. */
 684                                 u->gamelen = atoi(optval);
 685                         } else if (!strcasecmp(optname, "expand_p") && optval) {
 686                                 /* Expand UCT nodes after it has been
 687                                  * visited this many times. */
 688                                 u->expand_p = atoi(optval);
 689                         } else if (!strcasecmp(optname, "random_policy_chance") && optval) {
 690                                 /* If specified (N), with probability 1/N, random_policy policy
 691                                  * descend is used instead of main policy descend; useful
 692                                  * if specified policy (e.g. UCB1AMAF) can make unduly biased
 693                                  * choices sometimes, you can fall back to e.g.
 694                                  * random_policy=UCB1. */
 695                                 u->random_policy_chance = atoi(optval);
 696
 697                         /** General AMAF behavior */
 698                         /* (Only relevant if the policy supports AMAF.
 699                          * More variables can be tuned as policy
 700                          * parameters.) */
 701
 702                         } else if (!strcasecmp(optname, "playout_amaf")) {
 703                                 /* Whether to include random playout moves in
 704                                  * AMAF as well. (Otherwise, only tree moves
 705                                  * are included in AMAF. Of course makes sense
 706                                  * only in connection with an AMAF policy.) */
 707                                 /* with-without: 55.5% (+-4.1) */
 708                                 if (optval && *optval == '0')
 709                                         u->playout_amaf = false;
 710                                 else
 711                                         u->playout_amaf = true;
 712                         } else if (!strcasecmp(optname, "playout_amaf_nakade")) {
 713                                 /* Whether to include nakade moves from playouts
 714                                  * in the AMAF statistics; this tends to nullify
 715                                  * the playout_amaf effect by adding too much
 716                                  * noise. */
 717                                 if (optval && *optval == '0')
 718                                         u->playout_amaf_nakade = false;
 719                                 else
 720                                         u->playout_amaf_nakade = true;
 721                         } else if (!strcasecmp(optname, "playout_amaf_cutoff") && optval) {
 722                                 /* Keep only first N% of playout stage AMAF
 723                                  * information. */
 724                                 u->playout_amaf_cutoff = atoi(optval);
 725                         } else if (!strcasecmp(optname, "amaf_prior") && optval) {
 726                                 /* In node policy, consider prior values
 727                                  * part of the real result term or part
 728                                  * of the AMAF term? */
 729                                 u->amaf_prior = atoi(optval);
 730
 731                         /** Performance and memory management */
 732
 733                         } else if (!strcasecmp(optname, "threads") && optval) {
 734                                 /* By default, Pachi will run with only single
 735                                  * tree search thread! */
 736                                 u->threads = atoi(optval);
 737                         } else if (!strcasecmp(optname, "thread_model") && optval) {
 738                                 if (!strcasecmp(optval, "tree")) {
 739                                         /* Tree parallelization - all threads
 740                                          * grind on the same tree. */
 741                                         u->thread_model = TM_TREE;
 742                                         u->virtual_loss = 0;
 743                                 } else if (!strcasecmp(optval, "treevl")) {
 744                                         /* Tree parallelization, but also
 745                                          * with virtual losses - this discou-
 746                                          * rages most threads choosing the
 747                                          * same tree branches to read. */
 748                                         u->thread_model = TM_TREEVL;
 749                                 } else {
 750                                         fprintf(stderr, "UCT: Invalid thread model %s\n", optval);
 751                                         exit(1);
 752                                 }
 753                         } else if (!strcasecmp(optname, "virtual_loss")) {
 754                                 /* Number of virtual losses added before evaluating a node. */
 755                                 u->virtual_loss = !optval || atoi(optval);
 756                         } else if (!strcasecmp(optname, "pondering")) {
 757                                 /* Keep searching even during opponent's turn. */
 758                                 u->pondering_opt = !optval || atoi(optval);
 759                         } else if (!strcasecmp(optname, "max_tree_size") && optval) {
 760                                 /* Maximum amount of memory [MiB] consumed by the move tree.
 761                                  * For fast_alloc it includes the temp tree used for pruning.
 762                                  * Default is 3072 (3 GiB). */
 763                                 u->max_tree_size = atol(optval) * 1048576;
 764                         } else if (!strcasecmp(optname, "fast_alloc")) {
 765                                 u->fast_alloc = !optval || atoi(optval);
 766                         } else if (!strcasecmp(optname, "pruning_threshold") && optval) {
 767                                 /* Force pruning at beginning of a move if the tree consumes
 768                                  * more than this [MiB]. Default is 10% of max_tree_size.
 769                                  * Increase to reduce pruning time overhead if memory is plentiful.
 770                                  * This option is meaningful only for fast_alloc. */
 771                                 u->pruning_threshold = atol(optval) * 1048576;
 772
 773                         /** Time control */
 774
 775                         } else if (!strcasecmp(optname, "best2_ratio") && optval) {
 776                                 /* If set, prolong simulating while
 777                                  * first_best/second_best playouts ratio
 778                                  * is less than best2_ratio. */
 779                                 u->best2_ratio = atof(optval);
 780                         } else if (!strcasecmp(optname, "bestr_ratio") && optval) {
 781                                 /* If set, prolong simulating while
 782                                  * best,best_best_child values delta
 783                                  * is more than bestr_ratio. */
 784                                 u->bestr_ratio = atof(optval);
 785                         } else if (!strcasecmp(optname, "max_maintime_ratio") && optval) {
 786                                 /* If set and while not in byoyomi, prolong simulating no more than
 787                                  * max_maintime_ratio times the normal desired thinking time. */
 788                                 u->max_maintime_ratio = atof(optval);
 789                         } else if (!strcasecmp(optname, "fuseki_end") && optval) {
 790                                 /* At the very beginning it's not worth thinking
 791                                  * too long because the playout evaluations are
 792                                  * very noisy. So gradually increase the thinking
 793                                  * time up to maximum when fuseki_end percent
 794                                  * of the board has been played.
 795                                  * This only applies if we are not in byoyomi. */
 796                                 u->fuseki_end = atoi(optval);
 797                         } else if (!strcasecmp(optname, "yose_start") && optval) {
 798                                 /* When yose_start percent of the board has been
 799                                  * played, or if we are in byoyomi, stop spending
 800                                  * more time and spread the remaining time
 801                                  * uniformly.
 802                                  * Between fuseki_end and yose_start, we spend
 803                                  * a constant proportion of the remaining time
 804                                  * on each move. (yose_start should actually
 805                                  * be much earlier than when real yose start,
 806                                  * but "yose" is a good short name to convey
 807                                  * the idea.) */
 808                                 u->yose_start = atoi(optval);
 809
 810                         /** Dynamic komi */
 811
 812                         } else if (!strcasecmp(optname, "dynkomi") && optval) {
 813                                 /* Dynamic komi approach; there are multiple
 814                                  * ways to adjust komi dynamically throughout
 815                                  * play. We currently support two: */
 816                                 char *dynkomiarg = strchr(optval, ':');
 817                                 if (dynkomiarg)
 818                                         *dynkomiarg++ = 0;
 819                                 if (!strcasecmp(optval, "none")) {
 820                                         u->dynkomi = uct_dynkomi_init_none(u, dynkomiarg, b);
 821                                 } else if (!strcasecmp(optval, "linear")) {
 822                                         /* You should set dynkomi_mask=1
 823                                          * since this doesn't work well
 824                                          * for white handicaps! */
 825                                         u->dynkomi = uct_dynkomi_init_linear(u, dynkomiarg, b);
 826                                 } else if (!strcasecmp(optval, "adaptive")) {
 827                                         /* There are many more knobs to
 828                                          * crank - see uct/dynkomi.c. */
 829                                         u->dynkomi = uct_dynkomi_init_adaptive(u, dynkomiarg, b);
 830                                 } else {
 831                                         fprintf(stderr, "UCT: Invalid dynkomi mode %s\n", optval);
 832                                         exit(1);
 833                                 }
 834                         } else if (!strcasecmp(optname, "dynkomi_mask") && optval) {
 835                                 /* Bitmask of colors the player must be
 836                                  * for dynkomi be applied; you may want
 837                                  * to use dynkomi_mask=3 to allow dynkomi
 838                                  * even in games where Pachi is white. */
 839                                 u->dynkomi_mask = atoi(optval);
 840                         } else if (!strcasecmp(optname, "dynkomi_interval") && optval) {
 841                                 /* If non-zero, re-adjust dynamic komi
 842                                  * throughout a single genmove reading,
 843                                  * roughly every N simulations. */
 844                                 /* XXX: Does not work with tree
 845                                  * parallelization. */
 846                                 u->dynkomi_interval = atoi(optval);
 847
 848                         /** Node value result scaling */
 849
 850                         } else if (!strcasecmp(optname, "val_scale") && optval) {
 851                                 /* How much of the game result value should be
 852                                  * influenced by win size. Zero means it isn't. */
 853                                 u->val_scale = atof(optval);
 854                         } else if (!strcasecmp(optname, "val_points") && optval) {
 855                                 /* Maximum size of win to be scaled into game
 856                                  * result value. Zero means boardsize^2. */
 857                                 u->val_points = atoi(optval) * 2; // result values are doubled
 858                         } else if (!strcasecmp(optname, "val_extra")) {
 859                                 /* If false, the score coefficient will be simply
 860                                  * added to the value, instead of scaling the result
 861                                  * coefficient because of it. */
 862                                 u->val_extra = !optval || atoi(optval);
 863
 864                         /** Local trees */
 865                         /* (Purely experimental. Does not work - yet!) */
 866
 867                         } else if (!strcasecmp(optname, "local_tree") && optval) {
 868                                 /* Whether to bias exploration by local tree values
 869                                  * (must be supported by the used policy).
 870                                  * 0: Don't.
 871                                  * 1: Do, value = result.
 872                                  * Try to temper the result:
 873                                  * 2: Do, value = 0.5+(result-expected)/2.
 874                                  * 3: Do, value = 0.5+bzz((result-expected)^2).
 875                                  * 4: Do, value = 0.5+sqrt(result-expected)/2. */
 876                                 u->local_tree = atoi(optval);
 877                         } else if (!strcasecmp(optname, "tenuki_d") && optval) {
 878                                 /* Tenuki distance at which to break the local tree. */
 879                                 u->tenuki_d = atoi(optval);
 880                                 if (u->tenuki_d > TREE_NODE_D_MAX + 1) {
 881                                         fprintf(stderr, "uct: tenuki_d must not be larger than TREE_NODE_D_MAX+1 %d\n", TREE_NODE_D_MAX + 1);
 882                                         exit(1);
 883                                 }
 884                         } else if (!strcasecmp(optname, "local_tree_aging") && optval) {
 885                                 /* How much to reduce local tree values between moves. */
 886                                 u->local_tree_aging = atof(optval);
 887                         } else if (!strcasecmp(optname, "local_tree_depth_decay") && optval) {
 888                                 /* With value x>0, during the descent the node
 889                                  * contributes 1/x^depth playouts in
 890                                  * the local tree. I.e., with x>1, nodes more
 891                                  * distant from local situation contribute more
 892                                  * than nodes near the root. */
 893                                 u->local_tree_depth_decay = atof(optval);
 894                         } else if (!strcasecmp(optname, "local_tree_allseq")) {
 895                                 /* If disabled, only complete sequences are stored
 896                                  * in the local tree. If this is on, also
 897                                  * subsequences starting at each move are stored. */
 898                                 u->local_tree_allseq = !optval || atoi(optval);
 899                         } else if (!strcasecmp(optname, "local_tree_pseqroot")) {
 900                                 /* By default, when we have no sequence move
 901                                  * to suggest in-playout, we give up. If this
 902                                  * is on, we make probability distribution from
 903                                  * sequences first moves instead. */
 904                                 u->local_tree_pseqroot = !optval || atoi(optval);
 905                         } else if (!strcasecmp(optname, "local_tree_rootseqval")) {
 906                                 /* If disabled, expected node value is computed by
 907                                  * summing up values through the whole descent.
 908                                  * If enabled, expected node value for
 909                                  * each sequence is the value at the root of the
 910                                  * sequence. */
 911                                 u->local_tree_rootseqval = !optval || atoi(optval);
 912
 913                         /** Other heuristics */
 914                         } else if (!strcasecmp(optname, "significant_threshold") && optval) {
 915                                 /* Some heuristics (treepool) rely
 916                                  * on the knowledge of the last "significant"
 917                                  * node in the descent. Such a node is
 918                                  * considered reasonably trustworthy to carry
 919                                  * some meaningful information in the values
 920                                  * of the node and its children. */
 921                                 u->significant_threshold = atoi(optval);
 922                         } else if (!strcasecmp(optname, "treepool_chance") && optval) {
 923                                 /* Chance of applying the treepool heuristic:
 924                                  * one of the best N children of the last
 925                                  * significant node is tried on each turn
 926                                  * of the simulation. */
 927                                 /* This is in form of two numbers:
 928                                  * PREMOVE:POSTMOVE. Each is percentage
 929                                  * value, one is the chance of the move
 930                                  * tried before playout policy, one is the
 931                                  * chance of it being applied if the policy
 932                                  * has not picked anymove. */
 933                                 char *optval2 = strchr(optval, ':');
 934                                 if (!optval2) {
 935                                         fprintf(stderr, "uct: treepool_chance takes two comma-separated numbers\n");
 936                                         exit(1);
 937                                 }
 938                                 u->treepool_chance[0] = atoi(optval);
 939                                 optval = ++optval2;
 940                                 u->treepool_chance[1] = atoi(optval);
 941                         } else if (!strcasecmp(optname, "treepool_size") && optval) {
 942                                 /* Number of top significant children
 943                                  * to pick from. Too low means low effect,
 944                                  * too high means even lousy moves are
 945                                  * played. */
 946                                 u->treepool_size = atoi(optval);
 947                         } else if (!strcasecmp(optname, "treepool_type") && optval) {
 948                                 /* How to sort the children. */
 949                                 if (!strcasecmp(optval, "rave_playouts"))
 950                                         u->treepool_type = UTT_RAVE_PLAYOUTS;
 951                                 else if (!strcasecmp(optval, "rave_value"))
 952                                         u->treepool_type = UTT_RAVE_VALUE;
 953                                 else if (!strcasecmp(optval, "uct_playouts"))
 954                                         u->treepool_type = UTT_UCT_PLAYOUTS;
 955                                 else if (!strcasecmp(optval, "uct_value"))
 956                                         u->treepool_type = UTT_UCT_VALUE;
 957                                 else if (!strcasecmp(optval, "evaluate"))
 958                                         /* The proper combination of RAVE,
 959                                          * UCT and prior, as used through
 960                                          * descent. */
 961                                         u->treepool_type = UTT_EVALUATE;
 962                                 else {
 963                                         fprintf(stderr, "uct: unknown treepool_type %s\n", optval);
 964                                         exit(1);
 965                                 }
 966                         } else if (!strcasecmp(optname, "treepool_pickfactor") && optval) {
 967                                 /* Pick factor influencing children choice.
 968                                  * By default (if this is 0 or 10), coords
 969                                  * have uniform probability to be chosen;
 970                                  * otherwise, children are tried from best
 971                                  * to worst, each picked with probability
 972                                  * (1/n * pickfactor/10).
 973                                  * I.e., better children may be preferred
 974                                  * if pickfactor > 10. */
 975                                 u->treepool_pickfactor = atoi(optval);
 976
 977                         /** Distributed engine slaves setup */
 978
 979                         } else if (!strcasecmp(optname, "slave")) {
 980                                 /* Act as slave for the distributed engine. */
 981                                 u->slave = !optval || atoi(optval);
 982                         } else if (!strcasecmp(optname, "shared_nodes") && optval) {
 983                                 /* Share at most shared_nodes between master and slave at each genmoves.
 984                                  * Must use the same value in master and slaves. */
 985                                 u->shared_nodes = atoi(optval);
 986                         } else if (!strcasecmp(optname, "shared_levels") && optval) {
 987                                 /* Share only nodes of level <= shared_levels. */
 988                                 u->shared_levels = atoi(optval);
 989                         } else if (!strcasecmp(optname, "stats_hbits") && optval) {
 990                                 /* Set hash table size to 2^stats_hbits for the shared stats. */
 991                                 u->stats_hbits = atoi(optval);
 992
 993                         } else {
 994                                 fprintf(stderr, "uct: Invalid engine argument %s or missing value\n", optname);
 995                                 exit(1);
 996                         }
 997                 }
 998         }
 999
1000         if (!u->policy)
1001                 u->policy = policy_ucb1amaf_init(u, NULL);
1002
1003         if (!!u->random_policy_chance ^ !!u->random_policy) {
1004                 fprintf(stderr, "uct: Only one of random_policy and random_policy_chance is set\n");
1005                 exit(1);
1006         }
1007
1008         if (!u->local_tree) {
1009                 /* No ltree aging. */
1010                 u->local_tree_aging = 1.0f;
1011         }
1012
1013         if (u->fast_alloc) {
1014                 if (u->pruning_threshold < u->max_tree_size / 10)
1015                         u->pruning_threshold = u->max_tree_size / 10;
1016                 if (u->pruning_threshold > u->max_tree_size / 2)
1017                         u->pruning_threshold = u->max_tree_size / 2;
1018
1019                 /* Limit pruning temp space to 20% of memory. Beyond this we discard
1020                  * the nodes and recompute them at the next move if necessary. */
1021                 u->max_pruned_size = u->max_tree_size / 5;
1022                 u->max_tree_size -= u->max_pruned_size;
1023         } else {
1024                 /* Reserve 5% memory in case the background free() are slower
1025                  * than the concurrent allocations. */
1026                 u->max_tree_size -= u->max_tree_size / 20;
1027         }
1028
1029         if (!u->prior)
1030                 u->prior = uct_prior_init(NULL, b);
1031
1032         if (!u->playout)
1033                 u->playout = playout_moggy_init(NULL, b, u->jdict);
1034         if (!u->playout->debug_level)
1035                 u->playout->debug_level = u->debug_level;
1036
1037         u->ownermap.map = malloc2(board_size2(b) * sizeof(u->ownermap.map[0]));
1038
1039         if (u->slave) {
1040                 if (!u->stats_hbits) u->stats_hbits = DEFAULT_STATS_HBITS;
1041                 if (!u->shared_nodes) u->shared_nodes = DEFAULT_SHARED_NODES;
1042                 assert(u->shared_levels * board_bits2(b) <= 8 * (int)sizeof(path_t));
1043         }
1044
1045         if (!u->dynkomi)
1046                 u->dynkomi = uct_dynkomi_init_adaptive(u, NULL, b);
1047
1048         /* Some things remain uninitialized for now - the opening tbook
1049          * is not loaded and the tree not set up. */
1050         /* This will be initialized in setup_state() at the first move
1051          * received/requested. This is because right now we are not aware
1052          * about any komi or handicap setup and such. */
1053
1054         return u;
1055 }
1056
1057 struct engine *
1058 engine_uct_init(char *arg, struct board *b)
1059 {
1060         struct uct *u = uct_state_init(arg, b);
1061         struct engine *e = calloc2(1, sizeof(struct engine));
1062         e->name = "UCT Engine";
1063         e->printhook = uct_printhook_ownermap;
1064         e->notify_play = uct_notify_play;
1065         e->chat = uct_chat;
1066         e->result = uct_result;
1067         e->genmove = uct_genmove;
1068         e->genmoves = uct_genmoves;
1069         e->dead_group_list = uct_dead_group_list;
1070         e->done = uct_done;
1071         e->data = u;
1072         if (u->slave)
1073                 e->notify = uct_notify;
1074
1075         const char banner[] = "I'm playing UCT. When I'm losing, I will resign, "
1076                 "if I think I win, I play until you pass. "
1077                 "Anyone can send me 'winrate' in private chat to get my assessment of the position.";
1078         if (!u->banner) u->banner = "";
1079         e->comment = malloc2(sizeof(banner) + strlen(u->banner) + 1);
1080         sprintf(e->comment, "%s %s", banner, u->banner);
1081
1082         return e;
1083 }