uct/uct.c

   1 #include <assert.h>
   2 #include <pthread.h>
   3 #include <signal.h>
   4 #include <stdio.h>
   5 #include <stdlib.h>
   6 #include <string.h>
   7 #include <time.h>
   8
   9 #define DEBUG
  10
  11 #include "debug.h"
  12 #include "board.h"
  13 #include "gtp.h"
  14 #include "move.h"
  15 #include "mq.h"
  16 #include "playout.h"
  17 #include "playout/elo.h"
  18 #include "playout/moggy.h"
  19 #include "playout/light.h"
  20 #include "random.h"
  21 #include "timeinfo.h"
  22 #include "tactics.h"
  23 #include "uct/internal.h"
  24 #include "uct/prior.h"
  25 #include "uct/tree.h"
  26 #include "uct/uct.h"
  27 #include "uct/walk.h"
  28
  29 struct uct_policy *policy_ucb1_init(struct uct *u, char *arg);
  30 struct uct_policy *policy_ucb1amaf_init(struct uct *u, char *arg);
  31 static void uct_pondering_stop(struct uct *u);
  32
  33
  34 /* Default number of simulations to perform per move.
  35  * Note that this is now in total over all threads! (Unless TM_ROOT.) */
  36 #define MC_GAMES        80000
  37 #define MC_GAMELEN      MAX_GAMELEN
  38 static const struct time_info default_ti = {
  39         .period = TT_MOVE,
  40         .dim = TD_GAMES,
  41         .len = { .games = MC_GAMES },
  42 };
  43
  44 /* How big proportion of ownermap counts must be of one color to consider
  45  * the point sure. */
  46 #define GJ_THRES        0.8
  47 /* How many games to consider at minimum before judging groups. */
  48 #define GJ_MINGAMES     500
  49
  50 /* How often to inspect the tree from the main thread to check for playout
  51  * stop, progress reports, etc. (in seconds) */
  52 #define TREE_BUSYWAIT_INTERVAL 0.1 /* 100ms */
  53
  54 /* Once per how many simulations (per thread) to show a progress report line. */
  55 #define TREE_SIMPROGRESS_INTERVAL 10000
  56
  57 /* When terminating uct_search() early, the safety margin to add to the
  58  * remaining playout number estimate when deciding whether the result can
  59  * still change. */
  60 #define PLAYOUT_DELTA_SAFEMARGIN 1000
  61
  62
  63 static void
  64 setup_state(struct uct *u, struct board *b, enum stone color)
  65 {
  66         u->t = tree_init(b, color, u->fast_alloc ? u->max_tree_size: 0);
  67         if (u->force_seed)
  68                 fast_srandom(u->force_seed);
  69         if (UDEBUGL(0))
  70                 fprintf(stderr, "Fresh board with random seed %lu\n", fast_getseed());
  71         //board_print(b, stderr);
  72         if (!u->no_book && b->moves == 0) {
  73                 assert(color == S_BLACK);
  74                 tree_load(u->t, b);
  75         }
  76 }
  77
  78 static void
  79 reset_state(struct uct *u)
  80 {
  81         assert(u->t);
  82         tree_done(u->t); u->t = NULL;
  83 }
  84
  85 static void
  86 setup_dynkomi(struct uct *u, struct board *b, enum stone to_play)
  87 {
  88         if (u->dynkomi > b->moves && u->t->use_extra_komi)
  89                 u->t->extra_komi = uct_get_extra_komi(u, b);
  90         else
  91                 u->t->extra_komi = 0;
  92 }
  93
  94 static void
  95 prepare_move(struct engine *e, struct board *b, enum stone color)
  96 {
  97         struct uct *u = e->data;
  98
  99         if (u->t) {
 100                 /* Verify that we have sane state. */
 101                 assert(b->es == u);
 102                 assert(u->t && b->moves);
 103                 if (color != stone_other(u->t->root_color)) {
 104                         fprintf(stderr, "Fatal: Non-alternating play detected %d %d\n",
 105                                 color, u->t->root_color);
 106                         exit(1);
 107                 }
 108
 109         } else {
 110                 /* We need fresh state. */
 111                 b->es = u;
 112                 setup_state(u, b, color);
 113         }
 114
 115         u->ownermap.playouts = 0;
 116         memset(u->ownermap.map, 0, board_size2(b) * sizeof(u->ownermap.map[0]));
 117 }
 118
 119 static void
 120 dead_group_list(struct uct *u, struct board *b, struct move_queue *mq)
 121 {
 122         struct group_judgement gj;
 123         gj.thres = GJ_THRES;
 124         gj.gs = alloca(board_size2(b) * sizeof(gj.gs[0]));
 125         board_ownermap_judge_group(b, &u->ownermap, &gj);
 126         groups_of_status(b, &gj, GS_DEAD, mq);
 127 }
 128
 129 bool
 130 uct_pass_is_safe(struct uct *u, struct board *b, enum stone color, bool pass_all_alive)
 131 {
 132         if (u->ownermap.playouts < GJ_MINGAMES)
 133                 return false;
 134
 135         struct move_queue mq = { .moves = 0 };
 136         if (!pass_all_alive)
 137                 dead_group_list(u, b, &mq);
 138         return pass_is_safe(b, color, &mq);
 139 }
 140
 141
 142 static void
 143 uct_printhook_ownermap(struct board *board, coord_t c, FILE *f)
 144 {
 145         struct uct *u = board->es;
 146         assert(u);
 147         const char chr[] = ":XO,"; // dame, black, white, unclear
 148         const char chm[] = ":xo,";
 149         char ch = chr[board_ownermap_judge_point(&u->ownermap, c, GJ_THRES)];
 150         if (ch == ',') { // less precise estimate then?
 151                 ch = chm[board_ownermap_judge_point(&u->ownermap, c, 0.67)];
 152         }
 153         fprintf(f, "%c ", ch);
 154 }
 155
 156 static char *
 157 uct_notify_play(struct engine *e, struct board *b, struct move *m)
 158 {
 159         struct uct *u = e->data;
 160         if (!u->t) {
 161                 /* No state, create one - this is probably game beginning
 162                  * and we need to load the opening book right now. */
 163                 prepare_move(e, b, m->color);
 164                 assert(u->t);
 165         }
 166
 167         /* Stop pondering. */
 168         /* XXX: If we are about to receive multiple 'play' commands,
 169          * e.g. in a rengo, we will not ponder during the rest of them. */
 170         uct_pondering_stop(u);
 171
 172         if (is_resign(m->coord)) {
 173                 /* Reset state. */
 174                 reset_state(u);
 175                 return NULL;
 176         }
 177
 178         /* Promote node of the appropriate move to the tree root. */
 179         assert(u->t->root);
 180         if (!tree_promote_at(u->t, b, m->coord)) {
 181                 if (UDEBUGL(0))
 182                         fprintf(stderr, "Warning: Cannot promote move node! Several play commands in row?\n");
 183                 reset_state(u);
 184                 return NULL;
 185         }
 186         /* Setting up dynkomi is not necessary here, probably, but we
 187          * better do it anyway for consistency reasons. */
 188         setup_dynkomi(u, b, stone_other(m->color));
 189         return NULL;
 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                 /* Mock up some state and seed the ownermap by few
 236                  * simulations. */
 237                 prepare_move(e, b, S_BLACK); assert(u->t);
 238                 for (int i = 0; i < GJ_MINGAMES; i++)
 239                         uct_playout(u, b, S_BLACK, u->t);
 240                 mock_state = true;
 241         }
 242
 243         dead_group_list(u, b, mq);
 244
 245         if (mock_state) {
 246                 /* Clean up the mock state in case we will receive
 247                  * a genmove; we could get a non-alternating-move
 248                  * error from prepare_move() in that case otherwise. */
 249                 reset_state(u);
 250         }
 251 }
 252
 253 static void
 254 playout_policy_done(struct playout_policy *p)
 255 {
 256         if (p->done) p->done(p);
 257         if (p->data) free(p->data);
 258         free(p);
 259 }
 260
 261 static void
 262 uct_done(struct engine *e)
 263 {
 264         /* This is called on engine reset, especially when clear_board
 265          * is received and new game should begin. */
 266         struct uct *u = e->data;
 267         uct_pondering_stop(u);
 268         if (u->t) reset_state(u);
 269         free(u->ownermap.map);
 270
 271         free(u->policy);
 272         free(u->random_policy);
 273         playout_policy_done(u->playout);
 274         uct_prior_done(u->prior);
 275 }
 276
 277
 278 /* Pachi threading structure (if uct_playouts_parallel() is used):
 279  *
 280  * main thread
 281  *   |         main(), GTP communication, ...
 282  *   |         starts and stops the search managed by thread_manager
 283  *   |
 284  * thread_manager
 285  *   |         spawns and collects worker threads
 286  *   |
 287  * worker0
 288  * worker1
 289  * ...
 290  * workerK
 291  *             uct_playouts() loop, doing descend-playout until uct_halt
 292  *
 293  * Another way to look at it is by functions (lines denote thread boundaries):
 294  *
 295  * | uct_genmove()
 296  * | uct_search()            (uct_search_start() .. uct_search_stop())
 297  * | -----------------------
 298  * | spawn_thread_manager()
 299  * | -----------------------
 300  * | spawn_worker()
 301  * V uct_playouts() */
 302
 303 /* Set in thread manager in case the workers should stop. */
 304 volatile sig_atomic_t uct_halt = 0;
 305 /* ID of the running worker thread. */
 306 __thread int thread_id = -1;
 307 /* ID of the thread manager. */
 308 static pthread_t thread_manager;
 309 static bool thread_manager_running;
 310
 311 static pthread_mutex_t finish_mutex = PTHREAD_MUTEX_INITIALIZER;
 312 static pthread_cond_t finish_cond = PTHREAD_COND_INITIALIZER;
 313 static volatile int finish_thread;
 314 static pthread_mutex_t finish_serializer = PTHREAD_MUTEX_INITIALIZER;
 315
 316 struct spawn_ctx {
 317         int tid;
 318         struct uct *u;
 319         struct board *b;
 320         enum stone color;
 321         struct tree *t;
 322         unsigned long seed;
 323         int games;
 324 };
 325
 326 static void *
 327 spawn_worker(void *ctx_)
 328 {
 329         struct spawn_ctx *ctx = ctx_;
 330         /* Setup */
 331         fast_srandom(ctx->seed);
 332         thread_id = ctx->tid;
 333         /* Run */
 334         ctx->games = uct_playouts(ctx->u, ctx->b, ctx->color, ctx->t);
 335         /* Finish */
 336         pthread_mutex_lock(&finish_serializer);
 337         pthread_mutex_lock(&finish_mutex);
 338         finish_thread = ctx->tid;
 339         pthread_cond_signal(&finish_cond);
 340         pthread_mutex_unlock(&finish_mutex);
 341         return ctx;
 342 }
 343
 344 /* Thread manager, controlling worker threads. It must be called with
 345  * finish_mutex lock held, but it will unlock it itself before exiting;
 346  * this is necessary to be completely deadlock-free. */
 347 /* The finish_cond can be signalled for it to stop; in that case,
 348  * the caller should set finish_thread = -1. */
 349 /* After it is started, it will update mctx->t to point at some tree
 350  * used for the actual search (matters only for TM_ROOT), on return
 351  * it will set mctx->games to the number of performed simulations. */
 352 static void *
 353 spawn_thread_manager(void *ctx_)
 354 {
 355         /* In thread_manager, we use only some of the ctx fields. */
 356         struct spawn_ctx *mctx = ctx_;
 357         struct uct *u = mctx->u;
 358         struct tree *t = mctx->t;
 359         bool shared_tree = u->parallel_tree;
 360         fast_srandom(mctx->seed);
 361
 362         int played_games = 0;
 363         pthread_t threads[u->threads];
 364         int joined = 0;
 365
 366         uct_halt = 0;
 367
 368         /* Spawn threads... */
 369         for (int ti = 0; ti < u->threads; ti++) {
 370                 struct spawn_ctx *ctx = malloc(sizeof(*ctx));
 371                 ctx->u = u; ctx->b = mctx->b; ctx->color = mctx->color;
 372                 mctx->t = ctx->t = shared_tree ? t : tree_copy(t);
 373                 ctx->tid = ti; ctx->seed = fast_random(65536) + ti;
 374                 pthread_create(&threads[ti], NULL, spawn_worker, ctx);
 375                 if (UDEBUGL(2))
 376                         fprintf(stderr, "Spawned worker %d\n", ti);
 377         }
 378
 379         /* ...and collect them back: */
 380         while (joined < u->threads) {
 381                 /* Wait for some thread to finish... */
 382                 pthread_cond_wait(&finish_cond, &finish_mutex);
 383                 if (finish_thread < 0) {
 384                         /* Stop-by-caller. Tell the workers to wrap up. */
 385                         uct_halt = 1;
 386                         continue;
 387                 }
 388                 /* ...and gather its remnants. */
 389                 struct spawn_ctx *ctx;
 390                 pthread_join(threads[finish_thread], (void **) &ctx);
 391                 played_games += ctx->games;
 392                 joined++;
 393                 if (!shared_tree) {
 394                         if (ctx->t == mctx->t) mctx->t = t;
 395                         tree_merge(t, ctx->t);
 396                         tree_done(ctx->t);
 397                 }
 398                 free(ctx);
 399                 if (UDEBUGL(2))
 400                         fprintf(stderr, "Joined worker %d\n", finish_thread);
 401                 pthread_mutex_unlock(&finish_serializer);
 402         }
 403
 404         pthread_mutex_unlock(&finish_mutex);
 405
 406         if (!shared_tree)
 407                 tree_normalize(mctx->t, u->threads);
 408
 409         mctx->games = played_games;
 410         return mctx;
 411 }
 412
 413 static struct spawn_ctx *
 414 uct_search_start(struct uct *u, struct board *b, enum stone color, struct tree *t)
 415 {
 416         assert(u->threads > 0);
 417         assert(!thread_manager_running);
 418
 419         struct spawn_ctx ctx = { .u = u, .b = b, .color = color, .t = t, .seed = fast_random(65536) };
 420         static struct spawn_ctx mctx; mctx = ctx;
 421         pthread_mutex_lock(&finish_mutex);
 422         pthread_create(&thread_manager, NULL, spawn_thread_manager, &mctx);
 423         thread_manager_running = true;
 424         return &mctx;
 425 }
 426
 427 static struct spawn_ctx *
 428 uct_search_stop(void)
 429 {
 430         assert(thread_manager_running);
 431
 432         /* Signal thread manager to stop the workers. */
 433         pthread_mutex_lock(&finish_mutex);
 434         finish_thread = -1;
 435         pthread_cond_signal(&finish_cond);
 436         pthread_mutex_unlock(&finish_mutex);
 437
 438         /* Collect the thread manager. */
 439         struct spawn_ctx *pctx;
 440         thread_manager_running = false;
 441         pthread_join(thread_manager, (void **) &pctx);
 442         return pctx;
 443 }
 444
 445
 446 /* Determine whether we should terminate the search early. */
 447 static bool
 448 uct_search_stop_early(struct uct *u, struct tree *t, struct board *b,
 449                 struct time_info *ti, struct time_stop *stop,
 450                 struct tree_node *best, struct tree_node *best2,
 451                 int base_playouts, int i)
 452 {
 453         /* Early break in won situation. */
 454         if ((best->u.playouts >= 2000 && tree_node_get_value(t, 1, best->u.value) >= u->loss_threshold)
 455             || (best->u.playouts >= 500 && tree_node_get_value(t, 1, best->u.value) >= 0.95))
 456                 return true;
 457
 458         /* Break early if we estimate the second-best move cannot
 459          * catch up in assigned time anymore. We use all our time
 460          * if we are in byoyomi with single stone remaining in our
 461          * period, however. */
 462         if (best2 && ti->dim == TD_WALLTIME
 463             && (ti->len.t.main_time > 0 || ti->len.t.byoyomi_stones > 1)) {
 464                 double elapsed = time_now() - ti->len.t.timer_start;
 465                 double remaining = stop->worst.time - elapsed;
 466                 double pps = ((double)i - base_playouts) / elapsed;
 467                 double estplayouts = remaining * pps + PLAYOUT_DELTA_SAFEMARGIN;
 468                 if (best->u.playouts > best2->u.playouts + estplayouts) {
 469                         if (UDEBUGL(2))
 470                                 fprintf(stderr, "Early stop, result cannot change: "
 471                                         "best %d, best2 %d, estimated %f simulations to go\n",
 472                                         best->u.playouts, best2->u.playouts, estplayouts);
 473                         return true;
 474                 }
 475         }
 476
 477         return false;
 478 }
 479
 480 /* Run time-limited MCTS search on foreground. */
 481 static int
 482 uct_search(struct uct *u, struct board *b, struct time_info *ti, enum stone color, struct tree *t)
 483 {
 484         int base_playouts = u->t->root->u.playouts;
 485         if (UDEBUGL(2) && base_playouts > 0)
 486                 fprintf(stderr, "<pre-simulated %d games skipped>\n", base_playouts);
 487
 488         /* Set up time conditions. */
 489         if (ti->period == TT_NULL) *ti = default_ti;
 490         struct time_stop stop;
 491         time_stop_conditions(ti, b, u->fuseki_end, u->yose_start, &stop);
 492
 493         /* Number of last game with progress print. */
 494         int last_print = t->root->u.playouts;
 495         /* Number of simulations to wait before next print. */
 496         int print_interval = TREE_SIMPROGRESS_INTERVAL * (u->thread_model == TM_ROOT ? 1 : u->threads);
 497         /* Printed notification about full memory? */
 498         bool print_fullmem = false;
 499
 500         struct spawn_ctx *ctx = uct_search_start(u, b, color, t);
 501
 502         /* The search tree is ctx->t. This is normally == t, but in case of
 503          * TM_ROOT, it is one of the trees belonging to the independent
 504          * workers. It is important to reference ctx->t directly since the
 505          * thread manager will swap the tree pointer asynchronously. */
 506         /* XXX: This means TM_ROOT support is suboptimal since single stalled
 507          * thread can stall the others in case of limiting the search by game
 508          * count. However, TM_ROOT just does not deserve any more extra code
 509          * right now. */
 510
 511         struct tree_node *best = NULL, *prev_best;
 512         struct tree_node *best2 = NULL; // Second-best move.
 513         struct tree_node *winner = NULL, *prev_winner;
 514
 515         double busywait_interval = TREE_BUSYWAIT_INTERVAL;
 516
 517         /* Now, just periodically poll the search tree. */
 518         while (1) {
 519                 time_sleep(busywait_interval);
 520                 /* busywait_interval should never be less than desired time, or the
 521                  * time control is broken. But if it happens to be less, we still search
 522                  * at least 100ms otherwise the move is completely random. */
 523
 524                 int i = ctx->t->root->u.playouts;
 525
 526                 /* Print progress? */
 527                 if (i - last_print > print_interval) {
 528                         last_print += print_interval; // keep the numbers tidy
 529                         uct_progress_status(u, ctx->t, color, last_print);
 530                 }
 531                 if (!print_fullmem && ctx->t->nodes_size > u->max_tree_size) {
 532                         if (UDEBUGL(2))
 533                                 fprintf(stderr, "memory limit hit (%ld > %lu)\n", ctx->t->nodes_size, u->max_tree_size);
 534                         print_fullmem = true;
 535                 }
 536
 537                 /* Check against time settings. */
 538                 bool desired_done = false;
 539                 if (ti->dim == TD_WALLTIME) {
 540                         double elapsed = time_now() - ti->len.t.timer_start;
 541                         if (elapsed > stop.worst.time) break;
 542                         desired_done = elapsed > stop.desired.time;
 543
 544                 } else { assert(ti->dim == TD_GAMES);
 545                         if (i > stop.worst.playouts) break;
 546                         desired_done = i > stop.desired.playouts;
 547                 }
 548
 549                 prev_best = best;
 550                 best = u->policy->choose(u->policy, ctx->t->root, b, color, resign);
 551                 best2 = u->policy->choose(u->policy, ctx->t->root, b, color, best->coord);
 552
 553                 if (best && uct_search_stop_early(u, ctx->t, b, ti, &stop, best, best2, base_playouts, i))
 554                         break;
 555
 556                 /* We want to stop simulating, but are willing to keep trying
 557                  * if we aren't completely sure about the winner yet. */
 558                 if (desired_done) {
 559                         if (!best) {
 560                                 if (UDEBUGL(2))
 561                                         fprintf(stderr, "Did not find best move, still trying...\n");
 562                                 continue;
 563                         }
 564
 565                         if (u->best2_ratio > 0) {
 566                                 /* Check best/best2 simulations ratio. If the
 567                                  * two best moves give very similar results,
 568                                  * keep simulating. */
 569                                 if (best2 && best2->u.playouts
 570                                     && (double)best->u.playouts / best2->u.playouts < u->best2_ratio) {
 571                                         if (UDEBUGL(2))
 572                                                 fprintf(stderr, "Best2 ratio %f < threshold %f\n",
 573                                                         (double)best->u.playouts / best2->u.playouts,
 574                                                         u->best2_ratio);
 575                                         continue;
 576                                 }
 577                         }
 578
 579                         if (u->policy->winner && u->policy->evaluate) {
 580                                 prev_winner = winner;
 581                                 winner = u->policy->winner(u->policy, ctx->t, ctx->t->root);
 582                                 if (winner && winner != best) {
 583                                         /* Keep simulating if best explored
 584                                          * does not have also highest value. */
 585                                         if (UDEBUGL(2) && (best != prev_best || winner != prev_winner)) {
 586                                                 fprintf(stderr, "[%d] best %3s [%d] %f != winner %3s [%d] %f\n", i,
 587                                                         coord2sstr(best->coord, ctx->t->board),
 588                                                         best->u.playouts, tree_node_get_value(ctx->t, 1, best->u.value),
 589                                                         coord2sstr(winner->coord, ctx->t->board),
 590                                                         winner->u.playouts, tree_node_get_value(ctx->t, 1, winner->u.value));
 591                                         }
 592                                         continue;
 593                                 }
 594                         }
 595
 596                         /* No reason to keep simulating, bye. */
 597                         break;
 598                 }
 599
 600                 /* TODO: Early break if best->variance goes under threshold and we already
 601                  * have enough playouts (possibly thanks to book or to pondering)? */
 602         }
 603
 604         ctx = uct_search_stop();
 605
 606         if (UDEBUGL(2))
 607                 tree_dump(t, u->dumpthres);
 608         if (UDEBUGL(0))
 609                 uct_progress_status(u, t, color, ctx->games);
 610
 611         return ctx->games;
 612 }
 613
 614
 615 /* Start pondering background with @color to play. */
 616 static void
 617 uct_pondering_start(struct uct *u, struct board *b0, struct tree *t, enum stone color)
 618 {
 619         if (UDEBUGL(1))
 620                 fprintf(stderr, "Starting to ponder with color %s\n", stone2str(stone_other(color)));
 621         u->pondering = true;
 622
 623         /* We need a local board copy to ponder upon. */
 624         struct board *b = malloc(sizeof(*b)); board_copy(b, b0);
 625
 626         /* *b0 did not have the genmove'd move played yet. */
 627         struct move m = { t->root->coord, t->root_color };
 628         int res = board_play(b, &m);
 629         assert(res >= 0);
 630         setup_dynkomi(u, b, stone_other(m.color));
 631
 632         /* Start MCTS manager thread "headless". */
 633         uct_search_start(u, b, color, t);
 634 }
 635
 636 /* uct_search_stop() frontend for the pondering (non-genmove) mode. */
 637 static void
 638 uct_pondering_stop(struct uct *u)
 639 {
 640         u->pondering = false;
 641         if (!thread_manager_running)
 642                 return;
 643
 644         /* Stop the thread manager. */
 645         struct spawn_ctx *ctx = uct_search_stop();
 646         if (UDEBUGL(1)) {
 647                 fprintf(stderr, "(pondering) ");
 648                 uct_progress_status(u, ctx->t, ctx->color, ctx->games);
 649         }
 650         free(ctx->b);
 651 }
 652
 653
 654 static coord_t *
 655 uct_genmove(struct engine *e, struct board *b, struct time_info *ti, enum stone color, bool pass_all_alive)
 656 {
 657         double start_time = time_now();
 658         struct uct *u = e->data;
 659
 660         if (b->superko_violation) {
 661                 fprintf(stderr, "!!! WARNING: SUPERKO VIOLATION OCCURED BEFORE THIS MOVE\n");
 662                 fprintf(stderr, "Maybe you play with situational instead of positional superko?\n");
 663                 fprintf(stderr, "I'm going to ignore the violation, but note that I may miss\n");
 664                 fprintf(stderr, "some moves valid under this ruleset because of this.\n");
 665                 b->superko_violation = false;
 666         }
 667
 668         /* Seed the tree. */
 669         uct_pondering_stop(u);
 670         prepare_move(e, b, color);
 671         assert(u->t);
 672
 673         /* How to decide whether to use dynkomi in this game? Since we use
 674          * pondering, it's not simple "who-to-play" matter. Decide based on
 675          * the last genmove issued. */
 676         u->t->use_extra_komi = !!(u->dynkomi_mask & color);
 677         setup_dynkomi(u, b, color);
 678
 679         /* Perform the Monte Carlo Tree Search! */
 680         int played_games = uct_search(u, b, ti, color, u->t);
 681
 682         /* Choose the best move from the tree. */
 683         struct tree_node *best = u->policy->choose(u->policy, u->t->root, b, color, resign);
 684         if (!best) {
 685                 reset_state(u);
 686                 return coord_copy(pass);
 687         }
 688         if (UDEBUGL(1))
 689                 fprintf(stderr, "*** WINNER is %s (%d,%d) with score %1.4f (%d/%d:%d games)\n",
 690                         coord2sstr(best->coord, b), coord_x(best->coord, b), coord_y(best->coord, b),
 691                         tree_node_get_value(u->t, 1, best->u.value),
 692                         best->u.playouts, u->t->root->u.playouts, played_games);
 693
 694         /* Do not resign if we're so short of time that evaluation of best move is completely
 695          * unreliable, we might be winning actually. In this case best is almost random but
 696          * still better than resign. */
 697         if (tree_node_get_value(u->t, 1, best->u.value) < u->resign_ratio && !is_pass(best->coord)
 698             && best->u.playouts > GJ_MINGAMES) {
 699                 reset_state(u);
 700                 return coord_copy(resign);
 701         }
 702
 703         /* If the opponent just passed and we win counting, always
 704          * pass as well. */
 705         if (b->moves > 1 && is_pass(b->last_move.coord)) {
 706                 /* Make sure enough playouts are simulated. */
 707                 while (u->ownermap.playouts < GJ_MINGAMES)
 708                         uct_playout(u, b, color, u->t);
 709                 if (uct_pass_is_safe(u, b, color, u->pass_all_alive || pass_all_alive)) {
 710                         if (UDEBUGL(0))
 711                                 fprintf(stderr, "<Will rather pass, looks safe enough.>\n");
 712                         best->coord = pass;
 713                 }
 714         }
 715
 716         tree_promote_node(u->t, &best);
 717         /* After a pass, pondering is harmful for two reasons:
 718          * (i) We might keep pondering even when the game is over.
 719          * Of course this is the case for opponent resign as well.
 720          * (ii) More importantly, the ownermap will get skewed since
 721          * the UCT will start cutting off any playouts. */
 722         if (u->pondering_opt && !is_pass(best->coord)) {
 723                 uct_pondering_start(u, b, u->t, stone_other(color));
 724         }
 725         if (UDEBUGL(2)) {
 726                 double time = time_now() - start_time + 0.000001; /* avoid divide by zero */
 727                 fprintf(stderr, "genmove in %0.2fs (%d games/s, %d games/s/thread)\n",
 728                         time, (int)(played_games/time), (int)(played_games/time/u->threads));
 729         }
 730         return coord_copy(best->coord);
 731 }
 732
 733
 734 bool
 735 uct_genbook(struct engine *e, struct board *b, struct time_info *ti, enum stone color)
 736 {
 737         struct uct *u = e->data;
 738         if (!u->t) prepare_move(e, b, color);
 739         assert(u->t);
 740
 741         if (ti->dim == TD_GAMES) {
 742                 /* Don't count in games that already went into the book. */
 743                 ti->len.games += u->t->root->u.playouts;
 744         }
 745         uct_search(u, b, ti, color, u->t);
 746
 747         assert(ti->dim == TD_GAMES);
 748         tree_save(u->t, b, ti->len.games / 100);
 749
 750         return true;
 751 }
 752
 753 void
 754 uct_dumpbook(struct engine *e, struct board *b, enum stone color)
 755 {
 756         struct uct *u = e->data;
 757         struct tree *t = tree_init(b, color, u->fast_alloc ? u->max_tree_size: 0);
 758         tree_load(t, b);
 759         tree_dump(t, 0);
 760         tree_done(t);
 761 }
 762
 763
 764 struct uct *
 765 uct_state_init(char *arg, struct board *b)
 766 {
 767         struct uct *u = calloc(1, sizeof(struct uct));
 768
 769         u->debug_level = 3;
 770         u->gamelen = MC_GAMELEN;
 771         u->mercymin = 0;
 772         u->expand_p = 2;
 773         u->dumpthres = 1000;
 774         u->playout_amaf = true;
 775         u->playout_amaf_nakade = false;
 776         u->amaf_prior = false;
 777         u->max_tree_size = 3072ULL * 1048576;
 778
 779         if (board_size(b) - 2 >= 19)
 780                 u->dynkomi = 200;
 781         u->dynkomi_mask = S_BLACK;
 782
 783         u->threads = 1;
 784         u->thread_model = TM_TREEVL;
 785         u->parallel_tree = true;
 786         u->virtual_loss = true;
 787         u->fuseki_end = 20; // max time at 361*20% = 72 moves (our 36th move, still 99 to play)
 788         u->yose_start = 40; // (100-40-25)*361/100/2 = 63 moves still to play by us then
 789
 790         u->val_scale = 0.04; u->val_points = 40;
 791
 792         if (arg) {
 793                 char *optspec, *next = arg;
 794                 while (*next) {
 795                         optspec = next;
 796                         next += strcspn(next, ",");
 797                         if (*next) { *next++ = 0; } else { *next = 0; }
 798
 799                         char *optname = optspec;
 800                         char *optval = strchr(optspec, '=');
 801                         if (optval) *optval++ = 0;
 802
 803                         if (!strcasecmp(optname, "debug")) {
 804                                 if (optval)
 805                                         u->debug_level = atoi(optval);
 806                                 else
 807                                         u->debug_level++;
 808                         } else if (!strcasecmp(optname, "mercy") && optval) {
 809                                 /* Minimal difference of black/white captures
 810                                  * to stop playout - "Mercy Rule". Speeds up
 811                                  * hopeless playouts at the expense of some
 812                                  * accuracy. */
 813                                 u->mercymin = atoi(optval);
 814                         } else if (!strcasecmp(optname, "gamelen") && optval) {
 815                                 u->gamelen = atoi(optval);
 816                         } else if (!strcasecmp(optname, "expand_p") && optval) {
 817                                 u->expand_p = atoi(optval);
 818                         } else if (!strcasecmp(optname, "dumpthres") && optval) {
 819                                 u->dumpthres = atoi(optval);
 820                         } else if (!strcasecmp(optname, "best2_ratio") && optval) {
 821                                 /* If set, prolong simulating while
 822                                  * first_best/second_best playouts ratio
 823                                  * is less than best2_ratio. */
 824                                 u->best2_ratio = atof(optval);
 825                         } else if (!strcasecmp(optname, "playout_amaf")) {
 826                                 /* Whether to include random playout moves in
 827                                  * AMAF as well. (Otherwise, only tree moves
 828                                  * are included in AMAF. Of course makes sense
 829                                  * only in connection with an AMAF policy.) */
 830                                 /* with-without: 55.5% (+-4.1) */
 831                                 if (optval && *optval == '0')
 832                                         u->playout_amaf = false;
 833                                 else
 834                                         u->playout_amaf = true;
 835                         } else if (!strcasecmp(optname, "playout_amaf_nakade")) {
 836                                 /* Whether to include nakade moves from playouts
 837                                  * in the AMAF statistics; this tends to nullify
 838                                  * the playout_amaf effect by adding too much
 839                                  * noise. */
 840                                 if (optval && *optval == '0')
 841                                         u->playout_amaf_nakade = false;
 842                                 else
 843                                         u->playout_amaf_nakade = true;
 844                         } else if (!strcasecmp(optname, "playout_amaf_cutoff") && optval) {
 845                                 /* Keep only first N% of playout stage AMAF
 846                                  * information. */
 847                                 u->playout_amaf_cutoff = atoi(optval);
 848                         } else if ((!strcasecmp(optname, "policy") || !strcasecmp(optname, "random_policy")) && optval) {
 849                                 char *policyarg = strchr(optval, ':');
 850                                 struct uct_policy **p = !strcasecmp(optname, "policy") ? &u->policy : &u->random_policy;
 851                                 if (policyarg)
 852                                         *policyarg++ = 0;
 853                                 if (!strcasecmp(optval, "ucb1")) {
 854                                         *p = policy_ucb1_init(u, policyarg);
 855                                 } else if (!strcasecmp(optval, "ucb1amaf")) {
 856                                         *p = policy_ucb1amaf_init(u, policyarg);
 857                                 } else {
 858                                         fprintf(stderr, "UCT: Invalid tree policy %s\n", optval);
 859                                         exit(1);
 860                                 }
 861                         } else if (!strcasecmp(optname, "playout") && optval) {
 862                                 char *playoutarg = strchr(optval, ':');
 863                                 if (playoutarg)
 864                                         *playoutarg++ = 0;
 865                                 if (!strcasecmp(optval, "moggy")) {
 866                                         u->playout = playout_moggy_init(playoutarg);
 867                                 } else if (!strcasecmp(optval, "light")) {
 868                                         u->playout = playout_light_init(playoutarg);
 869                                 } else if (!strcasecmp(optval, "elo")) {
 870                                         u->playout = playout_elo_init(playoutarg);
 871                                 } else {
 872                                         fprintf(stderr, "UCT: Invalid playout policy %s\n", optval);
 873                                         exit(1);
 874                                 }
 875                         } else if (!strcasecmp(optname, "prior") && optval) {
 876                                 u->prior = uct_prior_init(optval, b);
 877                         } else if (!strcasecmp(optname, "amaf_prior") && optval) {
 878                                 u->amaf_prior = atoi(optval);
 879                         } else if (!strcasecmp(optname, "threads") && optval) {
 880                                 /* By default, Pachi will run with only single
 881                                  * tree search thread! */
 882                                 u->threads = atoi(optval);
 883                         } else if (!strcasecmp(optname, "thread_model") && optval) {
 884                                 if (!strcasecmp(optval, "root")) {
 885                                         /* Root parallelization - each thread
 886                                          * does independent search, trees are
 887                                          * merged at the end. */
 888                                         u->thread_model = TM_ROOT;
 889                                         u->parallel_tree = false;
 890                                         u->virtual_loss = false;
 891                                 } else if (!strcasecmp(optval, "tree")) {
 892                                         /* Tree parallelization - all threads
 893                                          * grind on the same tree. */
 894                                         u->thread_model = TM_TREE;
 895                                         u->parallel_tree = true;
 896                                         u->virtual_loss = false;
 897                                 } else if (!strcasecmp(optval, "treevl")) {
 898                                         /* Tree parallelization, but also
 899                                          * with virtual losses - this discou-
 900                                          * rages most threads choosing the
 901                                          * same tree branches to read. */
 902                                         u->thread_model = TM_TREEVL;
 903                                         u->parallel_tree = true;
 904                                         u->virtual_loss = true;
 905                                 } else {
 906                                         fprintf(stderr, "UCT: Invalid thread model %s\n", optval);
 907                                         exit(1);
 908                                 }
 909                         } else if (!strcasecmp(optname, "pondering")) {
 910                                 /* Keep searching even during opponent's turn. */
 911                                 u->pondering_opt = !optval || atoi(optval);
 912                         } else if (!strcasecmp(optname, "fuseki_end") && optval) {
 913                                 /* At the very beginning it's not worth thinking
 914                                  * too long because the playout evaluations are
 915                                  * very noisy. So gradually increase the thinking
 916                                  * time up to maximum when fuseki_end percent
 917                                  * of the board has been played.
 918                                  * This only applies if we are not in byoyomi. */
 919                                 u->fuseki_end = atoi(optval);
 920                         } else if (!strcasecmp(optname, "yose_start") && optval) {
 921                                 /* When yose_start percent of the board has been
 922                                  * played, or if we are in byoyomi, stop spending
 923                                  * more time and spread the remaining time
 924                                  * uniformly.
 925                                  * Between fuseki_end and yose_start, we spend
 926                                  * a constant proportion of the remaining time
 927                                  * on each move. (yose_start should actually
 928                                  * be much earlier than when real yose start,
 929                                  * but "yose" is a good short name to convey
 930                                  * the idea.) */
 931                                 u->yose_start = atoi(optval);
 932                         } else if (!strcasecmp(optname, "force_seed") && optval) {
 933                                 u->force_seed = atoi(optval);
 934                         } else if (!strcasecmp(optname, "no_book")) {
 935                                 u->no_book = true;
 936                         } else if (!strcasecmp(optname, "dynkomi")) {
 937                                 /* Dynamic komi in handicap game; linearly
 938                                  * decreases to basic settings until move
 939                                  * #optval. */
 940                                 u->dynkomi = optval ? atoi(optval) : 150;
 941                         } else if (!strcasecmp(optname, "dynkomi_mask") && optval) {
 942                                 /* Bitmask of colors the player must be
 943                                  * for dynkomi be applied; you may want
 944                                  * to use dynkomi_mask=3 to allow dynkomi
 945                                  * even in games where Pachi is white. */
 946                                 u->dynkomi_mask = atoi(optval);
 947                         } else if (!strcasecmp(optname, "val_scale") && optval) {
 948                                 /* How much of the game result value should be
 949                                  * influenced by win size. Zero means it isn't. */
 950                                 u->val_scale = atof(optval);
 951                         } else if (!strcasecmp(optname, "val_points") && optval) {
 952                                 /* Maximum size of win to be scaled into game
 953                                  * result value. Zero means boardsize^2. */
 954                                 u->val_points = atoi(optval) * 2; // result values are doubled
 955                         } else if (!strcasecmp(optname, "val_extra")) {
 956                                 /* If false, the score coefficient will be simply
 957                                  * added to the value, instead of scaling the result
 958                                  * coefficient because of it. */
 959                                 u->val_extra = !optval || atoi(optval);
 960                         } else if (!strcasecmp(optname, "root_heuristic") && optval) {
 961                                 /* Whether to bias exploration by root node values
 962                                  * (must be supported by the used policy).
 963                                  * 0: Don't.
 964                                  * 1: Do, value = result.
 965                                  * Try to temper the result:
 966                                  * 2: Do, value = 0.5+(result-expected)/2.
 967                                  * 3: Do, value = 0.5+bzz((result-expected)^2). */
 968                                 u->root_heuristic = atoi(optval);
 969                         } else if (!strcasecmp(optname, "pass_all_alive")) {
 970                                 /* Whether to consider all stones alive at the game
 971                                  * end instead of marking dead groupd. */
 972                                 u->pass_all_alive = !optval || atoi(optval);
 973                         } else if (!strcasecmp(optname, "random_policy_chance") && optval) {
 974                                 /* If specified (N), with probability 1/N, random_policy policy
 975                                  * descend is used instead of main policy descend; useful
 976                                  * if specified policy (e.g. UCB1AMAF) can make unduly biased
 977                                  * choices sometimes, you can fall back to e.g.
 978                                  * random_policy=UCB1. */
 979                                 u->random_policy_chance = atoi(optval);
 980                         } else if (!strcasecmp(optname, "max_tree_size") && optval) {
 981                                 /* Maximum amount of memory [MiB] consumed by the move tree.
 982                                  * Default is 3072 (3 GiB). Note that if you use TM_ROOT,
 983                                  * this limits size of only one of the trees, not all of them
 984                                  * together. */
 985                                 u->max_tree_size = atol(optval) * 1048576;
 986                         } else if (!strcasecmp(optname, "fast_alloc")) {
 987                                 u->fast_alloc = !optval || atoi(optval);
 988                         } else if (!strcasecmp(optname, "banner") && optval) {
 989                                 /* Additional banner string. This must come as the
 990                                  * last engine parameter. */
 991                                 if (*next) *--next = ',';
 992                                 u->banner = strdup(optval);
 993                                 break;
 994                         } else {
 995                                 fprintf(stderr, "uct: Invalid engine argument %s or missing value\n", optname);
 996                                 exit(1);
 997                         }
 998                 }
 999         }
1000
1001         u->resign_ratio = 0.2; /* Resign when most games are lost. */
1002         u->loss_threshold = 0.85; /* Stop reading if after at least 5000 playouts this is best value. */
1003         if (!u->policy)
1004                 u->policy = policy_ucb1amaf_init(u, NULL);
1005
1006         if (!!u->random_policy_chance ^ !!u->random_policy) {
1007                 fprintf(stderr, "uct: Only one of random_policy and random_policy_chance is set\n");
1008                 exit(1);
1009         }
1010
1011         if (u->fast_alloc && !u->parallel_tree) {
1012                 fprintf(stderr, "fast_alloc not supported with root parallelization.\n");
1013                 exit(1);
1014         }
1015
1016         if (!u->prior)
1017                 u->prior = uct_prior_init(NULL, b);
1018
1019         if (!u->playout)
1020                 u->playout = playout_moggy_init(NULL);
1021         u->playout->debug_level = u->debug_level;
1022
1023         u->ownermap.map = malloc(board_size2(b) * sizeof(u->ownermap.map[0]));
1024
1025         /* Some things remain uninitialized for now - the opening book
1026          * is not loaded and the tree not set up. */
1027         /* This will be initialized in setup_state() at the first move
1028          * received/requested. This is because right now we are not aware
1029          * about any komi or handicap setup and such. */
1030
1031         return u;
1032 }
1033
1034 struct engine *
1035 engine_uct_init(char *arg, struct board *b)
1036 {
1037         struct uct *u = uct_state_init(arg, b);
1038         struct engine *e = calloc(1, sizeof(struct engine));
1039         e->name = "UCT Engine";
1040         e->printhook = uct_printhook_ownermap;
1041         e->notify_play = uct_notify_play;
1042         e->chat = uct_chat;
1043         e->genmove = uct_genmove;
1044         e->dead_group_list = uct_dead_group_list;
1045         e->done = uct_done;
1046         e->data = u;
1047
1048         const char banner[] = "I'm playing UCT. When I'm losing, I will resign, "
1049                 "if I think I win, I play until you pass. "
1050                 "Anyone can send me 'winrate' in private chat to get my assessment of the position.";
1051         if (!u->banner) u->banner = "";
1052         e->comment = malloc(sizeof(banner) + strlen(u->banner) + 1);
1053         sprintf(e->comment, "%s %s", banner, u->banner);
1054
1055         return e;
1056 }