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
  39 /* How big proportion of ownermap counts must be of one color to consider
  40  * the point sure. */
  41 #define GJ_THRES        0.8
  42 /* How many games to consider at minimum before judging groups. */
  43 #define GJ_MINGAMES     500
  44
  45 /* How often to inspect the tree from the main thread to check for playout
  46  * stop, progress reports, etc. (in seconds) */
  47 #define TREE_BUSYWAIT_INTERVAL 0.1 /* 100ms */
  48
  49 /* For safety, use at most 3 times the desired time on a single move
  50  * in main time, and 1.1 times in byoyomi. */
  51 #define MAX_MAIN_TIME_EXTENSION 3.0
  52 #define MAX_BYOYOMI_TIME_EXTENSION 1.1
  53
  54 /* Once per how many simulations (per thread) to show a progress report line. */
  55 #define TREE_SIMPROGRESS_INTERVAL 10000
  56
  57
  58 static void
  59 setup_state(struct uct *u, struct board *b, enum stone color)
  60 {
  61         u->t = tree_init(b, color, u->fast_alloc ? u->max_tree_size: 0);
  62         if (u->force_seed)
  63                 fast_srandom(u->force_seed);
  64         if (UDEBUGL(0))
  65                 fprintf(stderr, "Fresh board with random seed %lu\n", fast_getseed());
  66         //board_print(b, stderr);
  67         if (!u->no_book && b->moves == 0) {
  68                 assert(color == S_BLACK);
  69                 tree_load(u->t, b);
  70         }
  71 }
  72
  73 static void
  74 reset_state(struct uct *u)
  75 {
  76         assert(u->t);
  77         tree_done(u->t); u->t = NULL;
  78 }
  79
  80 static void
  81 prepare_move(struct engine *e, struct board *b, enum stone color)
  82 {
  83         struct uct *u = e->data;
  84
  85         if (u->t) {
  86                 /* Verify that we have sane state. */
  87                 assert(b->es == u);
  88                 assert(u->t && b->moves);
  89                 if (color != stone_other(u->t->root_color)) {
  90                         fprintf(stderr, "Fatal: Non-alternating play detected %d %d\n",
  91                                 color, u->t->root_color);
  92                         exit(1);
  93                 }
  94
  95         } else {
  96                 /* We need fresh state. */
  97                 b->es = u;
  98                 setup_state(u, b, color);
  99         }
 100
 101         if (u->dynkomi && u->dynkomi > b->moves && (color & u->dynkomi_mask))
 102                 u->t->extra_komi = uct_get_extra_komi(u, b);
 103
 104         u->ownermap.playouts = 0;
 105         memset(u->ownermap.map, 0, board_size2(b) * sizeof(u->ownermap.map[0]));
 106 }
 107
 108 static void
 109 dead_group_list(struct uct *u, struct board *b, struct move_queue *mq)
 110 {
 111         struct group_judgement gj;
 112         gj.thres = GJ_THRES;
 113         gj.gs = alloca(board_size2(b) * sizeof(gj.gs[0]));
 114         board_ownermap_judge_group(b, &u->ownermap, &gj);
 115         groups_of_status(b, &gj, GS_DEAD, mq);
 116 }
 117
 118 bool
 119 uct_pass_is_safe(struct uct *u, struct board *b, enum stone color, bool pass_all_alive)
 120 {
 121         if (u->ownermap.playouts < GJ_MINGAMES)
 122                 return false;
 123
 124         struct move_queue mq = { .moves = 0 };
 125         if (!pass_all_alive)
 126                 dead_group_list(u, b, &mq);
 127         return pass_is_safe(b, color, &mq);
 128 }
 129
 130
 131 static void
 132 uct_printhook_ownermap(struct board *board, coord_t c, FILE *f)
 133 {
 134         struct uct *u = board->es;
 135         assert(u);
 136         const char chr[] = ":XO,"; // dame, black, white, unclear
 137         const char chm[] = ":xo,";
 138         char ch = chr[board_ownermap_judge_point(&u->ownermap, c, GJ_THRES)];
 139         if (ch == ',') { // less precise estimate then?
 140                 ch = chm[board_ownermap_judge_point(&u->ownermap, c, 0.67)];
 141         }
 142         fprintf(f, "%c ", ch);
 143 }
 144
 145 static char *
 146 uct_notify_play(struct engine *e, struct board *b, struct move *m)
 147 {
 148         struct uct *u = e->data;
 149         if (!u->t) {
 150                 /* No state, create one - this is probably game beginning
 151                  * and we need to load the opening book right now. */
 152                 prepare_move(e, b, m->color);
 153                 assert(u->t);
 154         }
 155
 156         /* Stop pondering. */
 157         /* XXX: If we are about to receive multiple 'play' commands,
 158          * e.g. in a rengo, we will not ponder during the rest of them. */
 159         uct_pondering_stop(u);
 160
 161         if (is_resign(m->coord)) {
 162                 /* Reset state. */
 163                 reset_state(u);
 164                 return NULL;
 165         }
 166
 167         /* Promote node of the appropriate move to the tree root. */
 168         assert(u->t->root);
 169         if (!tree_promote_at(u->t, b, m->coord)) {
 170                 if (UDEBUGL(0))
 171                         fprintf(stderr, "Warning: Cannot promote move node! Several play commands in row?\n");
 172                 reset_state(u);
 173                 return NULL;
 174         }
 175
 176         return NULL;
 177 }
 178
 179 static char *
 180 uct_chat(struct engine *e, struct board *b, char *cmd)
 181 {
 182         struct uct *u = e->data;
 183         static char reply[1024];
 184
 185         cmd += strspn(cmd, " \n\t");
 186         if (!strncasecmp(cmd, "winrate", 7)) {
 187                 if (!u->t)
 188                         return "no game context (yet?)";
 189                 enum stone color = u->t->root_color;
 190                 struct tree_node *n = u->t->root;
 191                 snprintf(reply, 1024, "In %d playouts at %d threads, %s %s can win with %.2f%% probability",
 192                          n->u.playouts, u->threads, stone2str(color), coord2sstr(n->coord, b),
 193                          tree_node_get_value(u->t, -1, n->u.value) * 100);
 194                 if (abs(u->t->extra_komi) >= 0.5) {
 195                         sprintf(reply + strlen(reply), ", while self-imposing extra komi %.1f",
 196                                 u->t->extra_komi);
 197                 }
 198                 strcat(reply, ".");
 199                 return reply;
 200         }
 201         return NULL;
 202 }
 203
 204 static void
 205 uct_dead_group_list(struct engine *e, struct board *b, struct move_queue *mq)
 206 {
 207         struct uct *u = e->data;
 208
 209         /* This means the game is probably over, no use pondering on. */
 210         uct_pondering_stop(u);
 211
 212         if (u->pass_all_alive)
 213                 return; // no dead groups
 214
 215         bool mock_state = false;
 216
 217         if (!u->t) {
 218                 /* No state, but we cannot just back out - we might
 219                  * have passed earlier, only assuming some stones are
 220                  * dead, and then re-connected, only to lose counting
 221                  * when all stones are assumed alive. */
 222                 /* Mock up some state and seed the ownermap by few
 223                  * simulations. */
 224                 prepare_move(e, b, S_BLACK); assert(u->t);
 225                 for (int i = 0; i < GJ_MINGAMES; i++)
 226                         uct_playout(u, b, S_BLACK, u->t);
 227                 mock_state = true;
 228         }
 229
 230         dead_group_list(u, b, mq);
 231
 232         if (mock_state) {
 233                 /* Clean up the mock state in case we will receive
 234                  * a genmove; we could get a non-alternating-move
 235                  * error from prepare_move() in that case otherwise. */
 236                 reset_state(u);
 237         }
 238 }
 239
 240 static void
 241 playout_policy_done(struct playout_policy *p)
 242 {
 243         if (p->done) p->done(p);
 244         if (p->data) free(p->data);
 245         free(p);
 246 }
 247
 248 static void
 249 uct_done(struct engine *e)
 250 {
 251         /* This is called on engine reset, especially when clear_board
 252          * is received and new game should begin. */
 253         struct uct *u = e->data;
 254         uct_pondering_stop(u);
 255         if (u->t) reset_state(u);
 256         free(u->ownermap.map);
 257
 258         free(u->policy);
 259         free(u->random_policy);
 260         playout_policy_done(u->playout);
 261         uct_prior_done(u->prior);
 262 }
 263
 264
 265 /* Pachi threading structure (if uct_playouts_parallel() is used):
 266  *
 267  * main thread
 268  *   |         main(), GTP communication, ...
 269  *   |         starts and stops the search managed by thread_manager
 270  *   |
 271  * thread_manager
 272  *   |         spawns and collects worker threads
 273  *   |
 274  * worker0
 275  * worker1
 276  * ...
 277  * workerK
 278  *             uct_playouts() loop, doing descend-playout until uct_halt
 279  *
 280  * Another way to look at it is by functions (lines denote thread boundaries):
 281  *
 282  * | uct_genmove()
 283  * | uct_search()            (uct_search_start() .. uct_search_stop())
 284  * | -----------------------
 285  * | spawn_thread_manager()
 286  * | -----------------------
 287  * | spawn_worker()
 288  * V uct_playouts() */
 289
 290 /* Set in thread manager in case the workers should stop. */
 291 volatile sig_atomic_t uct_halt = 0;
 292 /* ID of the running worker thread. */
 293 __thread int thread_id = -1;
 294 /* ID of the thread manager. */
 295 static pthread_t thread_manager;
 296 static bool thread_manager_running;
 297
 298 static pthread_mutex_t finish_mutex = PTHREAD_MUTEX_INITIALIZER;
 299 static pthread_cond_t finish_cond = PTHREAD_COND_INITIALIZER;
 300 static volatile int finish_thread;
 301 static pthread_mutex_t finish_serializer = PTHREAD_MUTEX_INITIALIZER;
 302
 303 struct spawn_ctx {
 304         int tid;
 305         struct uct *u;
 306         struct board *b;
 307         enum stone color;
 308         struct tree *t;
 309         unsigned long seed;
 310         int games;
 311 };
 312
 313 static void *
 314 spawn_worker(void *ctx_)
 315 {
 316         struct spawn_ctx *ctx = ctx_;
 317         /* Setup */
 318         fast_srandom(ctx->seed);
 319         thread_id = ctx->tid;
 320         /* Run */
 321         ctx->games = uct_playouts(ctx->u, ctx->b, ctx->color, ctx->t);
 322         /* Finish */
 323         pthread_mutex_lock(&finish_serializer);
 324         pthread_mutex_lock(&finish_mutex);
 325         finish_thread = ctx->tid;
 326         pthread_cond_signal(&finish_cond);
 327         pthread_mutex_unlock(&finish_mutex);
 328         return ctx;
 329 }
 330
 331 /* Thread manager, controlling worker threads. It must be called with
 332  * finish_mutex lock held, but it will unlock it itself before exiting;
 333  * this is necessary to be completely deadlock-free. */
 334 /* The finish_cond can be signalled for it to stop; in that case,
 335  * the caller should set finish_thread = -1. */
 336 /* After it is started, it will update mctx->t to point at some tree
 337  * used for the actual search (matters only for TM_ROOT), on return
 338  * it will set mctx->games to the number of performed simulations. */
 339 static void *
 340 spawn_thread_manager(void *ctx_)
 341 {
 342         /* In thread_manager, we use only some of the ctx fields. */
 343         struct spawn_ctx *mctx = ctx_;
 344         struct uct *u = mctx->u;
 345         struct tree *t = mctx->t;
 346         bool shared_tree = u->parallel_tree;
 347         fast_srandom(mctx->seed);
 348
 349         int played_games = 0;
 350         pthread_t threads[u->threads];
 351         int joined = 0;
 352
 353         uct_halt = 0;
 354
 355         /* Spawn threads... */
 356         for (int ti = 0; ti < u->threads; ti++) {
 357                 struct spawn_ctx *ctx = malloc(sizeof(*ctx));
 358                 ctx->u = u; ctx->b = mctx->b; ctx->color = mctx->color;
 359                 mctx->t = ctx->t = shared_tree ? t : tree_copy(t);
 360                 ctx->tid = ti; ctx->seed = fast_random(65536) + ti;
 361                 pthread_create(&threads[ti], NULL, spawn_worker, ctx);
 362                 if (UDEBUGL(2))
 363                         fprintf(stderr, "Spawned worker %d\n", ti);
 364         }
 365
 366         /* ...and collect them back: */
 367         while (joined < u->threads) {
 368                 /* Wait for some thread to finish... */
 369                 pthread_cond_wait(&finish_cond, &finish_mutex);
 370                 if (finish_thread < 0) {
 371                         /* Stop-by-caller. Tell the workers to wrap up. */
 372                         uct_halt = 1;
 373                         continue;
 374                 }
 375                 /* ...and gather its remnants. */
 376                 struct spawn_ctx *ctx;
 377                 pthread_join(threads[finish_thread], (void **) &ctx);
 378                 played_games += ctx->games;
 379                 joined++;
 380                 if (!shared_tree) {
 381                         if (ctx->t == mctx->t) mctx->t = t;
 382                         tree_merge(t, ctx->t);
 383                         tree_done(ctx->t);
 384                 }
 385                 free(ctx);
 386                 if (UDEBUGL(2))
 387                         fprintf(stderr, "Joined worker %d\n", finish_thread);
 388                 pthread_mutex_unlock(&finish_serializer);
 389         }
 390
 391         pthread_mutex_unlock(&finish_mutex);
 392
 393         if (!shared_tree)
 394                 tree_normalize(mctx->t, u->threads);
 395
 396         mctx->games = played_games;
 397         return mctx;
 398 }
 399
 400 static struct spawn_ctx *
 401 uct_search_start(struct uct *u, struct board *b, enum stone color, struct tree *t)
 402 {
 403         assert(u->threads > 0);
 404         assert(!thread_manager_running);
 405
 406         struct spawn_ctx ctx = { .u = u, .b = b, .color = color, .t = t, .seed = fast_random(65536) };
 407         static struct spawn_ctx mctx; mctx = ctx;
 408         pthread_mutex_lock(&finish_mutex);
 409         pthread_create(&thread_manager, NULL, spawn_thread_manager, &mctx);
 410         thread_manager_running = true;
 411         return &mctx;
 412 }
 413
 414 static struct spawn_ctx *
 415 uct_search_stop(void)
 416 {
 417         assert(thread_manager_running);
 418
 419         /* Signal thread manager to stop the workers. */
 420         pthread_mutex_lock(&finish_mutex);
 421         finish_thread = -1;
 422         pthread_cond_signal(&finish_cond);
 423         pthread_mutex_unlock(&finish_mutex);
 424
 425         /* Collect the thread manager. */
 426         struct spawn_ctx *pctx;
 427         thread_manager_running = false;
 428         pthread_join(thread_manager, (void **) &pctx);
 429         return pctx;
 430 }
 431
 432
 433 /* Search stopping conditions */
 434 union stop_conditions {
 435         struct { // TD_WALLTIME
 436                 double desired_stop; /* stop at that time if possible */
 437                 double worst_stop;   /* stop no later than this */
 438         } t;
 439         struct { // TD_GAMES
 440                 int desired_playouts;
 441                 int worst_playouts;
 442         } p;
 443 };
 444
 445 /* Pre-process time_info for search control and sets the desired stopping conditions. */
 446 static void
 447 time_prep(struct time_info *ti, struct uct *u, struct board *b, union stop_conditions *stop)
 448 {
 449         assert(ti->period != TT_TOTAL);
 450
 451         if (ti->period == TT_NULL) {
 452                 ti->period = TT_MOVE;
 453                 ti->dim = TD_GAMES;
 454                 ti->len.games = MC_GAMES;
 455         }
 456         if (ti->dim == TD_GAMES) {
 457                 stop->p.desired_playouts = ti->len.games;
 458                 /* We force worst == desired, so note that we will not loop
 459                  * until best == winner. */
 460                 stop->p.worst_playouts = ti->len.games;
 461
 462         } else {
 463                 double desired_time = ti->len.t.recommended_time;
 464                 double worst_time;
 465                 if (time_in_byoyomi(ti)) {
 466                         // make recommended == average(desired, worst)
 467                         worst_time = desired_time * MAX_BYOYOMI_TIME_EXTENSION;
 468                         desired_time *= (2 - MAX_BYOYOMI_TIME_EXTENSION);
 469
 470                 } else {
 471                         int bsize = (board_size(b)-2)*(board_size(b)-2);
 472                         int fuseki_end = u->fuseki_end * bsize / 100; // move nb at fuseki end
 473                         int yose_start = u->yose_start * bsize / 100; // move nb at yose start
 474
 475                         int left_at_yose_start = (b->moves - yose_start) / 2 + board_estimated_moves_left(b);
 476                         // ^- /2 because we only consider the moves we have to play ourselves
 477                         if (left_at_yose_start < MIN_MOVES_LEFT)
 478                                 left_at_yose_start = MIN_MOVES_LEFT;
 479                         double longest_time = ti->len.t.max_time / left_at_yose_start;
 480                         if (longest_time < desired_time) {
 481                                 // Should rarely happen, but keep desired_time anyway
 482                         } else if (b->moves < fuseki_end) {
 483                                 desired_time += ((longest_time - desired_time) * b->moves) / fuseki_end;
 484                                 /* In this branch fuseki_end can't be 0 */
 485                         } else if (b->moves < yose_start) {
 486                                 desired_time = longest_time;
 487                         }
 488                         worst_time = desired_time * MAX_MAIN_TIME_EXTENSION;
 489                 }
 490                 if (worst_time > ti->len.t.max_time)
 491                         worst_time = ti->len.t.max_time;
 492                 if (desired_time > worst_time)
 493                         desired_time = worst_time;
 494
 495                 stop->t.desired_stop = ti->len.t.timer_start + desired_time - ti->len.t.net_lag;
 496                 stop->t.worst_stop = ti->len.t.timer_start + worst_time - ti->len.t.net_lag;
 497                 // Both stop points may be in the past if too much lag.
 498
 499                 if (UDEBUGL(2))
 500                         fprintf(stderr, "desired time %.02f, worst %.02f\n", desired_time, worst_time);
 501         }
 502 }
 503
 504 /* Run time-limited MCTS search on foreground. */
 505 static int
 506 uct_search(struct uct *u, struct board *b, struct time_info *ti, enum stone color, struct tree *t)
 507 {
 508         union stop_conditions stop;
 509         time_prep(ti, u, b, &stop);
 510         if (UDEBUGL(2) && u->t->root->u.playouts > 0)
 511                 fprintf(stderr, "<pre-simulated %d games skipped>\n", u->t->root->u.playouts);
 512
 513         /* Number of last game with progress print. */
 514         int last_print = t->root->u.playouts;
 515         /* Number of simulations to wait before next print. */
 516         int print_interval = TREE_SIMPROGRESS_INTERVAL * (u->thread_model == TM_ROOT ? 1 : u->threads);
 517         /* Printed notification about full memory? */
 518         bool print_fullmem = false;
 519
 520         struct spawn_ctx *ctx = uct_search_start(u, b, color, t);
 521
 522         /* The search tree is ctx->t. This is normally == t, but in case of
 523          * TM_ROOT, it is one of the trees belonging to the independent
 524          * workers. It is important to reference ctx->t directly since the
 525          * thread manager will swap the tree pointer asynchronously. */
 526         /* XXX: This means TM_ROOT support is suboptimal since single stalled
 527          * thread can stall the others in case of limiting the search by game
 528          * count. However, TM_ROOT just does not deserve any more extra code
 529          * right now. */
 530
 531         struct tree_node *best = NULL, *prev_best;
 532         struct tree_node *winner = NULL, *prev_winner;
 533
 534         double busywait_interval = TREE_BUSYWAIT_INTERVAL;
 535
 536         /* Now, just periodically poll the search tree. */
 537         while (1) {
 538                 time_sleep(busywait_interval);
 539                 /* busywait_interval should never be less than desired time, or the
 540                  * time control is broken. But if it happens to be less, we still search
 541                  * at least 100ms otherwise the move is completely random. */
 542
 543                 int i = ctx->t->root->u.playouts;
 544
 545                 /* Print progress? */
 546                 if (i - last_print > print_interval) {
 547                         last_print += print_interval; // keep the numbers tidy
 548                         uct_progress_status(u, ctx->t, color, last_print);
 549                 }
 550                 if (!print_fullmem && ctx->t->nodes_size > u->max_tree_size) {
 551                         if (UDEBUGL(2))
 552                                 fprintf(stderr, "memory limit hit (%ld > %lu)\n", ctx->t->nodes_size, u->max_tree_size);
 553                         print_fullmem = true;
 554                 }
 555
 556                 /* Check against time settings. */
 557                 bool desired_done = false;
 558                 if (ti->dim == TD_WALLTIME) {
 559                         double now = time_now();
 560                         if (now > stop.t.worst_stop) break;
 561                         desired_done = now > stop.t.desired_stop;
 562                 } else {
 563                         assert(ti->dim == TD_GAMES);
 564                         if (i > stop.p.worst_playouts) break;
 565                         desired_done = i > stop.p.desired_playouts;
 566                 }
 567
 568                 /* Early break in won situation. */
 569                 prev_best = best;
 570                 best = u->policy->choose(u->policy, ctx->t->root, b, color);
 571                 if (best && ((best->u.playouts >= 2000 && tree_node_get_value(ctx->t, 1, best->u.value) >= u->loss_threshold)
 572                              || (best->u.playouts >= 500 && tree_node_get_value(ctx->t, 1, best->u.value) >= 0.95)))
 573                         break;
 574
 575                 if (desired_done) {
 576                         if (!u->policy->winner || !u->policy->evaluate)
 577                                 break;
 578                         /* Stop only if best explored has also highest value: */
 579                         prev_winner = winner;
 580                         winner = u->policy->winner(u->policy, ctx->t, ctx->t->root);
 581                         if (best && best == winner)
 582                                 break;
 583                         if (UDEBUGL(3) && (best != prev_best || winner != prev_winner)) {
 584                                 fprintf(stderr, "[%d] best", i);
 585                                 if (best)
 586                                         fprintf(stderr, " %3s [%d] %f", coord2sstr(best->coord, ctx->t->board),
 587                                                 best->u.playouts, tree_node_get_value(ctx->t, 1, best->u.value));
 588                                 fprintf(stderr, " != winner");
 589                                 if (winner)
 590                                         fprintf(stderr, " %3s [%d] %f ", coord2sstr(winner->coord, ctx->t->board),
 591                                                 winner->u.playouts, tree_node_get_value(ctx->t, 1, winner->u.value));
 592                                 fprintf(stderr, "\n");
 593                         }
 594                 }
 595
 596                 /* TODO: Early break if best->variance goes under threshold and we already
 597                  * have enough playouts (possibly thanks to book or to pondering). */
 598                 /* TODO: Early break if second best has no chance to catch up. */
 599         }
 600
 601         ctx = uct_search_stop();
 602
 603         if (UDEBUGL(2))
 604                 tree_dump(t, u->dumpthres);
 605         if (UDEBUGL(0))
 606                 uct_progress_status(u, t, color, ctx->games);
 607
 608         return ctx->games;
 609 }
 610
 611
 612 /* Start pondering background with @color to play. */
 613 static void
 614 uct_pondering_start(struct uct *u, struct board *b0, struct tree *t, enum stone color)
 615 {
 616         if (UDEBUGL(1))
 617                 fprintf(stderr, "Starting to ponder with color %s\n", stone2str(stone_other(color)));
 618         u->pondering = true;
 619
 620         /* We need a local board copy to ponder upon. */
 621         struct board *b = malloc(sizeof(*b)); board_copy(b, b0);
 622
 623         /* *b0 did not have the genmove'd move played yet. */
 624         struct move m = { t->root->coord, t->root_color };
 625         int res = board_play(b, &m);
 626         assert(res >= 0);
 627
 628         /* Start MCTS manager thread "headless". */
 629         uct_search_start(u, b, color, t);
 630 }
 631
 632 /* uct_search_stop() frontend for the pondering (non-genmove) mode. */
 633 static void
 634 uct_pondering_stop(struct uct *u)
 635 {
 636         u->pondering = false;
 637         if (!thread_manager_running)
 638                 return;
 639
 640         /* Stop the thread manager. */
 641         struct spawn_ctx *ctx = uct_search_stop();
 642         if (UDEBUGL(1)) {
 643                 fprintf(stderr, "(pondering) ");
 644                 uct_progress_status(u, ctx->t, ctx->color, ctx->games);
 645         }
 646         free(ctx->b);
 647 }
 648
 649
 650 static coord_t *
 651 uct_genmove(struct engine *e, struct board *b, struct time_info *ti, enum stone color, bool pass_all_alive)
 652 {
 653         double start_time = time_now();
 654         struct uct *u = e->data;
 655
 656         if (b->superko_violation) {
 657                 fprintf(stderr, "!!! WARNING: SUPERKO VIOLATION OCCURED BEFORE THIS MOVE\n");
 658                 fprintf(stderr, "Maybe you play with situational instead of positional superko?\n");
 659                 fprintf(stderr, "I'm going to ignore the violation, but note that I may miss\n");
 660                 fprintf(stderr, "some moves valid under this ruleset because of this.\n");
 661                 b->superko_violation = false;
 662         }
 663
 664         /* Seed the tree. */
 665         uct_pondering_stop(u);
 666         prepare_move(e, b, color);
 667         assert(u->t);
 668
 669         /* Perform the Monte Carlo Tree Search! */
 670         int played_games = uct_search(u, b, ti, color, u->t);
 671
 672         /* Choose the best move from the tree. */
 673         struct tree_node *best = u->policy->choose(u->policy, u->t->root, b, color);
 674         if (!best) {
 675                 reset_state(u);
 676                 return coord_copy(pass);
 677         }
 678         if (UDEBUGL(1))
 679                 fprintf(stderr, "*** WINNER is %s (%d,%d) with score %1.4f (%d/%d:%d games)\n",
 680                         coord2sstr(best->coord, b), coord_x(best->coord, b), coord_y(best->coord, b),
 681                         tree_node_get_value(u->t, 1, best->u.value),
 682                         best->u.playouts, u->t->root->u.playouts, played_games);
 683
 684         /* Do not resign if we're so short of time that evaluation of best move is completely
 685          * unreliable, we might be winning actually. In this case best is almost random but
 686          * still better than resign. */
 687         if (tree_node_get_value(u->t, 1, best->u.value) < u->resign_ratio && !is_pass(best->coord)
 688             && best->u.playouts > GJ_MINGAMES) {
 689                 reset_state(u);
 690                 return coord_copy(resign);
 691         }
 692
 693         /* If the opponent just passed and we win counting, always
 694          * pass as well. */
 695         if (b->moves > 1 && is_pass(b->last_move.coord)) {
 696                 /* Make sure enough playouts are simulated. */
 697                 while (u->ownermap.playouts < GJ_MINGAMES)
 698                         uct_playout(u, b, color, u->t);
 699                 if (uct_pass_is_safe(u, b, color, u->pass_all_alive || pass_all_alive)) {
 700                         if (UDEBUGL(0))
 701                                 fprintf(stderr, "<Will rather pass, looks safe enough.>\n");
 702                         best->coord = pass;
 703                 }
 704         }
 705
 706         tree_promote_node(u->t, &best);
 707         /* After a pass, pondering is harmful for two reasons:
 708          * (i) We might keep pondering even when the game is over.
 709          * Of course this is the case for opponent resign as well.
 710          * (ii) More importantly, the ownermap will get skewed since
 711          * the UCT will start cutting off any playouts. */
 712         if (u->pondering_opt && !is_pass(best->coord)) {
 713                 uct_pondering_start(u, b, u->t, stone_other(color));
 714         }
 715         if (UDEBUGL(2)) {
 716                 double time = time_now() - start_time + 0.000001; /* avoid divide by zero */
 717                 fprintf(stderr, "genmove in %0.2fs (%d games/s, %d games/s/thread)\n",
 718                         time, (int)(played_games/time), (int)(played_games/time/u->threads));
 719         }
 720         return coord_copy(best->coord);
 721 }
 722
 723
 724 bool
 725 uct_genbook(struct engine *e, struct board *b, struct time_info *ti, enum stone color)
 726 {
 727         struct uct *u = e->data;
 728         if (!u->t) prepare_move(e, b, color);
 729         assert(u->t);
 730
 731         if (ti->dim == TD_GAMES) {
 732                 /* Don't count in games that already went into the book. */
 733                 ti->len.games += u->t->root->u.playouts;
 734         }
 735         uct_search(u, b, ti, color, u->t);
 736
 737         assert(ti->dim == TD_GAMES);
 738         tree_save(u->t, b, ti->len.games / 100);
 739
 740         return true;
 741 }
 742
 743 void
 744 uct_dumpbook(struct engine *e, struct board *b, enum stone color)
 745 {
 746         struct uct *u = e->data;
 747         struct tree *t = tree_init(b, color, u->fast_alloc ? u->max_tree_size: 0);
 748         tree_load(t, b);
 749         tree_dump(t, 0);
 750         tree_done(t);
 751 }
 752
 753
 754 struct uct *
 755 uct_state_init(char *arg, struct board *b)
 756 {
 757         struct uct *u = calloc(1, sizeof(struct uct));
 758
 759         u->debug_level = 1;
 760         u->gamelen = MC_GAMELEN;
 761         u->mercymin = 0;
 762         u->expand_p = 2;
 763         u->dumpthres = 1000;
 764         u->playout_amaf = true;
 765         u->playout_amaf_nakade = false;
 766         u->amaf_prior = false;
 767         u->max_tree_size = 3072ULL * 1048576;
 768
 769         if (board_size(b) - 2 >= 19)
 770                 u->dynkomi = 200;
 771         u->dynkomi_mask = S_BLACK;
 772
 773         u->threads = 1;
 774         u->thread_model = TM_TREEVL;
 775         u->parallel_tree = true;
 776         u->virtual_loss = true;
 777         u->fuseki_end = 20; // max time at 361*20% = 72 moves (our 36th move, still 99 to play)
 778         u->yose_start = 40; // (100-40-25)*361/100/2 = 63 moves still to play by us then
 779
 780         u->val_scale = 0.04; u->val_points = 40;
 781
 782         if (arg) {
 783                 char *optspec, *next = arg;
 784                 while (*next) {
 785                         optspec = next;
 786                         next += strcspn(next, ",");
 787                         if (*next) { *next++ = 0; } else { *next = 0; }
 788
 789                         char *optname = optspec;
 790                         char *optval = strchr(optspec, '=');
 791                         if (optval) *optval++ = 0;
 792
 793                         if (!strcasecmp(optname, "debug")) {
 794                                 if (optval)
 795                                         u->debug_level = atoi(optval);
 796                                 else
 797                                         u->debug_level++;
 798                         } else if (!strcasecmp(optname, "mercy") && optval) {
 799                                 /* Minimal difference of black/white captures
 800                                  * to stop playout - "Mercy Rule". Speeds up
 801                                  * hopeless playouts at the expense of some
 802                                  * accuracy. */
 803                                 u->mercymin = atoi(optval);
 804                         } else if (!strcasecmp(optname, "gamelen") && optval) {
 805                                 u->gamelen = atoi(optval);
 806                         } else if (!strcasecmp(optname, "expand_p") && optval) {
 807                                 u->expand_p = atoi(optval);
 808                         } else if (!strcasecmp(optname, "dumpthres") && optval) {
 809                                 u->dumpthres = atoi(optval);
 810                         } else if (!strcasecmp(optname, "playout_amaf")) {
 811                                 /* Whether to include random playout moves in
 812                                  * AMAF as well. (Otherwise, only tree moves
 813                                  * are included in AMAF. Of course makes sense
 814                                  * only in connection with an AMAF policy.) */
 815                                 /* with-without: 55.5% (+-4.1) */
 816                                 if (optval && *optval == '0')
 817                                         u->playout_amaf = false;
 818                                 else
 819                                         u->playout_amaf = true;
 820                         } else if (!strcasecmp(optname, "playout_amaf_nakade")) {
 821                                 /* Whether to include nakade moves from playouts
 822                                  * in the AMAF statistics; this tends to nullify
 823                                  * the playout_amaf effect by adding too much
 824                                  * noise. */
 825                                 if (optval && *optval == '0')
 826                                         u->playout_amaf_nakade = false;
 827                                 else
 828                                         u->playout_amaf_nakade = true;
 829                         } else if (!strcasecmp(optname, "playout_amaf_cutoff") && optval) {
 830                                 /* Keep only first N% of playout stage AMAF
 831                                  * information. */
 832                                 u->playout_amaf_cutoff = atoi(optval);
 833                         } else if ((!strcasecmp(optname, "policy") || !strcasecmp(optname, "random_policy")) && optval) {
 834                                 char *policyarg = strchr(optval, ':');
 835                                 struct uct_policy **p = !strcasecmp(optname, "policy") ? &u->policy : &u->random_policy;
 836                                 if (policyarg)
 837                                         *policyarg++ = 0;
 838                                 if (!strcasecmp(optval, "ucb1")) {
 839                                         *p = policy_ucb1_init(u, policyarg);
 840                                 } else if (!strcasecmp(optval, "ucb1amaf")) {
 841                                         *p = policy_ucb1amaf_init(u, policyarg);
 842                                 } else {
 843                                         fprintf(stderr, "UCT: Invalid tree policy %s\n", optval);
 844                                         exit(1);
 845                                 }
 846                         } else if (!strcasecmp(optname, "playout") && optval) {
 847                                 char *playoutarg = strchr(optval, ':');
 848                                 if (playoutarg)
 849                                         *playoutarg++ = 0;
 850                                 if (!strcasecmp(optval, "moggy")) {
 851                                         u->playout = playout_moggy_init(playoutarg);
 852                                 } else if (!strcasecmp(optval, "light")) {
 853                                         u->playout = playout_light_init(playoutarg);
 854                                 } else if (!strcasecmp(optval, "elo")) {
 855                                         u->playout = playout_elo_init(playoutarg);
 856                                 } else {
 857                                         fprintf(stderr, "UCT: Invalid playout policy %s\n", optval);
 858                                         exit(1);
 859                                 }
 860                         } else if (!strcasecmp(optname, "prior") && optval) {
 861                                 u->prior = uct_prior_init(optval, b);
 862                         } else if (!strcasecmp(optname, "amaf_prior") && optval) {
 863                                 u->amaf_prior = atoi(optval);
 864                         } else if (!strcasecmp(optname, "threads") && optval) {
 865                                 /* By default, Pachi will run with only single
 866                                  * tree search thread! */
 867                                 u->threads = atoi(optval);
 868                         } else if (!strcasecmp(optname, "thread_model") && optval) {
 869                                 if (!strcasecmp(optval, "root")) {
 870                                         /* Root parallelization - each thread
 871                                          * does independent search, trees are
 872                                          * merged at the end. */
 873                                         u->thread_model = TM_ROOT;
 874                                         u->parallel_tree = false;
 875                                         u->virtual_loss = false;
 876                                 } else if (!strcasecmp(optval, "tree")) {
 877                                         /* Tree parallelization - all threads
 878                                          * grind on the same tree. */
 879                                         u->thread_model = TM_TREE;
 880                                         u->parallel_tree = true;
 881                                         u->virtual_loss = false;
 882                                 } else if (!strcasecmp(optval, "treevl")) {
 883                                         /* Tree parallelization, but also
 884                                          * with virtual losses - this discou-
 885                                          * rages most threads choosing the
 886                                          * same tree branches to read. */
 887                                         u->thread_model = TM_TREEVL;
 888                                         u->parallel_tree = true;
 889                                         u->virtual_loss = true;
 890                                 } else {
 891                                         fprintf(stderr, "UCT: Invalid thread model %s\n", optval);
 892                                         exit(1);
 893                                 }
 894                         } else if (!strcasecmp(optname, "pondering")) {
 895                                 /* Keep searching even during opponent's turn. */
 896                                 u->pondering_opt = !optval || atoi(optval);
 897                         } else if (!strcasecmp(optname, "fuseki_end") && optval) {
 898                                 /* At the very beginning it's not worth thinking
 899                                  * too long because the playout evaluations are
 900                                  * very noisy. So gradually increase the thinking
 901                                  * time up to maximum when fuseki_end percent
 902                                  * of the board has been played.
 903                                  * This only applies if we are not in byoyomi. */
 904                                 u->fuseki_end = atoi(optval);
 905                         } else if (!strcasecmp(optname, "yose_start") && optval) {
 906                                 /* When yose_start percent of the board has been
 907                                  * played, or if we are in byoyomi, stop spending
 908                                  * more time and spread the remaining time
 909                                  * uniformly.
 910                                  * Between fuseki_end and yose_start, we spend
 911                                  * a constant proportion of the remaining time
 912                                  * on each move. (yose_start should actually
 913                                  * be much earlier than when real yose start,
 914                                  * but "yose" is a good short name to convey
 915                                  * the idea.) */
 916                                 u->yose_start = atoi(optval);
 917                         } else if (!strcasecmp(optname, "force_seed") && optval) {
 918                                 u->force_seed = atoi(optval);
 919                         } else if (!strcasecmp(optname, "no_book")) {
 920                                 u->no_book = true;
 921                         } else if (!strcasecmp(optname, "dynkomi")) {
 922                                 /* Dynamic komi in handicap game; linearly
 923                                  * decreases to basic settings until move
 924                                  * #optval. */
 925                                 u->dynkomi = optval ? atoi(optval) : 150;
 926                         } else if (!strcasecmp(optname, "dynkomi_mask") && optval) {
 927                                 /* Bitmask of colors the player must be
 928                                  * for dynkomi be applied; you may want
 929                                  * to use dynkomi_mask=3 to allow dynkomi
 930                                  * even in games where Pachi is white. */
 931                                 u->dynkomi_mask = atoi(optval);
 932                         } else if (!strcasecmp(optname, "val_scale") && optval) {
 933                                 /* How much of the game result value should be
 934                                  * influenced by win size. Zero means it isn't. */
 935                                 u->val_scale = atof(optval);
 936                         } else if (!strcasecmp(optname, "val_points") && optval) {
 937                                 /* Maximum size of win to be scaled into game
 938                                  * result value. Zero means boardsize^2. */
 939                                 u->val_points = atoi(optval) * 2; // result values are doubled
 940                         } else if (!strcasecmp(optname, "val_extra")) {
 941                                 /* If false, the score coefficient will be simply
 942                                  * added to the value, instead of scaling the result
 943                                  * coefficient because of it. */
 944                                 u->val_extra = !optval || atoi(optval);
 945                         } else if (!strcasecmp(optname, "root_heuristic") && optval) {
 946                                 /* Whether to bias exploration by root node values
 947                                  * (must be supported by the used policy).
 948                                  * 0: Don't.
 949                                  * 1: Do, value = result.
 950                                  * Try to temper the result:
 951                                  * 2: Do, value = 0.5+(result-expected)/2.
 952                                  * 3: Do, value = 0.5+bzz((result-expected)^2). */
 953                                 u->root_heuristic = atoi(optval);
 954                         } else if (!strcasecmp(optname, "pass_all_alive")) {
 955                                 /* Whether to consider all stones alive at the game
 956                                  * end instead of marking dead groupd. */
 957                                 u->pass_all_alive = !optval || atoi(optval);
 958                         } else if (!strcasecmp(optname, "random_policy_chance") && optval) {
 959                                 /* If specified (N), with probability 1/N, random_policy policy
 960                                  * descend is used instead of main policy descend; useful
 961                                  * if specified policy (e.g. UCB1AMAF) can make unduly biased
 962                                  * choices sometimes, you can fall back to e.g.
 963                                  * random_policy=UCB1. */
 964                                 u->random_policy_chance = atoi(optval);
 965                         } else if (!strcasecmp(optname, "max_tree_size") && optval) {
 966                                 /* Maximum amount of memory [MiB] consumed by the move tree.
 967                                  * Default is 3072 (3 GiB). Note that if you use TM_ROOT,
 968                                  * this limits size of only one of the trees, not all of them
 969                                  * together. */
 970                                 u->max_tree_size = atol(optval) * 1048576;
 971                         } else if (!strcasecmp(optname, "banner") && optval) {
 972                                 /* Additional banner string. This must come as the
 973                                  * last engine parameter. */
 974                                 if (*next) *--next = ',';
 975                                 u->banner = strdup(optval);
 976                                 break;
 977                         } else {
 978                                 fprintf(stderr, "uct: Invalid engine argument %s or missing value\n", optname);
 979                                 exit(1);
 980                         }
 981                 }
 982         }
 983
 984         u->resign_ratio = 0.2; /* Resign when most games are lost. */
 985         u->loss_threshold = 0.85; /* Stop reading if after at least 5000 playouts this is best value. */
 986         if (!u->policy)
 987                 u->policy = policy_ucb1amaf_init(u, NULL);
 988
 989         if (!!u->random_policy_chance ^ !!u->random_policy) {
 990                 fprintf(stderr, "uct: Only one of random_policy and random_policy_chance is set\n");
 991                 exit(1);
 992         }
 993
 994         if (u->fast_alloc && !u->parallel_tree) {
 995                 fprintf(stderr, "fast_alloc not supported with root parallelization.\n");
 996                 exit(1);
 997         }
 998
 999         if (!u->prior)
1000                 u->prior = uct_prior_init(NULL, b);
1001
1002         if (!u->playout)
1003                 u->playout = playout_moggy_init(NULL);
1004         u->playout->debug_level = u->debug_level;
1005
1006         u->ownermap.map = malloc(board_size2(b) * sizeof(u->ownermap.map[0]));
1007
1008         /* Some things remain uninitialized for now - the opening book
1009          * is not loaded and the tree not set up. */
1010         /* This will be initialized in setup_state() at the first move
1011          * received/requested. This is because right now we are not aware
1012          * about any komi or handicap setup and such. */
1013
1014         return u;
1015 }
1016
1017 struct engine *
1018 engine_uct_init(char *arg, struct board *b)
1019 {
1020         struct uct *u = uct_state_init(arg, b);
1021         struct engine *e = calloc(1, sizeof(struct engine));
1022         e->name = "UCT Engine";
1023         e->printhook = uct_printhook_ownermap;
1024         e->notify_play = uct_notify_play;
1025         e->chat = uct_chat;
1026         e->genmove = uct_genmove;
1027         e->dead_group_list = uct_dead_group_list;
1028         e->done = uct_done;
1029         e->data = u;
1030
1031         const char banner[] = "I'm playing UCT. When I'm losing, I will resign, "
1032                 "if I think I win, I play until you pass. "
1033                 "Anyone can send me 'winrate' in private chat to get my assessment of the position.";
1034         if (!u->banner) u->banner = "";
1035         e->comment = malloc(sizeof(banner) + strlen(u->banner) + 1);
1036         sprintf(e->comment, "%s %s", banner, u->banner);
1037
1038         return e;
1039 }