uct/search.c

   1 #include <assert.h>
   2 #include <math.h>
   3 #include <pthread.h>
   4 #include <signal.h>
   5 #include <stdio.h>
   6 #include <stdlib.h>
   7 #include <string.h>
   8 #include <time.h>
   9
  10 #define DEBUG
  11
  12 #include "debug.h"
  13 #include "distributed/distributed.h"
  14 #include "move.h"
  15 #include "random.h"
  16 #include "timeinfo.h"
  17 #include "uct/dynkomi.h"
  18 #include "uct/internal.h"
  19 #include "uct/search.h"
  20 #include "uct/tree.h"
  21 #include "uct/uct.h"
  22 #include "uct/walk.h"
  23
  24
  25 /* Default number of simulations to perform per move.
  26  * Note that this is now in total over all threads!. */
  27 #define MC_GAMES        80000
  28 static const struct time_info default_ti = {
  29         .period = TT_MOVE,
  30         .dim = TD_GAMES,
  31         .len = { .games = MC_GAMES },
  32 };
  33
  34 /* Once per how many simulations (per thread) to show a progress report line. */
  35 #define TREE_SIMPROGRESS_INTERVAL 10000
  36
  37 /* When terminating UCT search early, the safety margin to add to the
  38  * remaining playout number estimate when deciding whether the result can
  39  * still change. */
  40 #define PLAYOUT_DELTA_SAFEMARGIN 1000
  41
  42 /* Minimal number of simulations to consider early break. */
  43 #define PLAYOUT_EARLY_BREAK_MIN 5000
  44
  45 /* Minimal time to consider early break (in seconds). */
  46 #define TIME_EARLY_BREAK_MIN 1.0
  47
  48
  49 /* Pachi threading structure:
  50  *
  51  * main thread
  52  *   |         main(), GTP communication, ...
  53  *   |         starts and stops the search managed by thread_manager
  54  *   |
  55  * thread_manager
  56  *   |         spawns and collects worker threads
  57  *   |
  58  * worker0
  59  * worker1
  60  * ...
  61  * workerK
  62  *             uct_playouts() loop, doing descend-playout until uct_halt
  63  *
  64  * Another way to look at it is by functions (lines denote thread boundaries):
  65  *
  66  * | uct_genmove()
  67  * | uct_search()            (uct_search_start() .. uct_search_stop())
  68  * | -----------------------
  69  * | spawn_thread_manager()
  70  * | -----------------------
  71  * | spawn_worker()
  72  * V uct_playouts() */
  73
  74 /* Set in thread manager in case the workers should stop. */
  75 volatile sig_atomic_t uct_halt = 0;
  76 /* ID of the thread manager. */
  77 static pthread_t thread_manager;
  78 bool thread_manager_running;
  79
  80 static pthread_mutex_t finish_mutex = PTHREAD_MUTEX_INITIALIZER;
  81 static pthread_cond_t finish_cond = PTHREAD_COND_INITIALIZER;
  82 static volatile int finish_thread;
  83 static pthread_mutex_t finish_serializer = PTHREAD_MUTEX_INITIALIZER;
  84
  85 static void *
  86 spawn_worker(void *ctx_)
  87 {
  88         struct uct_thread_ctx *ctx = ctx_;
  89         /* Setup */
  90         fast_srandom(ctx->seed);
  91         /* Run */
  92         ctx->games = uct_playouts(ctx->u, ctx->b, ctx->color, ctx->t, ctx->ti);
  93         /* Finish */
  94         pthread_mutex_lock(&finish_serializer);
  95         pthread_mutex_lock(&finish_mutex);
  96         finish_thread = ctx->tid;
  97         pthread_cond_signal(&finish_cond);
  98         pthread_mutex_unlock(&finish_mutex);
  99         return ctx;
 100 }
 101
 102 /* Thread manager, controlling worker threads. It must be called with
 103  * finish_mutex lock held, but it will unlock it itself before exiting;
 104  * this is necessary to be completely deadlock-free. */
 105 /* The finish_cond can be signalled for it to stop; in that case,
 106  * the caller should set finish_thread = -1. */
 107 /* After it is started, it will update mctx->t to point at some tree
 108  * used for the actual search, on return
 109  * it will set mctx->games to the number of performed simulations. */
 110 static void *
 111 spawn_thread_manager(void *ctx_)
 112 {
 113         /* In thread_manager, we use only some of the ctx fields. */
 114         struct uct_thread_ctx *mctx = ctx_;
 115         struct uct *u = mctx->u;
 116         struct tree *t = mctx->t;
 117         fast_srandom(mctx->seed);
 118
 119         int played_games = 0;
 120         pthread_t threads[u->threads];
 121         int joined = 0;
 122
 123         uct_halt = 0;
 124
 125         /* Garbage collect the tree by preference when pondering. */
 126         if (u->pondering && t->nodes && t->nodes_size >= t->pruning_threshold) {
 127                 t->root = tree_garbage_collect(t, t->root);
 128         }
 129
 130         /* Spawn threads... */
 131         for (int ti = 0; ti < u->threads; ti++) {
 132                 struct uct_thread_ctx *ctx = malloc2(sizeof(*ctx));
 133                 ctx->u = u; ctx->b = mctx->b; ctx->color = mctx->color;
 134                 mctx->t = ctx->t = t;
 135                 ctx->tid = ti; ctx->seed = fast_random(65536) + ti;
 136                 ctx->ti = mctx->ti;
 137                 pthread_create(&threads[ti], NULL, spawn_worker, ctx);
 138                 if (UDEBUGL(3))
 139                         fprintf(stderr, "Spawned worker %d\n", ti);
 140         }
 141
 142         /* ...and collect them back: */
 143         while (joined < u->threads) {
 144                 /* Wait for some thread to finish... */
 145                 pthread_cond_wait(&finish_cond, &finish_mutex);
 146                 if (finish_thread < 0) {
 147                         /* Stop-by-caller. Tell the workers to wrap up
 148                          * and unblock them from terminating. */
 149                         uct_halt = 1;
 150                         /* We need to make sure the workers do not complete
 151                          * the termination sequence before we get officially
 152                          * stopped - their wake and the stop wake could get
 153                          * coalesced. */
 154                         pthread_mutex_unlock(&finish_serializer);
 155                         continue;
 156                 }
 157                 /* ...and gather its remnants. */
 158                 struct uct_thread_ctx *ctx;
 159                 pthread_join(threads[finish_thread], (void **) &ctx);
 160                 played_games += ctx->games;
 161                 joined++;
 162                 free(ctx);
 163                 if (UDEBUGL(3))
 164                         fprintf(stderr, "Joined worker %d\n", finish_thread);
 165                 pthread_mutex_unlock(&finish_serializer);
 166         }
 167
 168         pthread_mutex_unlock(&finish_mutex);
 169
 170         mctx->games = played_games;
 171         return mctx;
 172 }
 173
 174
 175 /*** THREAD MANAGER end */
 176
 177 /*** Search infrastructure: */
 178
 179
 180 int
 181 uct_search_games(struct uct_search_state *s)
 182 {
 183         return s->ctx->t->root->u.playouts;
 184 }
 185
 186 void
 187 uct_search_start(struct uct *u, struct board *b, enum stone color,
 188                  struct tree *t, struct time_info *ti,
 189                  struct uct_search_state *s)
 190 {
 191         /* Set up search state. */
 192         s->base_playouts = s->last_dynkomi = s->last_print = t->root->u.playouts;
 193         s->print_interval = TREE_SIMPROGRESS_INTERVAL * u->threads;
 194         s->fullmem = false;
 195
 196         if (ti) {
 197                 if (ti->period == TT_NULL) *ti = default_ti;
 198                 time_stop_conditions(ti, b, u->fuseki_end, u->yose_start, u->max_maintime_ratio, &s->stop);
 199         }
 200
 201         /* Fire up the tree search thread manager, which will in turn
 202          * spawn the searching threads. */
 203         assert(u->threads > 0);
 204         assert(!thread_manager_running);
 205         static struct uct_thread_ctx mctx;
 206         mctx = (struct uct_thread_ctx) { .u = u, .b = b, .color = color, .t = t, .seed = fast_random(65536), .ti = ti };
 207         s->ctx = &mctx;
 208         pthread_mutex_lock(&finish_serializer);
 209         pthread_mutex_lock(&finish_mutex);
 210         pthread_create(&thread_manager, NULL, spawn_thread_manager, s->ctx);
 211         thread_manager_running = true;
 212 }
 213
 214 struct uct_thread_ctx *
 215 uct_search_stop(void)
 216 {
 217         assert(thread_manager_running);
 218
 219         /* Signal thread manager to stop the workers. */
 220         pthread_mutex_lock(&finish_mutex);
 221         finish_thread = -1;
 222         pthread_cond_signal(&finish_cond);
 223         pthread_mutex_unlock(&finish_mutex);
 224
 225         /* Collect the thread manager. */
 226         struct uct_thread_ctx *pctx;
 227         thread_manager_running = false;
 228         pthread_join(thread_manager, (void **) &pctx);
 229         return pctx;
 230 }
 231
 232
 233 void
 234 uct_search_progress(struct uct *u, struct board *b, enum stone color,
 235                     struct tree *t, struct time_info *ti,
 236                     struct uct_search_state *s, int i)
 237 {
 238         struct uct_thread_ctx *ctx = s->ctx;
 239
 240         /* Adjust dynkomi? */
 241         int di = u->dynkomi_interval * u->threads;
 242         if (ctx->t->use_extra_komi && u->dynkomi->permove
 243             && !u->pondering && di
 244             && i > s->last_dynkomi + di) {
 245                 s->last_dynkomi += di;
 246                 floating_t old_dynkomi = ctx->t->extra_komi;
 247                 ctx->t->extra_komi = u->dynkomi->permove(u->dynkomi, b, ctx->t);
 248                 if (UDEBUGL(3) && old_dynkomi != ctx->t->extra_komi)
 249                         fprintf(stderr, "dynkomi adjusted (%f -> %f)\n",
 250                                 old_dynkomi, ctx->t->extra_komi);
 251         }
 252
 253         /* Print progress? */
 254         if (i - s->last_print > s->print_interval) {
 255                 s->last_print += s->print_interval; // keep the numbers tidy
 256                 uct_progress_status(u, ctx->t, color, s->last_print);
 257         }
 258
 259         if (!s->fullmem && ctx->t->nodes_size > u->max_tree_size) {
 260                 if (UDEBUGL(2))
 261                         fprintf(stderr, "memory limit hit (%lu > %lu)\n",
 262                                 ctx->t->nodes_size, u->max_tree_size);
 263                 s->fullmem = true;
 264         }
 265 }
 266
 267
 268 /* Determine whether we should terminate the search early. */
 269 static bool
 270 uct_search_stop_early(struct uct *u, struct tree *t, struct board *b,
 271                 struct time_info *ti, struct time_stop *stop,
 272                 struct tree_node *best, struct tree_node *best2,
 273                 int played, bool fullmem)
 274 {
 275         /* If the memory is full, stop immediately. Since the tree
 276          * cannot grow anymore, some non-well-expanded nodes will
 277          * quickly take over with extremely high ratio since the
 278          * counters are not properly simulated (just as if we use
 279          * non-UCT MonteCarlo). */
 280         /* (XXX: A proper solution would be to prune the tree
 281          * on the spot.) */
 282         if (fullmem)
 283                 return true;
 284
 285         /* Think at least 100ms to avoid a random move. This is particularly
 286          * important in distributed mode, where this function is called frequently. */
 287         double elapsed = 0.0;
 288         if (ti->dim == TD_WALLTIME) {
 289                 elapsed = time_now() - ti->len.t.timer_start;
 290                 if (elapsed < TREE_BUSYWAIT_INTERVAL) return false;
 291         }
 292
 293         /* Break early if we estimate the second-best move cannot
 294          * catch up in assigned time anymore. We use all our time
 295          * if we are in byoyomi with single stone remaining in our
 296          * period, however - it's better to pre-ponder. */
 297         bool time_indulgent = (!ti->len.t.main_time && ti->len.t.byoyomi_stones == 1);
 298         if (best2 && ti->dim == TD_WALLTIME
 299             && played >= PLAYOUT_EARLY_BREAK_MIN && !time_indulgent) {
 300                 double remaining = stop->worst.time - elapsed;
 301                 double pps = ((double)played) / elapsed;
 302                 double estplayouts = remaining * pps + PLAYOUT_DELTA_SAFEMARGIN;
 303                 if (best->u.playouts > best2->u.playouts + estplayouts) {
 304                         if (UDEBUGL(2))
 305                                 fprintf(stderr, "Early stop, result cannot change: "
 306                                         "best %d, best2 %d, estimated %f simulations to go (%d/%f=%f pps)\n",
 307                                         best->u.playouts, best2->u.playouts, estplayouts, played, elapsed, pps);
 308                         return true;
 309                 }
 310         }
 311
 312         /* Early break in won situation. */
 313         if (best->u.playouts >= PLAYOUT_EARLY_BREAK_MIN
 314             && (ti->dim != TD_WALLTIME || elapsed > TIME_EARLY_BREAK_MIN)
 315             && tree_node_get_value(t, 1, best->u.value) >= u->sure_win_threshold) {
 316                 return true;
 317         }
 318
 319         return false;
 320 }
 321
 322 /* Determine whether we should terminate the search later than expected. */
 323 static bool
 324 uct_search_keep_looking(struct uct *u, struct tree *t, struct board *b,
 325                 struct time_info *ti, struct time_stop *stop,
 326                 struct tree_node *best, struct tree_node *best2,
 327                 struct tree_node *bestr, struct tree_node *winner, int i)
 328 {
 329         if (!best) {
 330                 if (UDEBUGL(2))
 331                         fprintf(stderr, "Did not find best move, still trying...\n");
 332                 return true;
 333         }
 334
 335         /* Do not waste time if we are winning. Spend up to worst time if
 336          * we are unsure, but only desired time if we are sure of winning. */
 337         floating_t beta = 2 * (tree_node_get_value(t, 1, best->u.value) - 0.5);
 338         if (ti->dim == TD_WALLTIME && beta > 0) {
 339                 double good_enough = stop->desired.time * beta + stop->worst.time * (1 - beta);
 340                 double elapsed = time_now() - ti->len.t.timer_start;
 341                 if (elapsed > good_enough) return false;
 342         }
 343
 344         if (u->best2_ratio > 0) {
 345                 /* Check best/best2 simulations ratio. If the
 346                  * two best moves give very similar results,
 347                  * keep simulating. */
 348                 if (best2 && best2->u.playouts
 349                     && (double)best->u.playouts / best2->u.playouts < u->best2_ratio) {
 350                         if (UDEBUGL(2))
 351                                 fprintf(stderr, "Best2 ratio %f < threshold %f\n",
 352                                         (double)best->u.playouts / best2->u.playouts,
 353                                         u->best2_ratio);
 354                         return true;
 355                 }
 356         }
 357
 358         if (u->bestr_ratio > 0) {
 359                 /* Check best, best_best value difference. If the best move
 360                  * and its best child do not give similar enough results,
 361                  * keep simulating. */
 362                 if (bestr && bestr->u.playouts
 363                     && fabs((double)best->u.value - bestr->u.value) > u->bestr_ratio) {
 364                         if (UDEBUGL(2))
 365                                 fprintf(stderr, "Bestr delta %f > threshold %f\n",
 366                                         fabs((double)best->u.value - bestr->u.value),
 367                                         u->bestr_ratio);
 368                         return true;
 369                 }
 370         }
 371
 372         if (winner && winner != best) {
 373                 /* Keep simulating if best explored
 374                  * does not have also highest value. */
 375                 if (UDEBUGL(2))
 376                         fprintf(stderr, "[%d] best %3s [%d] %f != winner %3s [%d] %f\n", i,
 377                                 coord2sstr(node_coord(best), t->board),
 378                                 best->u.playouts, tree_node_get_value(t, 1, best->u.value),
 379                                 coord2sstr(node_coord(winner), t->board),
 380                                 winner->u.playouts, tree_node_get_value(t, 1, winner->u.value));
 381                 return true;
 382         }
 383
 384         /* No reason to keep simulating, bye. */
 385         return false;
 386 }
 387
 388 bool
 389 uct_search_check_stop(struct uct *u, struct board *b, enum stone color,
 390                       struct tree *t, struct time_info *ti,
 391                       struct uct_search_state *s, int i)
 392 {
 393         struct uct_thread_ctx *ctx = s->ctx;
 394
 395         /* Never consider stopping if we played too few simulations.
 396          * Maybe we risk losing on time when playing in super-extreme
 397          * time pressure but the tree is going to be just too messed
 398          * up otherwise - we might even play invalid suicides or pass
 399          * when we mustn't. */
 400         assert(!(ti->dim == TD_GAMES && ti->len.games < GJ_MINGAMES));
 401         if (i < GJ_MINGAMES)
 402                 return false;
 403
 404         struct tree_node *best = NULL;
 405         struct tree_node *best2 = NULL; // Second-best move.
 406         struct tree_node *bestr = NULL; // best's best child.
 407         struct tree_node *winner = NULL;
 408
 409         best = u->policy->choose(u->policy, ctx->t->root, b, color, resign);
 410         if (best) best2 = u->policy->choose(u->policy, ctx->t->root, b, color, node_coord(best));
 411
 412         /* Possibly stop search early if it's no use to try on. */
 413         int played = u->played_all + i - s->base_playouts;
 414         if (best && uct_search_stop_early(u, ctx->t, b, ti, &s->stop, best, best2, played, s->fullmem))
 415                 return true;
 416
 417         /* Check against time settings. */
 418         bool desired_done;
 419         if (ti->dim == TD_WALLTIME) {
 420                 double elapsed = time_now() - ti->len.t.timer_start;
 421                 if (elapsed > s->stop.worst.time) return true;
 422                 desired_done = elapsed > s->stop.desired.time;
 423
 424         } else { assert(ti->dim == TD_GAMES);
 425                 if (i > s->stop.worst.playouts) return true;
 426                 desired_done = i > s->stop.desired.playouts;
 427         }
 428
 429         /* We want to stop simulating, but are willing to keep trying
 430          * if we aren't completely sure about the winner yet. */
 431         if (desired_done) {
 432                 if (u->policy->winner && u->policy->evaluate) {
 433                         struct uct_descent descent = { .node = ctx->t->root };
 434                         u->policy->winner(u->policy, ctx->t, &descent);
 435                         winner = descent.node;
 436                 }
 437                 if (best)
 438                         bestr = u->policy->choose(u->policy, best, b, stone_other(color), resign);
 439                 if (!uct_search_keep_looking(u, ctx->t, b, ti, &s->stop, best, best2, bestr, winner, i))
 440                         return true;
 441         }
 442
 443         /* TODO: Early break if best->variance goes under threshold
 444          * and we already have enough playouts (possibly thanks to tbook
 445          * or to pondering)? */
 446         return false;
 447 }
 448
 449
 450 struct tree_node *
 451 uct_search_result(struct uct *u, struct board *b, enum stone color,
 452                   bool pass_all_alive, int played_games, int base_playouts,
 453                   coord_t *best_coord)
 454 {
 455         /* Choose the best move from the tree. */
 456         struct tree_node *best = u->policy->choose(u->policy, u->t->root, b, color, resign);
 457         if (!best) {
 458                 *best_coord = pass;
 459                 return NULL;
 460         }
 461         *best_coord = node_coord(best);
 462         if (UDEBUGL(1))
 463                 fprintf(stderr, "*** WINNER is %s (%d,%d) with score %1.4f (%d/%d:%d/%d games), extra komi %f\n",
 464                         coord2sstr(node_coord(best), b), coord_x(node_coord(best), b), coord_y(node_coord(best), b),
 465                         tree_node_get_value(u->t, 1, best->u.value), best->u.playouts,
 466                         u->t->root->u.playouts, u->t->root->u.playouts - base_playouts, played_games,
 467                         u->t->extra_komi);
 468
 469         /* Do not resign if we're so short of time that evaluation of best
 470          * move is completely unreliable, we might be winning actually.
 471          * In this case best is almost random but still better than resign.
 472          * Also do not resign if we are getting bad results while actually
 473          * giving away extra komi points (dynkomi). */
 474         if (tree_node_get_value(u->t, 1, best->u.value) < u->resign_threshold
 475             && !is_pass(node_coord(best)) && best->u.playouts > GJ_MINGAMES
 476             && (!u->t->use_extra_komi || komi_by_color(u->t->extra_komi, color) < 0.5)) {
 477                 *best_coord = resign;
 478                 return NULL;
 479         }
 480
 481         /* If the opponent just passed and we win counting, always
 482          * pass as well. */
 483         if (b->moves > 1 && is_pass(b->last_move.coord)) {
 484                 /* Make sure enough playouts are simulated. */
 485                 while (u->ownermap.playouts < GJ_MINGAMES)
 486                         uct_playout(u, b, color, u->t);
 487                 if (uct_pass_is_safe(u, b, color, u->pass_all_alive || pass_all_alive)) {
 488                         if (UDEBUGL(0))
 489                                 fprintf(stderr, "<Will rather pass, looks safe enough; score %f>\n",
 490                                         board_official_score(b, NULL) / 2);
 491                         *best_coord = pass;
 492                         best = u->t->root->children; // pass is the first child
 493                         assert(is_pass(node_coord(best)));
 494                         return best;
 495                 }
 496         }
 497
 498         return best;
 499 }