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