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