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->max_tree_size/2) {
 124                 unsigned long temp_size = (MIN_FREE_MEM_PERCENT * t->max_tree_size) / 100;
 125                 t->root = tree_garbage_collect(t, temp_size, t->root);
 126         }
 127
 128         /* Spawn threads... */
 129         for (int ti = 0; ti < u->threads; ti++) {
 130                 struct uct_thread_ctx *ctx = malloc2(sizeof(*ctx));
 131                 ctx->u = u; ctx->b = mctx->b; ctx->color = mctx->color;
 132                 mctx->t = ctx->t = t;
 133                 ctx->tid = ti; ctx->seed = fast_random(65536) + 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                         uct_halt = 1;
 146                         continue;
 147                 }
 148                 /* ...and gather its remnants. */
 149                 struct uct_thread_ctx *ctx;
 150                 pthread_join(threads[finish_thread], (void **) &ctx);
 151                 played_games += ctx->games;
 152                 joined++;
 153                 free(ctx);
 154                 if (UDEBUGL(3))
 155                         fprintf(stderr, "Joined worker %d\n", finish_thread);
 156                 pthread_mutex_unlock(&finish_serializer);
 157         }
 158
 159         pthread_mutex_unlock(&finish_mutex);
 160
 161         mctx->games = played_games;
 162         return mctx;
 163 }
 164
 165
 166 /*** THREAD MANAGER end */
 167
 168 /*** Search infrastructure: */
 169
 170
 171 int
 172 uct_search_games(struct uct_search_state *s)
 173 {
 174         return s->ctx->t->root->u.playouts;
 175 }
 176
 177 void
 178 uct_search_start(struct uct *u, struct board *b, enum stone color,
 179                  struct tree *t, struct time_info *ti,
 180                  struct uct_search_state *s)
 181 {
 182         /* Set up search state. */
 183         s->base_playouts = s->last_dynkomi = s->last_print = t->root->u.playouts;
 184         s->print_interval = TREE_SIMPROGRESS_INTERVAL * u->threads;
 185         s->fullmem = false;
 186
 187         if (ti) {
 188                 if (ti->period == TT_NULL) *ti = default_ti;
 189                 time_stop_conditions(ti, b, u->fuseki_end, u->yose_start, &s->stop);
 190         }
 191
 192         /* Fire up the tree search thread manager, which will in turn
 193          * spawn the searching threads. */
 194         assert(u->threads > 0);
 195         assert(!thread_manager_running);
 196         static struct uct_thread_ctx mctx;
 197         mctx = (struct uct_thread_ctx) { .u = u, .b = b, .color = color, .t = t, .seed = fast_random(65536) };
 198         s->ctx = &mctx;
 199         pthread_mutex_lock(&finish_mutex);
 200         pthread_create(&thread_manager, NULL, spawn_thread_manager, s->ctx);
 201         thread_manager_running = true;
 202 }
 203
 204 struct uct_thread_ctx *
 205 uct_search_stop(void)
 206 {
 207         assert(thread_manager_running);
 208
 209         /* Signal thread manager to stop the workers. */
 210         pthread_mutex_lock(&finish_mutex);
 211         finish_thread = -1;
 212         pthread_cond_signal(&finish_cond);
 213         pthread_mutex_unlock(&finish_mutex);
 214
 215         /* Collect the thread manager. */
 216         struct uct_thread_ctx *pctx;
 217         thread_manager_running = false;
 218         pthread_join(thread_manager, (void **) &pctx);
 219         return pctx;
 220 }
 221
 222
 223 void
 224 uct_search_progress(struct uct *u, struct board *b, enum stone color,
 225                     struct tree *t, struct time_info *ti,
 226                     struct uct_search_state *s, int i)
 227 {
 228         struct uct_thread_ctx *ctx = s->ctx;
 229
 230         /* Adjust dynkomi? */
 231         int di = u->dynkomi_interval * u->threads;
 232         if (ctx->t->use_extra_komi && u->dynkomi->permove
 233             && !u->pondering && di
 234             && i > s->last_dynkomi + di) {
 235                 s->last_dynkomi += di;
 236                 float old_dynkomi = ctx->t->extra_komi;
 237                 ctx->t->extra_komi = u->dynkomi->permove(u->dynkomi, b, ctx->t);
 238                 if (UDEBUGL(3) && old_dynkomi != ctx->t->extra_komi)
 239                         fprintf(stderr, "dynkomi adjusted (%f -> %f)\n",
 240                                 old_dynkomi, ctx->t->extra_komi);
 241         }
 242
 243         /* Print progress? */
 244         if (i - s->last_print > s->print_interval) {
 245                 s->last_print += s->print_interval; // keep the numbers tidy
 246                 uct_progress_status(u, ctx->t, color, s->last_print);
 247         }
 248
 249         if (!s->fullmem && ctx->t->nodes_size > u->max_tree_size) {
 250                 if (UDEBUGL(2))
 251                         fprintf(stderr, "memory limit hit (%lu > %lu)\n",
 252                                 ctx->t->nodes_size, u->max_tree_size);
 253                 s->fullmem = true;
 254         }
 255 }
 256
 257
 258 /* Determine whether we should terminate the search early. */
 259 static bool
 260 uct_search_stop_early(struct uct *u, struct tree *t, struct board *b,
 261                 struct time_info *ti, struct time_stop *stop,
 262                 struct tree_node *best, struct tree_node *best2,
 263                 int played, bool fullmem)
 264 {
 265         /* If the memory is full, stop immediately. Since the tree
 266          * cannot grow anymore, some non-well-expanded nodes will
 267          * quickly take over with extremely high ratio since the
 268          * counters are not properly simulated (just as if we use
 269          * non-UCT MonteCarlo). */
 270         /* (XXX: A proper solution would be to prune the tree
 271          * on the spot.) */
 272         if (fullmem)
 273                 return true;
 274
 275         /* Break early if we estimate the second-best move cannot
 276          * catch up in assigned time anymore. We use all our time
 277          * if we are in byoyomi with single stone remaining in our
 278          * period, however - it's better to pre-ponder. */
 279         bool time_indulgent = (!ti->len.t.main_time && ti->len.t.byoyomi_stones == 1);
 280         if (best2 && ti->dim == TD_WALLTIME && !time_indulgent) {
 281                 double elapsed = time_now() - ti->len.t.timer_start;
 282                 double remaining = stop->worst.time - elapsed;
 283                 double pps = ((double)played) / elapsed;
 284                 double estplayouts = remaining * pps + PLAYOUT_DELTA_SAFEMARGIN;
 285                 if (best->u.playouts > best2->u.playouts + estplayouts) {
 286                         if (UDEBUGL(2))
 287                                 fprintf(stderr, "Early stop, result cannot change: "
 288                                         "best %d, best2 %d, estimated %f simulations to go\n",
 289                                         best->u.playouts, best2->u.playouts, estplayouts);
 290                         return true;
 291                 }
 292         }
 293
 294         /* Early break in won situation. */
 295         if (best->u.playouts >= PLAYOUT_EARLY_BREAK_MIN
 296             && tree_node_get_value(t, 1, best->u.value) >= u->sure_win_threshold) {
 297                 return true;
 298         }
 299
 300         return false;
 301 }
 302
 303 /* Determine whether we should terminate the search later than expected. */
 304 static bool
 305 uct_search_keep_looking(struct uct *u, struct tree *t, struct board *b,
 306                 struct time_info *ti, struct time_stop *stop,
 307                 struct tree_node *best, struct tree_node *best2,
 308                 struct tree_node *bestr, struct tree_node *winner, int i)
 309 {
 310         if (!best) {
 311                 if (UDEBUGL(2))
 312                         fprintf(stderr, "Did not find best move, still trying...\n");
 313                 return true;
 314         }
 315
 316         /* Do not waste time if we are winning. Spend up to worst time if
 317          * we are unsure, but only desired time if we are sure of winning. */
 318         float beta = 2 * (tree_node_get_value(t, 1, best->u.value) - 0.5);
 319         if (ti->dim == TD_WALLTIME && beta > 0) {
 320                 double good_enough = stop->desired.time * beta + stop->worst.time * (1 - beta);
 321                 double elapsed = time_now() - ti->len.t.timer_start;
 322                 if (elapsed > good_enough) return false;
 323         }
 324
 325         if (u->best2_ratio > 0) {
 326                 /* Check best/best2 simulations ratio. If the
 327                  * two best moves give very similar results,
 328                  * keep simulating. */
 329                 if (best2 && best2->u.playouts
 330                     && (double)best->u.playouts / best2->u.playouts < u->best2_ratio) {
 331                         if (UDEBUGL(2))
 332                                 fprintf(stderr, "Best2 ratio %f < threshold %f\n",
 333                                         (double)best->u.playouts / best2->u.playouts,
 334                                         u->best2_ratio);
 335                         return true;
 336                 }
 337         }
 338
 339         if (u->bestr_ratio > 0) {
 340                 /* Check best, best_best value difference. If the best move
 341                  * and its best child do not give similar enough results,
 342                  * keep simulating. */
 343                 if (bestr && bestr->u.playouts
 344                     && fabs((double)best->u.value - bestr->u.value) > u->bestr_ratio) {
 345                         if (UDEBUGL(2))
 346                                 fprintf(stderr, "Bestr delta %f > threshold %f\n",
 347                                         fabs((double)best->u.value - bestr->u.value),
 348                                         u->bestr_ratio);
 349                         return true;
 350                 }
 351         }
 352
 353         if (winner && winner != best) {
 354                 /* Keep simulating if best explored
 355                  * does not have also highest value. */
 356                 if (UDEBUGL(2))
 357                         fprintf(stderr, "[%d] best %3s [%d] %f != winner %3s [%d] %f\n", i,
 358                                 coord2sstr(best->coord, t->board),
 359                                 best->u.playouts, tree_node_get_value(t, 1, best->u.value),
 360                                 coord2sstr(winner->coord, t->board),
 361                                 winner->u.playouts, tree_node_get_value(t, 1, winner->u.value));
 362                 return true;
 363         }
 364
 365         /* No reason to keep simulating, bye. */
 366         return false;
 367 }
 368
 369 bool
 370 uct_search_check_stop(struct uct *u, struct board *b, enum stone color,
 371                       struct tree *t, struct time_info *ti,
 372                       struct uct_search_state *s, int i)
 373 {
 374         struct uct_thread_ctx *ctx = s->ctx;
 375
 376         /* Never consider stopping if we played too few simulations.
 377          * Maybe we risk losing on time when playing in super-extreme
 378          * time pressure but the tree is going to be just too messed
 379          * up otherwise - we might even play invalid suicides or pass
 380          * when we mustn't. */
 381         if (i < GJ_MINGAMES)
 382                 return false;
 383
 384         struct tree_node *best = NULL;
 385         struct tree_node *best2 = NULL; // Second-best move.
 386         struct tree_node *bestr = NULL; // best's best child.
 387         struct tree_node *winner = NULL;
 388
 389         best = u->policy->choose(u->policy, ctx->t->root, b, color, resign);
 390         if (best) best2 = u->policy->choose(u->policy, ctx->t->root, b, color, best->coord);
 391
 392         /* Possibly stop search early if it's no use to try on. */
 393         int played = u->played_all + i - s->base_playouts;
 394         if (best && uct_search_stop_early(u, ctx->t, b, ti, &s->stop, best, best2, played, s->fullmem))
 395                 return true;
 396
 397         /* Check against time settings. */
 398         bool desired_done;
 399         if (ti->dim == TD_WALLTIME) {
 400                 double elapsed = time_now() - ti->len.t.timer_start;
 401                 if (elapsed > s->stop.worst.time) return true;
 402                 desired_done = elapsed > s->stop.desired.time;
 403
 404         } else { assert(ti->dim == TD_GAMES);
 405                 if (i > s->stop.worst.playouts) return true;
 406                 desired_done = i > s->stop.desired.playouts;
 407         }
 408
 409         /* We want to stop simulating, but are willing to keep trying
 410          * if we aren't completely sure about the winner yet. */
 411         if (desired_done) {
 412                 if (u->policy->winner && u->policy->evaluate) {
 413                         struct uct_descent descent = { .node = ctx->t->root };
 414                         u->policy->winner(u->policy, ctx->t, &descent);
 415                         winner = descent.node;
 416                 }
 417                 if (best)
 418                         bestr = u->policy->choose(u->policy, best, b, stone_other(color), resign);
 419                 if (!uct_search_keep_looking(u, ctx->t, b, ti, &s->stop, best, best2, bestr, winner, i))
 420                         return true;
 421         }
 422
 423         /* TODO: Early break if best->variance goes under threshold
 424          * and we already have enough playouts (possibly thanks to book
 425          * or to pondering)? */
 426         return false;
 427 }
 428
 429
 430 struct tree_node *
 431 uct_search_result(struct uct *u, struct board *b, enum stone color,
 432                   bool pass_all_alive, int played_games, int base_playouts,
 433                   coord_t *best_coord)
 434 {
 435         /* Choose the best move from the tree. */
 436         struct tree_node *best = u->policy->choose(u->policy, u->t->root, b, color, resign);
 437         if (!best) {
 438                 *best_coord = pass;
 439                 return NULL;
 440         }
 441         *best_coord = best->coord;
 442         if (UDEBUGL(1))
 443                 fprintf(stderr, "*** WINNER is %s (%d,%d) with score %1.4f (%d/%d:%d/%d games), extra komi %f\n",
 444                         coord2sstr(best->coord, b), coord_x(best->coord, b), coord_y(best->coord, b),
 445                         tree_node_get_value(u->t, 1, best->u.value), best->u.playouts,
 446                         u->t->root->u.playouts, u->t->root->u.playouts - base_playouts, played_games,
 447                         u->t->extra_komi);
 448
 449         /* Do not resign if we're so short of time that evaluation of best
 450          * move is completely unreliable, we might be winning actually.
 451          * In this case best is almost random but still better than resign.
 452          * Also do not resign if we are getting bad results while actually
 453          * giving away extra komi points (dynkomi). */
 454         if (tree_node_get_value(u->t, 1, best->u.value) < u->resign_threshold
 455             && !is_pass(best->coord) && best->u.playouts > GJ_MINGAMES
 456             && (!u->t->use_extra_komi || komi_by_color(u->t->extra_komi, color) < 0.5)) {
 457                 *best_coord = resign;
 458                 return NULL;
 459         }
 460
 461         /* If the opponent just passed and we win counting, always
 462          * pass as well. */
 463         if (b->moves > 1 && is_pass(b->last_move.coord)) {
 464                 /* Make sure enough playouts are simulated. */
 465                 while (u->ownermap.playouts < GJ_MINGAMES)
 466                         uct_playout(u, b, color, u->t);
 467                 if (uct_pass_is_safe(u, b, color, u->pass_all_alive || pass_all_alive)) {
 468                         if (UDEBUGL(0))
 469                                 fprintf(stderr, "<Will rather pass, looks safe enough; score %f>\n",
 470                                         board_official_score(b, NULL) / 2);
 471                         *best_coord = pass;
 472                         best = u->t->root->children; // pass is the first child
 473                         assert(is_pass(best->coord));
 474                         return best;
 475                 }
 476         }
 477
 478         return best;
 479 }