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 running worker thread. */
  74 __thread int thread_id = -1;
  75 /* ID of the thread manager. */
  76 static pthread_t thread_manager;
  77 bool thread_manager_running;
  78
  79 static pthread_mutex_t finish_mutex = PTHREAD_MUTEX_INITIALIZER;
  80 static pthread_cond_t finish_cond = PTHREAD_COND_INITIALIZER;
  81 static volatile int finish_thread;
  82 static pthread_mutex_t finish_serializer = PTHREAD_MUTEX_INITIALIZER;
  83
  84 static void *
  85 spawn_worker(void *ctx_)
  86 {
  87         struct uct_thread_ctx *ctx = ctx_;
  88         /* Setup */
  89         fast_srandom(ctx->seed);
  90         thread_id = ctx->tid;
  91         /* Run */
  92         ctx->games = uct_playouts(ctx->u, ctx->b, ctx->color, ctx->t);
  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->max_tree_size/2) {
 127                 unsigned long temp_size = (MIN_FREE_MEM_PERCENT * t->max_tree_size) / 100;
 128                 t->root = tree_garbage_collect(t, temp_size, t->root);
 129         }
 130
 131         /* Spawn threads... */
 132         for (int ti = 0; ti < u->threads; ti++) {
 133                 struct uct_thread_ctx *ctx = malloc2(sizeof(*ctx));
 134                 ctx->u = u; ctx->b = mctx->b; ctx->color = mctx->color;
 135                 mctx->t = ctx->t = t;
 136                 ctx->tid = ti; ctx->seed = fast_random(65536) + 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                         uct_halt = 1;
 149                         continue;
 150                 }
 151                 /* ...and gather its remnants. */
 152                 struct uct_thread_ctx *ctx;
 153                 pthread_join(threads[finish_thread], (void **) &ctx);
 154                 played_games += ctx->games;
 155                 joined++;
 156                 free(ctx);
 157                 if (UDEBUGL(3))
 158                         fprintf(stderr, "Joined worker %d\n", finish_thread);
 159                 pthread_mutex_unlock(&finish_serializer);
 160         }
 161
 162         pthread_mutex_unlock(&finish_mutex);
 163
 164         mctx->games = played_games;
 165         return mctx;
 166 }
 167
 168
 169 /*** THREAD MANAGER end */
 170
 171 /*** Search infrastructure: */
 172
 173
 174 int
 175 uct_search_games(struct uct_search_state *s)
 176 {
 177         return s->ctx->t->root->u.playouts;
 178 }
 179
 180 void
 181 uct_search_start(struct uct *u, struct board *b, enum stone color,
 182                  struct tree *t, struct time_info *ti,
 183                  struct uct_search_state *s)
 184 {
 185         /* Set up search state. */
 186         s->base_playouts = s->last_dynkomi = s->last_print = t->root->u.playouts;
 187         s->print_interval = TREE_SIMPROGRESS_INTERVAL * u->threads;
 188         s->fullmem = false;
 189
 190         if (ti) {
 191                 if (ti->period == TT_NULL) *ti = default_ti;
 192                 time_stop_conditions(ti, b, u->fuseki_end, u->yose_start, &s->stop);
 193         }
 194
 195         /* Fire up the tree search thread manager, which will in turn
 196          * spawn the searching threads. */
 197         assert(u->threads > 0);
 198         assert(!thread_manager_running);
 199         static struct uct_thread_ctx mctx;
 200         mctx = (struct uct_thread_ctx) { .u = u, .b = b, .color = color, .t = t, .seed = fast_random(65536) };
 201         s->ctx = &mctx;
 202         pthread_mutex_lock(&finish_mutex);
 203         pthread_create(&thread_manager, NULL, spawn_thread_manager, s->ctx);
 204         thread_manager_running = true;
 205 }
 206
 207 struct uct_thread_ctx *
 208 uct_search_stop(void)
 209 {
 210         assert(thread_manager_running);
 211
 212         /* Signal thread manager to stop the workers. */
 213         pthread_mutex_lock(&finish_mutex);
 214         finish_thread = -1;
 215         pthread_cond_signal(&finish_cond);
 216         pthread_mutex_unlock(&finish_mutex);
 217
 218         /* Collect the thread manager. */
 219         struct uct_thread_ctx *pctx;
 220         thread_manager_running = false;
 221         pthread_join(thread_manager, (void **) &pctx);
 222         return pctx;
 223 }
 224
 225
 226 void
 227 uct_search_progress(struct uct *u, struct board *b, enum stone color,
 228                     struct tree *t, struct time_info *ti,
 229                     struct uct_search_state *s, int i)
 230 {
 231         struct uct_thread_ctx *ctx = s->ctx;
 232
 233         /* Adjust dynkomi? */
 234         int di = u->dynkomi_interval * u->threads;
 235         if (ctx->t->use_extra_komi && u->dynkomi->permove
 236             && !u->pondering && di
 237             && i > s->last_dynkomi + di) {
 238                 s->last_dynkomi += di;
 239                 float old_dynkomi = ctx->t->extra_komi;
 240                 ctx->t->extra_komi = u->dynkomi->permove(u->dynkomi, b, ctx->t);
 241                 if (UDEBUGL(3) && old_dynkomi != ctx->t->extra_komi)
 242                         fprintf(stderr, "dynkomi adjusted (%f -> %f)\n",
 243                                 old_dynkomi, ctx->t->extra_komi);
 244         }
 245
 246         /* Print progress? */
 247         if (i - s->last_print > s->print_interval) {
 248                 s->last_print += s->print_interval; // keep the numbers tidy
 249                 uct_progress_status(u, ctx->t, color, s->last_print);
 250         }
 251
 252         if (!s->fullmem && ctx->t->nodes_size > u->max_tree_size) {
 253                 if (UDEBUGL(2))
 254                         fprintf(stderr, "memory limit hit (%lu > %lu)\n",
 255                                 ctx->t->nodes_size, u->max_tree_size);
 256                 s->fullmem = true;
 257         }
 258 }
 259
 260
 261 /* Determine whether we should terminate the search early. */
 262 static bool
 263 uct_search_stop_early(struct uct *u, struct tree *t, struct board *b,
 264                 struct time_info *ti, struct time_stop *stop,
 265                 struct tree_node *best, struct tree_node *best2,
 266                 int played, bool fullmem)
 267 {
 268         /* If the memory is full, stop immediately. Since the tree
 269          * cannot grow anymore, some non-well-expanded nodes will
 270          * quickly take over with extremely high ratio since the
 271          * counters are not properly simulated (just as if we use
 272          * non-UCT MonteCarlo). */
 273         /* (XXX: A proper solution would be to prune the tree
 274          * on the spot.) */
 275         if (fullmem)
 276                 return true;
 277
 278         /* Break early if we estimate the second-best move cannot
 279          * catch up in assigned time anymore. We use all our time
 280          * if we are in byoyomi with single stone remaining in our
 281          * period, however - it's better to pre-ponder. */
 282         bool time_indulgent = (!ti->len.t.main_time && ti->len.t.byoyomi_stones == 1);
 283         if (best2 && ti->dim == TD_WALLTIME && !time_indulgent) {
 284                 double elapsed = time_now() - ti->len.t.timer_start;
 285                 double remaining = stop->worst.time - elapsed;
 286                 double pps = ((double)played) / elapsed;
 287                 double estplayouts = remaining * pps + PLAYOUT_DELTA_SAFEMARGIN;
 288                 if (best->u.playouts > best2->u.playouts + estplayouts) {
 289                         if (UDEBUGL(2))
 290                                 fprintf(stderr, "Early stop, result cannot change: "
 291                                         "best %d, best2 %d, estimated %f simulations to go\n",
 292                                         best->u.playouts, best2->u.playouts, estplayouts);
 293                         return true;
 294                 }
 295         }
 296
 297         /* Early break in won situation. */
 298         if (best->u.playouts >= PLAYOUT_EARLY_BREAK_MIN
 299             && tree_node_get_value(t, 1, best->u.value) >= u->loss_threshold) {
 300                 return true;
 301         }
 302
 303         return false;
 304 }
 305
 306 /* Determine whether we should terminate the search later than expected. */
 307 static bool
 308 uct_search_keep_looking(struct uct *u, struct tree *t, struct board *b,
 309                 struct time_info *ti, struct time_stop *stop,
 310                 struct tree_node *best, struct tree_node *best2,
 311                 struct tree_node *bestr, struct tree_node *winner, int i)
 312 {
 313         if (!best) {
 314                 if (UDEBUGL(2))
 315                         fprintf(stderr, "Did not find best move, still trying...\n");
 316                 return true;
 317         }
 318
 319         /* Do not waste time if we are winning. Spend up to worst time if
 320          * we are unsure, but only desired time if we are sure of winning. */
 321         float beta = 2 * (tree_node_get_value(t, 1, best->u.value) - 0.5);
 322         if (ti->dim == TD_WALLTIME && beta > 0) {
 323                 double good_enough = stop->desired.time * beta + stop->worst.time * (1 - beta);
 324                 double elapsed = time_now() - ti->len.t.timer_start;
 325                 if (elapsed > good_enough) return false;
 326         }
 327
 328         if (u->best2_ratio > 0) {
 329                 /* Check best/best2 simulations ratio. If the
 330                  * two best moves give very similar results,
 331                  * keep simulating. */
 332                 if (best2 && best2->u.playouts
 333                     && (double)best->u.playouts / best2->u.playouts < u->best2_ratio) {
 334                         if (UDEBUGL(2))
 335                                 fprintf(stderr, "Best2 ratio %f < threshold %f\n",
 336                                         (double)best->u.playouts / best2->u.playouts,
 337                                         u->best2_ratio);
 338                         return true;
 339                 }
 340         }
 341
 342         if (u->bestr_ratio > 0) {
 343                 /* Check best, best_best value difference. If the best move
 344                  * and its best child do not give similar enough results,
 345                  * keep simulating. */
 346                 if (bestr && bestr->u.playouts
 347                     && fabs((double)best->u.value - bestr->u.value) > u->bestr_ratio) {
 348                         if (UDEBUGL(2))
 349                                 fprintf(stderr, "Bestr delta %f > threshold %f\n",
 350                                         fabs((double)best->u.value - bestr->u.value),
 351                                         u->bestr_ratio);
 352                         return true;
 353                 }
 354         }
 355
 356         if (winner && winner != best) {
 357                 /* Keep simulating if best explored
 358                  * does not have also highest value. */
 359                 if (UDEBUGL(2))
 360                         fprintf(stderr, "[%d] best %3s [%d] %f != winner %3s [%d] %f\n", i,
 361                                 coord2sstr(best->coord, t->board),
 362                                 best->u.playouts, tree_node_get_value(t, 1, best->u.value),
 363                                 coord2sstr(winner->coord, t->board),
 364                                 winner->u.playouts, tree_node_get_value(t, 1, winner->u.value));
 365                 return true;
 366         }
 367
 368         /* No reason to keep simulating, bye. */
 369         return false;
 370 }
 371
 372 bool
 373 uct_search_check_stop(struct uct *u, struct board *b, enum stone color,
 374                       struct tree *t, struct time_info *ti,
 375                       struct uct_search_state *s, int i)
 376 {
 377         struct uct_thread_ctx *ctx = s->ctx;
 378
 379         /* Never consider stopping if we played too few simulations.
 380          * Maybe we risk losing on time when playing in super-extreme
 381          * time pressure but the tree is going to be just too messed
 382          * up otherwise - we might even play invalid suicides or pass
 383          * when we mustn't. */
 384         if (i < GJ_MINGAMES)
 385                 return false;
 386
 387         struct tree_node *best = NULL;
 388         struct tree_node *best2 = NULL; // Second-best move.
 389         struct tree_node *bestr = NULL; // best's best child.
 390         struct tree_node *winner = NULL;
 391
 392         best = u->policy->choose(u->policy, ctx->t->root, b, color, resign);
 393         if (best) best2 = u->policy->choose(u->policy, ctx->t->root, b, color, best->coord);
 394
 395         /* Possibly stop search early if it's no use to try on. */
 396         int played = u->played_all + i - s->base_playouts;
 397         if (best && uct_search_stop_early(u, ctx->t, b, ti, &s->stop, best, best2, played, s->fullmem))
 398                 return true;
 399
 400         /* Check against time settings. */
 401         bool desired_done;
 402         if (ti->dim == TD_WALLTIME) {
 403                 double elapsed = time_now() - ti->len.t.timer_start;
 404                 if (elapsed > s->stop.worst.time) return true;
 405                 desired_done = elapsed > s->stop.desired.time;
 406
 407         } else { assert(ti->dim == TD_GAMES);
 408                 if (i > s->stop.worst.playouts) return true;
 409                 desired_done = i > s->stop.desired.playouts;
 410         }
 411
 412         /* We want to stop simulating, but are willing to keep trying
 413          * if we aren't completely sure about the winner yet. */
 414         if (desired_done) {
 415                 if (u->policy->winner && u->policy->evaluate) {
 416                         struct uct_descent descent = { .node = ctx->t->root };
 417                         u->policy->winner(u->policy, ctx->t, &descent);
 418                         winner = descent.node;
 419                 }
 420                 if (best)
 421                         bestr = u->policy->choose(u->policy, best, b, stone_other(color), resign);
 422                 if (!uct_search_keep_looking(u, ctx->t, b, ti, &s->stop, best, best2, bestr, winner, i))
 423                         return true;
 424         }
 425
 426         /* TODO: Early break if best->variance goes under threshold
 427          * and we already have enough playouts (possibly thanks to book
 428          * or to pondering)? */
 429         return false;
 430 }
 431
 432
 433 struct tree_node *
 434 uct_search_result(struct uct *u, struct board *b, enum stone color,
 435                   bool pass_all_alive, int played_games, int base_playouts,
 436                   coord_t *best_coord)
 437 {
 438         /* Choose the best move from the tree. */
 439         struct tree_node *best = u->policy->choose(u->policy, u->t->root, b, color, resign);
 440         if (!best) {
 441                 *best_coord = pass;
 442                 return NULL;
 443         }
 444         *best_coord = best->coord;
 445         if (UDEBUGL(1))
 446                 fprintf(stderr, "*** WINNER is %s (%d,%d) with score %1.4f (%d/%d:%d/%d games), extra komi %f\n",
 447                         coord2sstr(best->coord, b), coord_x(best->coord, b), coord_y(best->coord, b),
 448                         tree_node_get_value(u->t, 1, best->u.value), best->u.playouts,
 449                         u->t->root->u.playouts, u->t->root->u.playouts - base_playouts, played_games,
 450                         u->t->extra_komi);
 451
 452         /* Do not resign if we're so short of time that evaluation of best
 453          * move is completely unreliable, we might be winning actually.
 454          * In this case best is almost random but still better than resign.
 455          * Also do not resign if we are getting bad results while actually
 456          * giving away extra komi points (dynkomi). */
 457         if (tree_node_get_value(u->t, 1, best->u.value) < u->resign_ratio
 458             && !is_pass(best->coord) && best->u.playouts > GJ_MINGAMES
 459             && komi_by_color(u->t->extra_komi, color) < 0.5) {
 460                 *best_coord = resign;
 461                 return NULL;
 462         }
 463
 464         /* If the opponent just passed and we win counting, always
 465          * pass as well. */
 466         if (b->moves > 1 && is_pass(b->last_move.coord)) {
 467                 /* Make sure enough playouts are simulated. */
 468                 while (u->ownermap.playouts < GJ_MINGAMES)
 469                         uct_playout(u, b, color, u->t);
 470                 if (uct_pass_is_safe(u, b, color, u->pass_all_alive || pass_all_alive)) {
 471                         if (UDEBUGL(0))
 472                                 fprintf(stderr, "<Will rather pass, looks safe enough; score %f>\n",
 473                                         board_official_score(b, NULL) / 2);
 474                         *best_coord = pass;
 475                         best = u->t->root->children; // pass is the first child
 476                         assert(is_pass(best->coord));
 477                         return best;
 478                 }
 479         }
 480
 481         return best;
 482 }