uct/uct.c

   1 #include <assert.h>
   2 #include <pthread.h>
   3 #include <signal.h>
   4 #include <stdio.h>
   5 #include <stdlib.h>
   6 #include <string.h>
   7
   8 #define DEBUG
   9
  10 #include "debug.h"
  11 #include "board.h"
  12 #include "move.h"
  13 #include "playout.h"
  14 #include "playout/moggy.h"
  15 #include "playout/old.h"
  16 #include "playout/light.h"
  17 #include "random.h"
  18 #include "uct/internal.h"
  19 #include "uct/tree.h"
  20 #include "uct/uct.h"
  21
  22 struct uct_policy *policy_ucb1_init(struct uct *u, char *arg);
  23 struct uct_policy *policy_ucb1tuned_init(struct uct *u, char *arg);
  24 struct uct_policy *policy_ucb1amaf_init(struct uct *u, char *arg);
  25
  26
  27 #define MC_GAMES        80000
  28 #define MC_GAMELEN      400
  29
  30
  31 static void
  32 progress_status(struct uct *u, struct tree *t, enum stone color, int playouts)
  33 {
  34         if (!UDEBUGL(0))
  35                 return;
  36
  37         /* Best move */
  38         struct tree_node *best = u->policy->choose(u->policy, t->root, t->board, color);
  39         if (!best) {
  40                 fprintf(stderr, "... No moves left\n");
  41                 return;
  42         }
  43         fprintf(stderr, "[%d] ", playouts);
  44         fprintf(stderr, "best %f ", best->u.value);
  45
  46         /* Max depth */
  47         fprintf(stderr, "deepest % 2d ", t->max_depth - t->root->depth);
  48
  49         /* Best sequence */
  50         fprintf(stderr, "| seq ");
  51         for (int depth = 0; depth < 6; depth++) {
  52                 if (best && best->u.playouts >= 25) {
  53                         fprintf(stderr, "%3s ", coord2sstr(best->coord, t->board));
  54                         best = u->policy->choose(u->policy, best, t->board, color);
  55                 } else {
  56                         fprintf(stderr, "    ");
  57                 }
  58         }
  59
  60         /* Best candidates */
  61         fprintf(stderr, "| can ");
  62         int cans = 4;
  63         struct tree_node *can[cans];
  64         memset(can, 0, sizeof(can));
  65         best = t->root->children;
  66         while (best) {
  67                 int c = 0;
  68                 while ((!can[c] || best->u.playouts > can[c]->u.playouts) && ++c < cans);
  69                 for (int d = 0; d < c; d++) can[d] = can[d + 1];
  70                 if (c > 0) can[c - 1] = best;
  71                 best = best->sibling;
  72         }
  73         while (--cans >= 0) {
  74                 if (can[cans]) {
  75                         fprintf(stderr, "%3s(%.3f) ", coord2sstr(can[cans]->coord, t->board), can[cans]->u.value);
  76                 } else {
  77                         fprintf(stderr, "           ");
  78                 }
  79         }
  80
  81         fprintf(stderr, "\n");
  82 }
  83
  84
  85 static int
  86 uct_leaf_node(struct uct *u, struct board *b, enum stone player_color,
  87               struct playout_amafmap *amaf,
  88               struct tree *t, struct tree_node *n, enum stone node_color,
  89               char *spaces)
  90 {
  91         enum stone next_color = stone_other(node_color);
  92         if (n->u.playouts >= u->expand_p)
  93                 tree_expand_node(t, n, b, next_color, u->radar_d, u->policy,
  94                                  (next_color == player_color ? 1 : -1));
  95         if (UDEBUGL(7))
  96                 fprintf(stderr, "%s*-- UCT playout #%d start [%s] %f\n",
  97                         spaces, n->u.playouts, coord2sstr(n->coord, t->board), n->u.value);
  98
  99         int result = play_random_game(&b2, next_color, u->gamelen, u->playout_amaf ? amaf : NULL, u->playout);
 100         if (player_color != next_color && result >= 0)
 101                 result = !result;
 102         if (UDEBUGL(7))
 103                 fprintf(stderr, "%s -- [%d..%d] %s random playout result %d\n",
 104                         spaces, player_color, next_color, coord2sstr(n->coord, t->board), result);
 105
 106         return result;
 107 }
 108
 109 static int
 110 uct_playout(struct uct *u, struct board *b, enum stone player_color, struct tree *t)
 111 {
 112         struct board b2;
 113         board_copy(&b2, b);
 114
 115         struct playout_amafmap *amaf = NULL;
 116         if (u->policy->wants_amaf) {
 117                 amaf = calloc(1, sizeof(*amaf));
 118                 amaf->map = calloc(board_size2(&b2) + 1, sizeof(*amaf->map));
 119                 amaf->map++; // -1 is pass
 120         }
 121
 122         /* Walk the tree until we find a leaf, then expand it and do
 123          * a random playout. */
 124         struct tree_node *n = t->root;
 125         enum stone node_color = stone_other(player_color);
 126         int result;
 127         int pass_limit = (board_size(&b2) - 2) * (board_size(&b2) - 2) / 2;
 128         int passes = is_pass(b->last_move.coord);
 129         /* debug */
 130         int depth = 0;
 131         static char spaces[] = "\0                                                      ";
 132         /* /debug */
 133         if (UDEBUGL(8))
 134                 fprintf(stderr, "--- UCT walk with color %d\n", player_color);
 135
 136         while (!tree_leaf_node(n) && passes < 2) {
 137                 spaces[depth++] = ' '; spaces[depth] = 0;
 138
 139                 n = u->policy->descend(u->policy, t, n, (node_color == player_color ? -1 : 1), pass_limit);
 140                 node_color = stone_other(node_color);
 141                 assert(n == t->root || n->parent);
 142                 if (UDEBUGL(7))
 143                         fprintf(stderr, "%s+-- UCT sent us to [%s:%d] %f\n",
 144                                 spaces, coord2sstr(n->coord, t->board), n->coord, n->u.value);
 145
 146                 if (amaf && n->coord >= -1 && !is_pass(n->coord)) {
 147                         if (amaf->map[n->coord] == S_NONE) {
 148                                 amaf->map[n->coord] = node_color;
 149                         } else {
 150                                 amaf_op(amaf->map[n->coord], +);
 151                         }
 152                 }
 153
 154                 struct move m = { n->coord, node_color };
 155                 int res = board_play(&b2, &m);
 156
 157                 if (res < 0 || (!is_pass(m.coord) && !group_at(&b2, m.coord)) /* suicide */
 158                     || b2.superko_violation) {
 159                         if (UDEBUGL(3)) {
 160                                 for (struct tree_node *ni = n; ni; ni = ni->parent)
 161                                         fprintf(stderr, "%s ", coord2sstr(ni->coord, t->board));
 162                                 fprintf(stderr, "deleting invalid %s node %d,%d res %d group %d spk %d\n",
 163                                         stone2str(node_color), coord_x(n->coord,b), coord_y(n->coord,b),
 164                                         res, group_at(&b2, m.coord), b2.superko_violation);
 165                         }
 166                         tree_delete_node(t, n);
 167                         result = -1;
 168                         goto end;
 169                 }
 170
 171                 if (is_pass(n->coord))
 172                         passes++;
 173                 else
 174                         passes = 0;
 175         }
 176
 177         if (tree_leaf_node(n)) {
 178                 result = uct_leaf_node(u, &b2, player_color, amaf, t, n, node_color, spaces);
 179
 180         } else { assert(passes >= 2);
 181
 182                 float score = board_official_score(&b2);
 183                 result = (player_color == S_BLACK) ? score < 0 : score > 0;
 184
 185                 if (UDEBUGL(5))
 186                         fprintf(stderr, "[%d..%d] %s p-p scoring playout result %d (W %f)\n",
 187                                 player_color, node_color, coord2sstr(n->coord, t->board), result, score);
 188                 if (UDEBUGL(6))
 189                         board_print(&b2, stderr);
 190         }
 191
 192         assert(n == t->root || n->parent);
 193         if (result >= 0)
 194                 u->policy->update(u->policy, t, n, node_color, player_color, amaf, result);
 195
 196 end:
 197         if (amaf) {
 198                 free(amaf->map - 1);
 199                 free(amaf);
 200         }
 201         board_done_noalloc(&b2);
 202         return result;
 203 }
 204
 205 static void
 206 prepare_move(struct engine *e, struct board *b, enum stone color, coord_t promote)
 207 {
 208         struct uct *u = e->data;
 209
 210         if (!b->moves && u->t) {
 211                 /* Stale state from last game */
 212                 tree_done(u->t);
 213                 u->t = NULL;
 214         }
 215
 216         if (!u->t) {
 217                 u->t = tree_init(b, color);
 218                 if (u->force_seed)
 219                         fast_srandom(u->force_seed);
 220                 if (UDEBUGL(0))
 221                         fprintf(stderr, "Fresh board with random seed %lu\n", fast_getseed());
 222                 //board_print(b, stderr);
 223                 tree_load(u->t, b, color);
 224         }
 225
 226         /* XXX: We hope that the opponent didn't suddenly play
 227          * several moves in the row. */
 228         if (!is_resign(promote) && !tree_promote_at(u->t, b, promote)) {
 229                 if (UDEBUGL(2))
 230                         fprintf(stderr, "<cannot find node to promote>\n");
 231                 /* Reset tree */
 232                 tree_done(u->t);
 233                 u->t = tree_init(b, color);
 234         }
 235 }
 236
 237 /* Set in main thread in case the playouts should stop. */
 238 static volatile sig_atomic_t halt = 0;
 239
 240 static int
 241 uct_playouts(struct uct *u, struct board *b, enum stone color, struct tree *t)
 242 {
 243         int i, games = u->games;
 244         if (t->root->children)
 245                 games -= t->root->u.playouts / 1.5;
 246         /* else this is highly read-out but dead-end branch of opening book;
 247          * we need to start from scratch; XXX: Maybe actually base the readout
 248          * count based on number of playouts of best node? */
 249         for (i = 0; i < games; i++) {
 250                 int result = uct_playout(u, b, color, t);
 251                 if (result < 0) {
 252                         /* Tree descent has hit invalid move. */
 253                         continue;
 254                 }
 255
 256                 if (i > 0 && !(i % 10000)) {
 257                         progress_status(u, t, color, i);
 258                 }
 259
 260                 if (i > 0 && !(i % 500)) {
 261                         struct tree_node *best = u->policy->choose(u->policy, t->root, b, color);
 262                         if (best && best->u.playouts >= 1500 && best->u.value >= u->loss_threshold)
 263                                 break;
 264                 }
 265
 266                 if (halt) {
 267                         if (UDEBUGL(2))
 268                                 fprintf(stderr, "<halting early, %d games skipped>\n", games - i);
 269                         break;
 270                 }
 271         }
 272
 273         progress_status(u, t, color, i);
 274         if (UDEBUGL(3))
 275                 tree_dump(t, u->dumpthres);
 276         return i;
 277 }
 278
 279 static pthread_mutex_t finish_mutex = PTHREAD_MUTEX_INITIALIZER;
 280 static pthread_cond_t finish_cond = PTHREAD_COND_INITIALIZER;
 281 static volatile int finish_thread;
 282 static pthread_mutex_t finish_serializer = PTHREAD_MUTEX_INITIALIZER;
 283
 284 struct spawn_ctx {
 285         int tid;
 286         struct uct *u;
 287         struct board *b;
 288         enum stone color;
 289         struct tree *t;
 290         unsigned long seed;
 291         int games;
 292 };
 293
 294 static void *
 295 spawn_helper(void *ctx_)
 296 {
 297         struct spawn_ctx *ctx = ctx_;
 298         /* Setup */
 299         fast_srandom(ctx->seed);
 300         /* Run */
 301         ctx->games = uct_playouts(ctx->u, ctx->b, ctx->color, ctx->t);
 302         /* Finish */
 303         pthread_mutex_lock(&finish_serializer);
 304         pthread_mutex_lock(&finish_mutex);
 305         finish_thread = ctx->tid;
 306         pthread_cond_signal(&finish_cond);
 307         pthread_mutex_unlock(&finish_mutex);
 308         return ctx;
 309 }
 310
 311 static void
 312 uct_notify_play(struct engine *e, struct board *b, struct move *m)
 313 {
 314         prepare_move(e, b, stone_other(m->color), m->coord);
 315 }
 316
 317 static coord_t *
 318 uct_genmove(struct engine *e, struct board *b, enum stone color)
 319 {
 320         struct uct *u = e->data;
 321
 322         /* Seed the tree. */
 323         prepare_move(e, b, color, resign);
 324
 325         int played_games = 0;
 326         if (!u->threads) {
 327                 played_games = uct_playouts(u, b, color, u->t);
 328         } else {
 329                 pthread_t threads[u->threads];
 330                 int joined = 0;
 331                 halt = 0;
 332                 pthread_mutex_lock(&finish_mutex);
 333                 /* Spawn threads... */
 334                 for (int ti = 0; ti < u->threads; ti++) {
 335                         struct spawn_ctx *ctx = malloc(sizeof(*ctx));
 336                         ctx->u = u; ctx->b = b; ctx->color = color;
 337                         ctx->t = tree_copy(u->t); ctx->tid = ti;
 338                         ctx->seed = fast_random(65536) + ti;
 339                         pthread_create(&threads[ti], NULL, spawn_helper, ctx);
 340                         if (UDEBUGL(2))
 341                                 fprintf(stderr, "Spawned thread %d\n", ti);
 342                 }
 343                 /* ...and collect them back: */
 344                 while (joined < u->threads) {
 345                         /* Wait for some thread to finish... */
 346                         pthread_cond_wait(&finish_cond, &finish_mutex);
 347                         /* ...and gather its remnants. */
 348                         struct spawn_ctx *ctx;
 349                         pthread_join(threads[finish_thread], (void **) &ctx);
 350                         played_games += ctx->games;
 351                         joined++;
 352                         tree_merge(u->t, ctx->t);
 353                         tree_done(ctx->t);
 354                         free(ctx);
 355                         if (UDEBUGL(2))
 356                                 fprintf(stderr, "Joined thread %d\n", finish_thread);
 357                         /* Do not get stalled by slow threads. */
 358                         if (joined >= u->threads / 2)
 359                                 halt = 1;
 360                         pthread_mutex_unlock(&finish_serializer);
 361                 }
 362                 pthread_mutex_unlock(&finish_mutex);
 363         }
 364
 365         if (UDEBUGL(2))
 366                 tree_dump(u->t, u->dumpthres);
 367
 368         struct tree_node *best = u->policy->choose(u->policy, u->t->root, b, color);
 369         if (!best) {
 370                 tree_done(u->t); u->t = NULL;
 371                 return coord_copy(pass);
 372         }
 373         if (UDEBUGL(0))
 374                 fprintf(stderr, "*** WINNER is %s (%d,%d) with score %1.4f (%d/%d:%d games)\n", coord2sstr(best->coord, b), coord_x(best->coord, b), coord_y(best->coord, b), best->u.value, best->u.playouts, u->t->root->u.playouts, played_games);
 375         if (best->u.value < u->resign_ratio && !is_pass(best->coord)) {
 376                 tree_done(u->t); u->t = NULL;
 377                 return coord_copy(resign);
 378         }
 379         tree_promote_node(u->t, best);
 380         return coord_copy(best->coord);
 381 }
 382
 383 bool
 384 uct_genbook(struct engine *e, struct board *b, enum stone color)
 385 {
 386         struct uct *u = e->data;
 387         u->t = tree_init(b, color);
 388         tree_load(u->t, b, color);
 389
 390         int i;
 391         for (i = 0; i < u->games; i++) {
 392                 int result = uct_playout(u, b, color, u->t);
 393                 if (result < 0) {
 394                         /* Tree descent has hit invalid move. */
 395                         continue;
 396                 }
 397
 398                 if (i > 0 && !(i % 10000)) {
 399                         progress_status(u, u->t, color, i);
 400                 }
 401         }
 402         progress_status(u, u->t, color, i);
 403
 404         tree_save(u->t, b, u->games / 100);
 405
 406         tree_done(u->t);
 407
 408         return true;
 409 }
 410
 411 void
 412 uct_dumpbook(struct engine *e, struct board *b, enum stone color)
 413 {
 414         struct uct *u = e->data;
 415         u->t = tree_init(b, color);
 416         tree_load(u->t, b, color);
 417         tree_dump(u->t, 0);
 418         tree_done(u->t);
 419 }
 420
 421
 422 struct uct *
 423 uct_state_init(char *arg)
 424 {
 425         struct uct *u = calloc(1, sizeof(struct uct));
 426
 427         u->debug_level = 1;
 428         u->games = MC_GAMES;
 429         u->gamelen = MC_GAMELEN;
 430         u->expand_p = 2;
 431         u->dumpthres = 1000;
 432         u->playout_amaf = false;
 433
 434         if (arg) {
 435                 char *optspec, *next = arg;
 436                 while (*next) {
 437                         optspec = next;
 438                         next += strcspn(next, ",");
 439                         if (*next) { *next++ = 0; } else { *next = 0; }
 440
 441                         char *optname = optspec;
 442                         char *optval = strchr(optspec, '=');
 443                         if (optval) *optval++ = 0;
 444
 445                         if (!strcasecmp(optname, "debug")) {
 446                                 if (optval)
 447                                         u->debug_level = atoi(optval);
 448                                 else
 449                                         u->debug_level++;
 450                         } else if (!strcasecmp(optname, "games") && optval) {
 451                                 u->games = atoi(optval);
 452                         } else if (!strcasecmp(optname, "gamelen") && optval) {
 453                                 u->gamelen = atoi(optval);
 454                         } else if (!strcasecmp(optname, "expand_p") && optval) {
 455                                 u->expand_p = atoi(optval);
 456                         } else if (!strcasecmp(optname, "radar_d") && optval) {
 457                                 /* For 19x19, it is good idea to set this to 3. */
 458                                 u->radar_d = atoi(optval);
 459                         } else if (!strcasecmp(optname, "dumpthres") && optval) {
 460                                 u->dumpthres = atoi(optval);
 461                         } else if (!strcasecmp(optname, "playout_amaf")) {
 462                                 /* Whether to include random playout moves in
 463                                  * AMAF as well. (Otherwise, only tree moves
 464                                  * are included in AMAF. Of course makes sense
 465                                  * only in connection with an AMAF policy.) */
 466                                 /* with-without: 55.5% (+-4.1) */
 467                                 if (optval && *optval == '0')
 468                                         u->playout_amaf = false;
 469                                 else
 470                                         u->playout_amaf = true;
 471                         } else if (!strcasecmp(optname, "policy") && optval) {
 472                                 char *policyarg = strchr(optval, ':');
 473                                 if (policyarg)
 474                                         *policyarg++ = 0;
 475                                 if (!strcasecmp(optval, "ucb1")) {
 476                                         u->policy = policy_ucb1_init(u, policyarg);
 477                                 } else if (!strcasecmp(optval, "ucb1tuned")) {
 478                                         u->policy = policy_ucb1tuned_init(u, policyarg);
 479                                 } else if (!strcasecmp(optval, "ucb1amaf")) {
 480                                         u->policy = policy_ucb1amaf_init(u, policyarg);
 481                                 } else {
 482                                         fprintf(stderr, "UCT: Invalid tree policy %s\n", optval);
 483                                 }
 484                         } else if (!strcasecmp(optname, "playout") && optval) {
 485                                 char *playoutarg = strchr(optval, ':');
 486                                 if (playoutarg)
 487                                         *playoutarg++ = 0;
 488                                 if (!strcasecmp(optval, "old")) {
 489                                         u->playout = playout_old_init(playoutarg);
 490                                 } else if (!strcasecmp(optval, "moggy")) {
 491                                         u->playout = playout_moggy_init(playoutarg);
 492                                 } else if (!strcasecmp(optval, "light")) {
 493                                         u->playout = playout_light_init(playoutarg);
 494                                 } else {
 495                                         fprintf(stderr, "UCT: Invalid playout policy %s\n", optval);
 496                                 }
 497                         } else if (!strcasecmp(optname, "threads") && optval) {
 498                                 u->threads = atoi(optval);
 499                         } else if (!strcasecmp(optname, "force_seed") && optval) {
 500                                 u->force_seed = atoi(optval);
 501                         } else {
 502                                 fprintf(stderr, "uct: Invalid engine argument %s or missing value\n", optname);
 503                         }
 504                 }
 505         }
 506
 507         u->resign_ratio = 0.2; /* Resign when most games are lost. */
 508         u->loss_threshold = 0.85; /* Stop reading if after at least 1500 playouts this is best value. */
 509         if (!u->policy)
 510                 u->policy = policy_ucb1amaf_init(u, NULL);
 511
 512         if (!u->playout)
 513                 u->playout = playout_moggy_init(NULL);
 514         u->playout->debug_level = u->debug_level;
 515
 516         return u;
 517 }
 518
 519
 520 struct engine *
 521 engine_uct_init(char *arg)
 522 {
 523         struct uct *u = uct_state_init(arg);
 524         struct engine *e = calloc(1, sizeof(struct engine));
 525         e->name = "UCT Engine";
 526         e->comment = "I'm playing UCT. When we both pass, I will consider all the stones on the board alive. If you are reading this, write 'yes'. Please bear with me at the game end, I need to fill the whole board; if you help me, we will both be happier. Filling the board will not lose points (NZ rules).";
 527         e->genmove = uct_genmove;
 528         e->notify_play = uct_notify_play;
 529         e->data = u;
 530
 531         return e;
 532 }