uct/uct.c

   1 #include <assert.h>
   2 #include <stdio.h>
   3 #include <stdlib.h>
   4 #include <string.h>
   5
   6 #define DEBUG
   7
   8 #include "debug.h"
   9 #include "board.h"
  10 #include "move.h"
  11 #include "playout.h"
  12 #include "playout/moggy.h"
  13 #include "playout/old.h"
  14 #include "playout/light.h"
  15 #include "uct/internal.h"
  16 #include "uct/tree.h"
  17 #include "uct/uct.h"
  18
  19 struct uct_policy *policy_ucb1_init(struct uct *u, char *arg);
  20 struct uct_policy *policy_ucb1tuned_init(struct uct *u, char *arg);
  21 struct uct_policy *policy_ucb1amaf_init(struct uct *u, char *arg);
  22
  23
  24 #define MC_GAMES        40000
  25 #define MC_GAMELEN      400
  26
  27
  28 static void
  29 progress_status(struct uct *u, struct tree *t, enum stone color, int playouts)
  30 {
  31         if (!UDEBUGL(0))
  32                 return;
  33
  34         /* Best move */
  35         struct tree_node *best = u->policy->choose(u->policy, t->root, t->board, color);
  36         if (!best) {
  37                 fprintf(stderr, "... No moves left\n");
  38                 return;
  39         }
  40         fprintf(stderr, "[%d] ", playouts);
  41         fprintf(stderr, "best %f ", best->u.value);
  42
  43         /* Max depth */
  44         fprintf(stderr, "deepest % 2d ", t->max_depth - t->root->depth);
  45
  46         /* Best sequence */
  47         fprintf(stderr, "| seq ");
  48         for (int depth = 0; depth < 6; depth++) {
  49                 if (best && best->u.playouts >= 25) {
  50                         fprintf(stderr, "%3s ", coord2sstr(best->coord, t->board));
  51                         best = u->policy->choose(u->policy, best, t->board, color);
  52                 } else {
  53                         fprintf(stderr, "    ");
  54                 }
  55         }
  56
  57         /* Best candidates */
  58         fprintf(stderr, "| can ");
  59         int cans = 4;
  60         struct tree_node *can[cans];
  61         memset(can, 0, sizeof(can));
  62         best = t->root->children;
  63         while (best) {
  64                 int c = 0;
  65                 while ((!can[c] || best->u.playouts > can[c]->u.playouts) && ++c < cans);
  66                 for (int d = 0; d < c; d++) can[d] = can[d + 1];
  67                 if (c > 0) can[c - 1] = best;
  68                 best = best->sibling;
  69         }
  70         while (--cans >= 0) {
  71                 if (can[cans]) {
  72                         fprintf(stderr, "%3s(%.3f) ", coord2sstr(can[cans]->coord, t->board), can[cans]->u.value);
  73                 } else {
  74                         fprintf(stderr, "           ");
  75                 }
  76         }
  77
  78         fprintf(stderr, "\n");
  79 }
  80
  81
  82 static int
  83 uct_playout(struct uct *u, struct board *b, enum stone color, struct tree *t)
  84 {
  85         struct board b2;
  86         board_copy(&b2, b);
  87
  88         struct playout_amafmap *amaf = NULL;
  89         if (u->policy->wants_amaf) {
  90                 amaf = calloc(1, sizeof(*amaf));
  91                 amaf->map = calloc(board_size2(&b2) + 1, sizeof(*amaf->map));
  92                 amaf->map++; // -1 is pass
  93         }
  94
  95         /* Walk the tree until we find a leaf, then expand it and do
  96          * a random playout. */
  97         struct tree_node *n = t->root;
  98         enum stone orig_color = color;
  99         int result;
 100         int pass_limit = (board_size(&b2) - 2) * (board_size(&b2) - 2) / 2;
 101         int passes = is_pass(b->last_move.coord);
 102         if (UDEBUGL(8))
 103                 fprintf(stderr, "--- UCT walk with color %d\n", color);
 104         for (; pass; color = stone_other(color)) {
 105                 if (tree_leaf_node(n)) {
 106                         if (n->u.playouts >= u->expand_p)
 107                                 tree_expand_node(t, n, &b2, color, u->radar_d, u->policy, (color == orig_color ? 1 : -1));
 108
 109                         result = play_random_game(&b2, color, u->gamelen, u->playout_amaf ? amaf : NULL, u->playout);
 110                         if (orig_color != color && result >= 0)
 111                                 result = !result;
 112                         if (UDEBUGL(7))
 113                                 fprintf(stderr, "[%d..%d] %s random playout result %d\n", orig_color, color, coord2sstr(n->coord, t->board), result);
 114
 115                         /* Reset color to the @n color. */
 116                         color = stone_other(color);
 117                         break;
 118                 }
 119
 120                 n = u->policy->descend(u->policy, t, n, (color == orig_color ? 1 : -1), pass_limit);
 121                 assert(n == t->root || n->parent);
 122                 if (UDEBUGL(7))
 123                         fprintf(stderr, "-- UCT sent us to [%s] %f\n", coord2sstr(n->coord, t->board), n->u.value);
 124                 if (amaf && n->coord >= -1)
 125                         amaf->map[n->coord] = color;
 126                 struct move m = { n->coord, color };
 127                 int res = board_play(&b2, &m);
 128                 if (res < 0 || (!is_pass(m.coord) && !group_at(&b2, m.coord)) /* suicide */
 129                     || b2.superko_violation) {
 130                         if (UDEBUGL(6))
 131                                 fprintf(stderr, "deleting invalid node %d,%d\n", coord_x(n->coord,b), coord_y(n->coord,b));
 132                         tree_delete_node(t, n);
 133                         result = -1;
 134                         goto end;
 135                 }
 136
 137                 if (is_pass(n->coord)) {
 138                         passes++;
 139                         if (passes >= 2) {
 140                                 float score = board_official_score(&b2);
 141                                 result = (orig_color == S_BLACK) ? score < 0 : score > 0;
 142                                 if (UDEBUGL(5))
 143                                         fprintf(stderr, "[%d..%d] %s p-p scoring playout result %d (W %f)\n", orig_color, color, coord2sstr(n->coord, t->board), result, score);
 144                                 if (UDEBUGL(6))
 145                                         board_print(&b2, stderr);
 146                                 break;
 147                         }
 148                 } else {
 149                         passes = 0;
 150                 }
 151         }
 152
 153         assert(n == t->root || n->parent);
 154         if (result >= 0)
 155                 u->policy->update(u->policy, t, n, color, amaf, result);
 156
 157 end:
 158         if (amaf) {
 159                 free(amaf->map - 1);
 160                 free(amaf);
 161         }
 162         board_done_noalloc(&b2);
 163         return result;
 164 }
 165
 166 static coord_t *
 167 uct_genmove(struct engine *e, struct board *b, enum stone color)
 168 {
 169         struct uct *u = e->data;
 170
 171         if (!u->t) {
 172 tree_init:
 173                 u->t = tree_init(b, color);
 174                 //board_print(b, stderr);
 175
 176                 if (!b->moves)
 177                         tree_load(u->t, b);
 178         } else {
 179                 /* XXX: We hope that the opponent didn't suddenly play
 180                  * several moves in the row. */
 181                 for (struct tree_node *ni = u->t->root->children; ni; ni = ni->sibling)
 182                         if (ni->coord == b->last_move.coord) {
 183                                 tree_promote_node(u->t, ni);
 184                                 goto promoted;
 185                         }
 186                 fprintf(stderr, "CANNOT FIND NODE TO PROMOTE!\n");
 187                 tree_done(u->t);
 188                 goto tree_init;
 189 promoted:;
 190         }
 191
 192         int i, games = u->games - (u->t->root->u.playouts / 1.5);
 193         for (i = 0; i < games; i++) {
 194                 int result = uct_playout(u, b, color, u->t);
 195                 if (result < 0) {
 196                         /* Tree descent has hit invalid move. */
 197                         continue;
 198                 }
 199
 200                 if (i > 0 && !(i % 10000)) {
 201                         progress_status(u, u->t, color, i);
 202                 }
 203
 204                 if (i > 0 && !(i % 500)) {
 205                         struct tree_node *best = u->policy->choose(u->policy, u->t->root, b, color);
 206                         if (best && best->u.playouts >= 1000 && best->u.value >= u->loss_threshold)
 207                                 break;
 208                 }
 209         }
 210
 211         progress_status(u, u->t, color, i);
 212         if (UDEBUGL(2))
 213                 tree_dump(u->t, u->dumpthres);
 214
 215         struct tree_node *best = u->policy->choose(u->policy, u->t->root, b, color);
 216         if (!best) {
 217                 tree_done(u->t); u->t = NULL;
 218                 return coord_copy(pass);
 219         }
 220         if (UDEBUGL(0))
 221                 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, i);
 222         if (best->u.value < u->resign_ratio && !is_pass(best->coord)) {
 223                 tree_done(u->t); u->t = NULL;
 224                 return coord_copy(resign);
 225         }
 226         tree_promote_node(u->t, best);
 227         return coord_copy(best->coord);
 228 }
 229
 230 bool
 231 uct_genbook(struct engine *e, struct board *b, enum stone color)
 232 {
 233         struct uct *u = e->data;
 234         u->t = tree_init(b, color);
 235
 236         int i;
 237         for (i = 0; i < u->games; i++) {
 238                 int result = uct_playout(u, b, color, u->t);
 239                 if (result < 0) {
 240                         /* Tree descent has hit invalid move. */
 241                         continue;
 242                 }
 243
 244                 if (i > 0 && !(i % 10000)) {
 245                         progress_status(u, u->t, color, i);
 246                 }
 247         }
 248         progress_status(u, u->t, color, i);
 249
 250         tree_save(u->t, b, u->games / 1000);
 251
 252         tree_done(u->t);
 253
 254         return true;
 255 }
 256
 257 void
 258 uct_dumpbook(struct engine *e, struct board *b, enum stone color)
 259 {
 260         struct uct *u = e->data;
 261         u->t = tree_init(b, color);
 262         tree_load(u->t, b);
 263         tree_dump(u->t, 0);
 264         tree_done(u->t);
 265 }
 266
 267
 268 struct uct *
 269 uct_state_init(char *arg)
 270 {
 271         struct uct *u = calloc(1, sizeof(struct uct));
 272
 273         u->debug_level = 1;
 274         u->games = MC_GAMES;
 275         u->gamelen = MC_GAMELEN;
 276         u->expand_p = 2;
 277         u->dumpthres = 500;
 278
 279         if (arg) {
 280                 char *optspec, *next = arg;
 281                 while (*next) {
 282                         optspec = next;
 283                         next += strcspn(next, ",");
 284                         if (*next) { *next++ = 0; } else { *next = 0; }
 285
 286                         char *optname = optspec;
 287                         char *optval = strchr(optspec, '=');
 288                         if (optval) *optval++ = 0;
 289
 290                         if (!strcasecmp(optname, "debug")) {
 291                                 if (optval)
 292                                         u->debug_level = atoi(optval);
 293                                 else
 294                                         u->debug_level++;
 295                         } else if (!strcasecmp(optname, "games") && optval) {
 296                                 u->games = atoi(optval);
 297                         } else if (!strcasecmp(optname, "gamelen") && optval) {
 298                                 u->gamelen = atoi(optval);
 299                         } else if (!strcasecmp(optname, "expand_p") && optval) {
 300                                 u->expand_p = atoi(optval);
 301                         } else if (!strcasecmp(optname, "radar_d") && optval) {
 302                                 /* For 19x19, it is good idea to set this to 3. */
 303                                 u->radar_d = atoi(optval);
 304                         } else if (!strcasecmp(optname, "dumpthres") && optval) {
 305                                 u->dumpthres = atoi(optval);
 306                         } else if (!strcasecmp(optname, "playout_amaf")) {
 307                                 /* Whether to include random playout moves in
 308                                  * AMAF as well. (Otherwise, only tree moves
 309                                  * are included in AMAF. Of course makes sense
 310                                  * only in connection with an AMAF policy.) */
 311                                 u->playout_amaf = true;
 312                         } else if (!strcasecmp(optname, "policy") && optval) {
 313                                 char *policyarg = strchr(optval, ':');
 314                                 if (policyarg)
 315                                         *policyarg++ = 0;
 316                                 if (!strcasecmp(optval, "ucb1")) {
 317                                         u->policy = policy_ucb1_init(u, policyarg);
 318                                 } else if (!strcasecmp(optval, "ucb1tuned")) {
 319                                         u->policy = policy_ucb1tuned_init(u, policyarg);
 320                                 } else if (!strcasecmp(optval, "ucb1amaf")) {
 321                                         u->policy = policy_ucb1amaf_init(u, policyarg);
 322                                 } else {
 323                                         fprintf(stderr, "UCT: Invalid tree policy %s\n", optval);
 324                                 }
 325                         } else if (!strcasecmp(optname, "playout") && optval) {
 326                                 char *playoutarg = strchr(optval, ':');
 327                                 if (playoutarg)
 328                                         *playoutarg++ = 0;
 329                                 if (!strcasecmp(optval, "old")) {
 330                                         u->playout = playout_old_init(playoutarg);
 331                                 } else if (!strcasecmp(optval, "moggy")) {
 332                                         u->playout = playout_moggy_init(playoutarg);
 333                                 } else if (!strcasecmp(optval, "light")) {
 334                                         u->playout = playout_light_init(playoutarg);
 335                                 } else {
 336                                         fprintf(stderr, "UCT: Invalid playout policy %s\n", optval);
 337                                 }
 338                         } else {
 339                                 fprintf(stderr, "uct: Invalid engine argument %s or missing value\n", optname);
 340                         }
 341                 }
 342         }
 343
 344         u->resign_ratio = 0.2; /* Resign when most games are lost. */
 345         u->loss_threshold = 0.95; /* Stop reading if after at least 500 playouts this is best value. */
 346         if (!u->policy)
 347                 u->policy = policy_ucb1_init(u, NULL);
 348
 349         if (!u->playout)
 350                 u->playout = playout_moggy_init(NULL);
 351         u->playout->debug_level = u->debug_level;
 352
 353         return u;
 354 }
 355
 356
 357 struct engine *
 358 engine_uct_init(char *arg)
 359 {
 360         struct uct *u = uct_state_init(arg);
 361         struct engine *e = calloc(1, sizeof(struct engine));
 362         e->name = "UCT Engine";
 363         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).";
 364         e->genmove = uct_genmove;
 365         e->data = u;
 366
 367         return e;
 368 }