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