uct/walk.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 "probdist.h"
  15 #include "random.h"
  16 #include "tactics.h"
  17 #include "uct/internal.h"
  18 #include "uct/tree.h"
  19 #include "uct/uct.h"
  20 #include "uct/walk.h"
  21
  22 float
  23 uct_get_extra_komi(struct uct *u, struct board *b)
  24 {
  25         float extra_komi = board_effective_handicap(b) * (u->dynkomi - b->moves) / u->dynkomi;
  26         return extra_komi;
  27 }
  28
  29 void
  30 uct_progress_status(struct uct *u, struct tree *t, enum stone color, int playouts)
  31 {
  32         if (!UDEBUGL(0))
  33                 return;
  34
  35         /* Best move */
  36         struct tree_node *best = u->policy->choose(u->policy, t->root, t->board, color);
  37         if (!best) {
  38                 fprintf(stderr, "... No moves left\n");
  39                 return;
  40         }
  41         fprintf(stderr, "[%d] ", playouts);
  42         fprintf(stderr, "best %f ", tree_node_get_value(t, 1, best->u.value));
  43
  44         /* Max depth */
  45         fprintf(stderr, "deepest % 2d ", t->max_depth - t->root->depth);
  46
  47         /* Best sequence */
  48         fprintf(stderr, "| seq ");
  49         for (int depth = 0; depth < 6; depth++) {
  50                 if (best && best->u.playouts >= 25) {
  51                         fprintf(stderr, "%3s ", coord2sstr(best->coord, t->board));
  52                         best = u->policy->choose(u->policy, best, t->board, color);
  53                 } else {
  54                         fprintf(stderr, "    ");
  55                 }
  56         }
  57
  58         /* Best candidates */
  59         fprintf(stderr, "| can ");
  60         int cans = 4;
  61         struct tree_node *can[cans];
  62         memset(can, 0, sizeof(can));
  63         best = t->root->children;
  64         while (best) {
  65                 int c = 0;
  66                 while ((!can[c] || best->u.playouts > can[c]->u.playouts) && ++c < cans);
  67                 for (int d = 0; d < c; d++) can[d] = can[d + 1];
  68                 if (c > 0) can[c - 1] = best;
  69                 best = best->sibling;
  70         }
  71         while (--cans >= 0) {
  72                 if (can[cans]) {
  73                         fprintf(stderr, "%3s(%.3f) ",
  74                                 coord2sstr(can[cans]->coord, t->board),
  75                                 tree_node_get_value(t, 1, 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         int parity = (next_color == player_color ? 1 : -1);
  93
  94         /* We need to make sure only one thread expands the node. If
  95          * we are unlucky enough for two threads to meet in the same
  96          * node, the latter one will simply do another simulation from
  97          * the node itself, no big deal. t->nodes_size may exceed
  98          * the maximum in multi-threaded case but not by much so it's ok.
  99          * The size test must be before the test&set not after, to allow
 100          * expansion of the node later if enough nodes have been freed. */
 101         if (n->u.playouts >= u->expand_p && t->nodes_size < u->max_tree_size
 102             && !__sync_lock_test_and_set(&n->is_expanded, 1)) {
 103                 tree_expand_node(t, n, b, next_color, u, parity);
 104         }
 105         if (UDEBUGL(7))
 106                 fprintf(stderr, "%s*-- UCT playout #%d start [%s] %f\n",
 107                         spaces, n->u.playouts, coord2sstr(n->coord, t->board),
 108                         tree_node_get_value(t, parity, n->u.value));
 109
 110         struct playout_setup ps = { .gamelen = u->gamelen, .mercymin = u->mercymin };
 111         int result = play_random_game(&ps, b, next_color,
 112                                       u->playout_amaf ? amaf : NULL,
 113                                       &u->ownermap, u->playout);
 114         if (next_color == S_WHITE) {
 115                 /* We need the result from black's perspective. */
 116                 result = - result;
 117         }
 118         if (UDEBUGL(7))
 119                 fprintf(stderr, "%s -- [%d..%d] %s random playout result %d\n",
 120                         spaces, player_color, next_color, coord2sstr(n->coord, t->board), result);
 121
 122         return result;
 123 }
 124
 125 static float
 126 scale_value(struct uct *u, struct board *b, int result)
 127 {
 128         float rval = result > 0;
 129         if (u->val_scale) {
 130                 int vp = u->val_points;
 131                 if (!vp) {
 132                         vp = board_size(b) - 1; vp *= vp; vp *= 2;
 133                 }
 134
 135                 float sval = (float) abs(result) / vp;
 136                 sval = sval > 1 ? 1 : sval;
 137                 if (result < 0) sval = 1 - sval;
 138                 if (u->val_extra)
 139                         rval += u->val_scale * sval;
 140                 else
 141                         rval = (1 - u->val_scale) * rval + u->val_scale * sval;
 142                 // fprintf(stderr, "score %d => sval %f, rval %f\n", result, sval, rval);
 143         }
 144         return rval;
 145 }
 146
 147
 148 int
 149 uct_playout(struct uct *u, struct board *b, enum stone player_color, struct tree *t)
 150 {
 151         struct board b2;
 152         board_copy(&b2, b);
 153
 154         struct playout_amafmap *amaf = NULL;
 155         if (u->policy->wants_amaf) {
 156                 amaf = calloc(1, sizeof(*amaf));
 157                 amaf->map = calloc(board_size2(&b2) + 1, sizeof(*amaf->map));
 158                 amaf->map++; // -1 is pass
 159         }
 160
 161         /* Walk the tree until we find a leaf, then expand it and do
 162          * a random playout. */
 163         struct tree_node *n = t->root;
 164         enum stone node_color = stone_other(player_color);
 165         assert(node_color == t->root_color);
 166
 167         void *dstate = NULL, *dstater = NULL;
 168
 169         int result;
 170         int pass_limit = (board_size(&b2) - 2) * (board_size(&b2) - 2) / 2;
 171         int passes = is_pass(b->last_move.coord) && b->moves > 0;
 172
 173         /* debug */
 174         int depth = 0;
 175         static char spaces[] = "\0                                                      ";
 176         /* /debug */
 177         if (UDEBUGL(8))
 178                 fprintf(stderr, "--- UCT walk with color %d\n", player_color);
 179
 180         while (!tree_leaf_node(n) && passes < 2) {
 181                 spaces[depth++] = ' '; spaces[depth] = 0;
 182
 183                 /* Parity is chosen already according to the child color, since
 184                  * it is applied to children. */
 185                 node_color = stone_other(node_color);
 186                 int parity = (node_color == player_color ? 1 : -1);
 187                 n = (!u->random_policy_chance || fast_random(u->random_policy_chance))
 188                         ? u->policy->descend(u->policy, &dstate, t, n, parity, pass_limit)
 189                         : u->random_policy->descend(u->random_policy, &dstater, t, n, parity, pass_limit);
 190
 191                 assert(n == t->root || n->parent);
 192                 if (UDEBUGL(7))
 193                         fprintf(stderr, "%s+-- UCT sent us to [%s:%d] %f\n",
 194                                 spaces, coord2sstr(n->coord, t->board), n->coord,
 195                                 tree_node_get_value(t, parity, n->u.value));
 196
 197                 /* Add virtual loss if we need to; this is used to discourage
 198                  * other threads from visiting this node in case of multiple
 199                  * threads doing the tree search. */
 200                 if (u->virtual_loss)
 201                         stats_add_result(&n->u, tree_parity(t, parity) > 0 ? 0 : 1, 1);
 202
 203                 assert(n->coord >= -1);
 204                 if (amaf && !is_pass(n->coord)) {
 205                         if (amaf->map[n->coord] == S_NONE || amaf->map[n->coord] == node_color) {
 206                                 amaf->map[n->coord] = node_color;
 207                         } else { // XXX: Respect amaf->record_nakade
 208                                 amaf_op(amaf->map[n->coord], +);
 209                         }
 210                         amaf->game[amaf->gamelen].coord = n->coord;
 211                         amaf->game[amaf->gamelen].color = node_color;
 212                         amaf->gamelen++;
 213                         assert(amaf->gamelen < sizeof(amaf->game) / sizeof(amaf->game[0]));
 214                 }
 215
 216                 struct move m = { n->coord, node_color };
 217                 int res = board_play(&b2, &m);
 218
 219                 if (res < 0 || (!is_pass(m.coord) && !group_at(&b2, m.coord)) /* suicide */
 220                     || b2.superko_violation) {
 221                         if (UDEBUGL(3)) {
 222                                 for (struct tree_node *ni = n; ni; ni = ni->parent)
 223                                         fprintf(stderr, "%s<%"PRIhash"> ", coord2sstr(ni->coord, t->board), ni->hash);
 224                                 fprintf(stderr, "marking invalid %s node %d,%d res %d group %d spk %d\n",
 225                                         stone2str(node_color), coord_x(n->coord,b), coord_y(n->coord,b),
 226                                         res, group_at(&b2, m.coord), b2.superko_violation);
 227                         }
 228                         n->hints |= TREE_HINT_INVALID;
 229                         result = 0;
 230                         goto end;
 231                 }
 232
 233                 if (is_pass(n->coord))
 234                         passes++;
 235                 else
 236                         passes = 0;
 237         }
 238
 239         if (amaf) {
 240                 amaf->game_baselen = amaf->gamelen;
 241                 amaf->record_nakade = u->playout_amaf_nakade;
 242         }
 243
 244         if (u->dynkomi > b2.moves && (player_color & u->dynkomi_mask))
 245                 b2.komi += uct_get_extra_komi(u, &b2);
 246
 247         if (passes >= 2) {
 248                 /* XXX: No dead groups support. */
 249                 float score = board_official_score(&b2, NULL);
 250                 /* Result from black's perspective (no matter who
 251                  * the player; black's perspective is always
 252                  * what the tree stores. */
 253                 result = - (score * 2);
 254
 255                 if (UDEBUGL(5))
 256                         fprintf(stderr, "[%d..%d] %s p-p scoring playout result %d (W %f)\n",
 257                                 player_color, node_color, coord2sstr(n->coord, t->board), result, score);
 258                 if (UDEBUGL(6))
 259                         board_print(&b2, stderr);
 260
 261                 board_ownermap_fill(&u->ownermap, &b2);
 262
 263         } else { assert(u->parallel_tree || tree_leaf_node(n));
 264                 /* In case of parallel tree search, the assertion might
 265                  * not hold if two threads chew on the same node. */
 266                 result = uct_leaf_node(u, &b2, player_color, amaf, t, n, node_color, spaces);
 267         }
 268
 269         if (amaf && u->playout_amaf_cutoff) {
 270                 int cutoff = amaf->game_baselen;
 271                 cutoff += (amaf->gamelen - amaf->game_baselen) * u->playout_amaf_cutoff / 100;
 272                 /* Now, reconstruct the amaf map. */
 273                 memset(amaf->map, 0, board_size2(&b2) * sizeof(*amaf->map));
 274                 for (int i = 0; i < cutoff; i++) {
 275                         coord_t coord = amaf->game[i].coord;
 276                         enum stone color = amaf->game[i].color;
 277                         if (amaf->map[coord] == S_NONE || amaf->map[coord] == color) {
 278                                 amaf->map[coord] = color;
 279                         /* Nakade always recorded for in-tree part */
 280                         } else if (amaf->record_nakade || i <= amaf->game_baselen) {
 281                                 amaf_op(amaf->map[n->coord], +);
 282                         }
 283                 }
 284         }
 285
 286         assert(n == t->root || n->parent);
 287         if (result != 0) {
 288                 float rval = scale_value(u, b, result);
 289
 290                 u->policy->update(u->policy, t, n, node_color, player_color, amaf, rval);
 291
 292                 if (u->root_heuristic && n->parent) {
 293                         if (!t->chvals) {
 294                                 t->chvals = calloc(board_size2(b), sizeof(t->chvals[0]));
 295                                 t->chchvals = calloc(board_size2(b), sizeof(t->chchvals[0]));
 296                         }
 297
 298                         /* Possibly transform the rval appropriately. */
 299                         rval = stats_temper_value(rval, n->parent->u.value, u->root_heuristic);
 300
 301                         struct tree_node *ni = n;
 302                         while (ni->parent->parent && ni->parent->parent->parent)
 303                                 ni = ni->parent;
 304                         if (ni->parent->parent) {
 305                                 if (likely(!is_pass(ni->coord)))
 306                                         stats_add_result(&t->chchvals[ni->coord], rval, 1);
 307                                 ni = ni->parent;
 308                         }
 309                         assert(ni->parent && !ni->parent->parent);
 310
 311                         if (likely(!is_pass(ni->coord)))
 312                                 stats_add_result(&t->chvals[ni->coord], rval, 1);
 313                 }
 314         }
 315
 316 end:
 317         /* We need to undo the virtual loss we added during descend. */
 318         if (u->virtual_loss) {
 319                 int parity = (node_color == player_color ? 1 : -1);
 320                 for (; n->parent; n = n->parent) {
 321                         stats_rm_result(&n->u, tree_parity(t, parity) > 0 ? 0 : 1, 1);
 322                         parity = -parity;
 323                 }
 324         }
 325
 326         if (dstater) free(dstater);
 327         if (dstate) free(dstate);
 328         if (amaf) {
 329                 free(amaf->map - 1);
 330                 free(amaf);
 331         }
 332         board_done_noalloc(&b2);
 333         return result;
 334 }
 335
 336 int
 337 uct_playouts(struct uct *u, struct board *b, enum stone color, struct tree *t, int games)
 338 {
 339         /* Should we print progress info? In case all threads work on the same
 340          * tree, only the first thread does. */
 341         /* XXX: The thread manager should do things like this. */
 342         #define ok_to_talk (!u->parallel_tree || !thread_id)
 343
 344         int i;
 345         for (i = 0; !uct_halt && (!games || i < games); i++) {
 346                 int result = uct_playout(u, b, color, t);
 347                 if (result == 0) {
 348                         /* Tree descent has hit invalid move. */
 349                         continue;
 350                 }
 351
 352                 if (unlikely(ok_to_talk && i > 0 && !(i % 10000))) {
 353                         uct_progress_status(u, t, color, i);
 354                 }
 355
 356                 if (i > 0 && !(i % 500)) {
 357                         struct tree_node *best = u->policy->choose(u->policy, t->root, b, color);
 358                         if (best && ((best->u.playouts >= 2000 && tree_node_get_value(t, 1, best->u.value) >= u->loss_threshold)
 359                                      || (best->u.playouts >= 500 && tree_node_get_value(t, 1, best->u.value) >= 0.95)))
 360                                 break;
 361                 }
 362         }
 363
 364         if (ok_to_talk) {
 365                 uct_progress_status(u, t, color, i);
 366                 if (UDEBUGL(3))
 367                         tree_dump(t, u->dumpthres);
 368         }
 369         return i;
 370 }