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 // This should become a dynamic parameter. 7G is suitable for 20k*23 threads.
  23 #define MAX_NODE_SIZES (7L*1024*1024*1024)
  24
  25 float
  26 uct_get_extra_komi(struct uct *u, struct board *b)
  27 {
  28         float extra_komi = board_effective_handicap(b) * (u->dynkomi - b->moves) / u->dynkomi;
  29         return extra_komi;
  30 }
  31
  32 void
  33 uct_progress_status(struct uct *u, struct tree *t, enum stone color, int playouts)
  34 {
  35         if (!UDEBUGL(0))
  36                 return;
  37
  38         /* Best move */
  39         struct tree_node *best = u->policy->choose(u->policy, t->root, t->board, color);
  40         if (!best) {
  41                 fprintf(stderr, "... No moves left\n");
  42                 return;
  43         }
  44         fprintf(stderr, "[%d] ", playouts);
  45         fprintf(stderr, "best %f ", tree_node_get_value(t, 1, best->u.value));
  46
  47         /* Max depth */
  48         fprintf(stderr, "deepest % 2d ", t->max_depth - t->root->depth);
  49
  50         /* Best sequence */
  51         fprintf(stderr, "| seq ");
  52         for (int depth = 0; depth < 6; depth++) {
  53                 if (best && best->u.playouts >= 25) {
  54                         fprintf(stderr, "%3s ", coord2sstr(best->coord, t->board));
  55                         best = u->policy->choose(u->policy, best, t->board, color);
  56                 } else {
  57                         fprintf(stderr, "    ");
  58                 }
  59         }
  60
  61         /* Best candidates */
  62         fprintf(stderr, "| can ");
  63         int cans = 4;
  64         struct tree_node *can[cans];
  65         memset(can, 0, sizeof(can));
  66         best = t->root->children;
  67         while (best) {
  68                 int c = 0;
  69                 while ((!can[c] || best->u.playouts > can[c]->u.playouts) && ++c < cans);
  70                 for (int d = 0; d < c; d++) can[d] = can[d + 1];
  71                 if (c > 0) can[c - 1] = best;
  72                 best = best->sibling;
  73         }
  74         while (--cans >= 0) {
  75                 if (can[cans]) {
  76                         fprintf(stderr, "%3s(%.3f) ",
  77                                 coord2sstr(can[cans]->coord, t->board),
  78                                 tree_node_get_value(t, 1, can[cans]->u.value));
  79                 } else {
  80                         fprintf(stderr, "           ");
  81                 }
  82         }
  83
  84         fprintf(stderr, "\n");
  85 }
  86
  87
  88 static int
  89 uct_leaf_node(struct uct *u, struct board *b, enum stone player_color,
  90               struct playout_amafmap *amaf,
  91               struct tree *t, struct tree_node *n, enum stone node_color,
  92               char *spaces)
  93 {
  94         enum stone next_color = stone_other(node_color);
  95         int parity = (next_color == player_color ? 1 : -1);
  96
  97         /* We need to make sure only one thread expands the node. If
  98          * we are unlucky enough for two threads to meet in the same
  99          * node, the latter one will simply do another simulation from
 100          * the node itself, no big deal. t->node_sizes may exceed
 101          * MAX_NODE_SIZES in multi-threaded case but not by much so it's ok.
 102          * The size test must be before the test&set not after, to allow
 103          * expansion of the node later if enough nodes have been freed. */
 104         if (n->u.playouts >= u->expand_p && t->node_sizes < MAX_NODE_SIZES
 105             && !__sync_lock_test_and_set(&n->is_expanded, 1)) {
 106                 tree_expand_node(t, n, b, next_color, u, parity);
 107         }
 108         if (UDEBUGL(7))
 109                 fprintf(stderr, "%s*-- UCT playout #%d start [%s] %f\n",
 110                         spaces, n->u.playouts, coord2sstr(n->coord, t->board),
 111                         tree_node_get_value(t, parity, n->u.value));
 112
 113         struct playout_setup ps = { .gamelen = u->gamelen };
 114         int result = play_random_game(&ps, b, next_color,
 115                                       u->playout_amaf ? amaf : NULL,
 116                                       &u->ownermap, u->playout);
 117         if (next_color == S_WHITE) {
 118                 /* We need the result from black's perspective. */
 119                 result = - result;
 120         }
 121         if (UDEBUGL(7))
 122                 fprintf(stderr, "%s -- [%d..%d] %s random playout result %d\n",
 123                         spaces, player_color, next_color, coord2sstr(n->coord, t->board), result);
 124
 125         return result;
 126 }
 127
 128 static float
 129 scale_value(struct uct *u, struct board *b, int result)
 130 {
 131         float rval = result > 0;
 132         if (u->val_scale) {
 133                 int vp = u->val_points;
 134                 if (!vp) {
 135                         vp = board_size(b) - 1; vp *= vp; vp *= 2;
 136                 }
 137
 138                 float sval = (float) abs(result) / vp;
 139                 sval = sval > 1 ? 1 : sval;
 140                 if (result < 0) sval = 1 - sval;
 141                 if (u->val_extra)
 142                         rval += u->val_scale * sval;
 143                 else
 144                         rval = (1 - u->val_scale) * rval + u->val_scale * sval;
 145                 // fprintf(stderr, "score %d => sval %f, rval %f\n", result, sval, rval);
 146         }
 147         return rval;
 148 }
 149
 150
 151 int
 152 uct_playout(struct uct *u, struct board *b, enum stone player_color, struct tree *t)
 153 {
 154         struct board b2;
 155         board_copy(&b2, b);
 156
 157         struct playout_amafmap *amaf = NULL;
 158         if (u->policy->wants_amaf) {
 159                 amaf = calloc(1, sizeof(*amaf));
 160                 amaf->map = calloc(board_size2(&b2) + 1, sizeof(*amaf->map));
 161                 amaf->map++; // -1 is pass
 162         }
 163
 164         /* Walk the tree until we find a leaf, then expand it and do
 165          * a random playout. */
 166         struct tree_node *n = t->root;
 167         enum stone node_color = stone_other(player_color);
 168         assert(node_color == t->root_color);
 169
 170         void *dstate = NULL, *dstater = NULL;
 171
 172         int result;
 173         int pass_limit = (board_size(&b2) - 2) * (board_size(&b2) - 2) / 2;
 174         int passes = is_pass(b->last_move.coord) && b->moves > 0;
 175
 176         /* debug */
 177         int depth = 0;
 178         static char spaces[] = "\0                                                      ";
 179         /* /debug */
 180         if (UDEBUGL(8))
 181                 fprintf(stderr, "--- UCT walk with color %d\n", player_color);
 182
 183         while (!tree_leaf_node(n) && passes < 2) {
 184                 spaces[depth++] = ' '; spaces[depth] = 0;
 185
 186                 /* Parity is chosen already according to the child color, since
 187                  * it is applied to children. */
 188                 node_color = stone_other(node_color);
 189                 int parity = (node_color == player_color ? 1 : -1);
 190                 n = (!u->random_policy_chance || fast_random(u->random_policy_chance))
 191                         ? u->policy->descend(u->policy, &dstate, t, n, parity, pass_limit)
 192                         : u->random_policy->descend(u->random_policy, &dstater, t, n, parity, pass_limit);
 193
 194                 assert(n == t->root || n->parent);
 195                 if (UDEBUGL(7))
 196                         fprintf(stderr, "%s+-- UCT sent us to [%s:%d] %f\n",
 197                                 spaces, coord2sstr(n->coord, t->board), n->coord,
 198                                 tree_node_get_value(t, parity, n->u.value));
 199
 200                 /* Add virtual loss if we need to; this is used to discourage
 201                  * other threads from visiting this node in case of multiple
 202                  * threads doing the tree search. */
 203                 if (u->virtual_loss)
 204                         stats_add_result(&n->u, tree_parity(t, parity) > 0 ? 0 : 1, 1);
 205
 206                 assert(n->coord >= -1);
 207                 if (amaf && !is_pass(n->coord)) {
 208                         if (amaf->map[n->coord] == S_NONE || amaf->map[n->coord] == node_color) {
 209                                 amaf->map[n->coord] = node_color;
 210                         } else { // XXX: Respect amaf->record_nakade
 211                                 amaf_op(amaf->map[n->coord], +);
 212                         }
 213                         amaf->game[amaf->gamelen].coord = n->coord;
 214                         amaf->game[amaf->gamelen].color = node_color;
 215                         amaf->gamelen++;
 216                         assert(amaf->gamelen < sizeof(amaf->game) / sizeof(amaf->game[0]));
 217                 }
 218
 219                 struct move m = { n->coord, node_color };
 220                 int res = board_play(&b2, &m);
 221
 222                 if (res < 0 || (!is_pass(m.coord) && !group_at(&b2, m.coord)) /* suicide */
 223                     || b2.superko_violation) {
 224                         if (UDEBUGL(3)) {
 225                                 for (struct tree_node *ni = n; ni; ni = ni->parent)
 226                                         fprintf(stderr, "%s<%"PRIhash"> ", coord2sstr(ni->coord, t->board), ni->hash);
 227                                 fprintf(stderr, "marking invalid %s node %d,%d res %d group %d spk %d\n",
 228                                         stone2str(node_color), coord_x(n->coord,b), coord_y(n->coord,b),
 229                                         res, group_at(&b2, m.coord), b2.superko_violation);
 230                         }
 231                         n->hints |= TREE_HINT_INVALID;
 232                         result = 0;
 233                         goto end;
 234                 }
 235
 236                 if (is_pass(n->coord))
 237                         passes++;
 238                 else
 239                         passes = 0;
 240         }
 241
 242         if (amaf) {
 243                 amaf->game_baselen = amaf->gamelen;
 244                 amaf->record_nakade = u->playout_amaf_nakade;
 245         }
 246
 247         if (u->dynkomi > b2.moves && (player_color & u->dynkomi_mask))
 248                 b2.komi += uct_get_extra_komi(u, &b2);
 249
 250         if (passes >= 2) {
 251                 /* XXX: No dead groups support. */
 252                 float score = board_official_score(&b2, NULL);
 253                 /* Result from black's perspective (no matter who
 254                  * the player; black's perspective is always
 255                  * what the tree stores. */
 256                 result = - (score * 2);
 257
 258                 if (UDEBUGL(5))
 259                         fprintf(stderr, "[%d..%d] %s p-p scoring playout result %d (W %f)\n",
 260                                 player_color, node_color, coord2sstr(n->coord, t->board), result, score);
 261                 if (UDEBUGL(6))
 262                         board_print(&b2, stderr);
 263
 264                 board_ownermap_fill(&u->ownermap, &b2);
 265
 266         } else { assert(u->parallel_tree || tree_leaf_node(n));
 267                 /* In case of parallel tree search, the assertion might
 268                  * not hold if two threads chew on the same node. */
 269                 result = uct_leaf_node(u, &b2, player_color, amaf, t, n, node_color, spaces);
 270         }
 271
 272         if (amaf && u->playout_amaf_cutoff) {
 273                 int cutoff = amaf->game_baselen;
 274                 cutoff += (amaf->gamelen - amaf->game_baselen) * u->playout_amaf_cutoff / 100;
 275                 /* Now, reconstruct the amaf map. */
 276                 memset(amaf->map, 0, board_size2(&b2) * sizeof(*amaf->map));
 277                 for (int i = 0; i < cutoff; i++) {
 278                         coord_t coord = amaf->game[i].coord;
 279                         enum stone color = amaf->game[i].color;
 280                         if (amaf->map[coord] == S_NONE || amaf->map[coord] == color) {
 281                                 amaf->map[coord] = color;
 282                         /* Nakade always recorded for in-tree part */
 283                         } else if (amaf->record_nakade || i <= amaf->game_baselen) {
 284                                 amaf_op(amaf->map[n->coord], +);
 285                         }
 286                 }
 287         }
 288
 289         assert(n == t->root || n->parent);
 290         if (result != 0) {
 291                 float rval = scale_value(u, b, result);
 292
 293                 u->policy->update(u->policy, t, n, node_color, player_color, amaf, rval);
 294
 295                 if (u->root_heuristic && n->parent) {
 296                         if (!t->chvals) {
 297                                 t->chvals = calloc(board_size2(b), sizeof(t->chvals[0]));
 298                                 t->chchvals = calloc(board_size2(b), sizeof(t->chchvals[0]));
 299                         }
 300
 301                         /* Possibly transform the rval appropriately. */
 302                         rval = stats_temper_value(rval, n->parent->u.value, u->root_heuristic);
 303
 304                         struct tree_node *ni = n;
 305                         while (ni->parent->parent && ni->parent->parent->parent)
 306                                 ni = ni->parent;
 307                         if (ni->parent->parent) {
 308                                 if (likely(!is_pass(ni->coord)))
 309                                         stats_add_result(&t->chchvals[ni->coord], rval, 1);
 310                                 ni = ni->parent;
 311                         }
 312                         assert(ni->parent && !ni->parent->parent);
 313
 314                         if (likely(!is_pass(ni->coord)))
 315                                 stats_add_result(&t->chvals[ni->coord], rval, 1);
 316                 }
 317         }
 318
 319 end:
 320         /* We need to undo the virtual loss we added during descend. */
 321         if (u->virtual_loss) {
 322                 int parity = (node_color == player_color ? 1 : -1);
 323                 for (; n->parent; n = n->parent) {
 324                         stats_rm_result(&n->u, tree_parity(t, parity) > 0 ? 0 : 1, 1);
 325                         parity = -parity;
 326                 }
 327         }
 328
 329         if (dstater) free(dstater);
 330         if (dstate) free(dstate);
 331         if (amaf) {
 332                 free(amaf->map - 1);
 333                 free(amaf);
 334         }
 335         board_done_noalloc(&b2);
 336         return result;
 337 }
 338
 339 int
 340 uct_playouts(struct uct *u, struct board *b, enum stone color, struct tree *t, int games)
 341 {
 342         /* Should we print progress info? In case all threads work on the same
 343          * tree, only the first thread does. */
 344         #define ok_to_talk (!u->parallel_tree || !thread_id)
 345
 346         int i;
 347         for (i = 0; i < games; i++) {
 348                 int result = uct_playout(u, b, color, t);
 349                 if (result == 0) {
 350                         /* Tree descent has hit invalid move. */
 351                         continue;
 352                 }
 353
 354                 if (unlikely(ok_to_talk && i > 0 && !(i % 10000))) {
 355                         uct_progress_status(u, t, color, i);
 356                 }
 357
 358                 if (i > 0 && !(i % 500)) {
 359                         struct tree_node *best = u->policy->choose(u->policy, t->root, b, color);
 360                         if (best && ((best->u.playouts >= 2000 && tree_node_get_value(t, 1, best->u.value) >= u->loss_threshold)
 361                                      || (best->u.playouts >= 500 && tree_node_get_value(t, 1, best->u.value) >= 0.95)))
 362                                 break;
 363                 }
 364
 365                 if (uct_halt) {
 366                         if (UDEBUGL(2))
 367                                 fprintf(stderr, "<halting early, %d games skipped>\n", games - i);
 368                         break;
 369                 }
 370         }
 371
 372         if (ok_to_talk) {
 373                 uct_progress_status(u, t, color, i);
 374                 if (UDEBUGL(3))
 375                         tree_dump(t, u->dumpthres);
 376         }
 377         return i;
 378 }