patternsp.c

   1 #define DEBUG
   2 #include <assert.h>
   3 #include <ctype.h>
   4 #include <inttypes.h>
   5 #include <stdio.h>
   6 #include <stdlib.h>
   7
   8 #include "board.h"
   9 #include "debug.h"
  10 #include "pattern.h"
  11 #include "patternsp.h"
  12
  13 /* Mapping from point sequence to coordinate offsets (to determine
  14  * coordinates relative to pattern center). The array is ordered
  15  * in the gridcular metric order so that we can go through it
  16  * and incrementally match spatial features in nested circles.
  17  * Within one circle, coordinates are ordered by rows to keep
  18  * good cache behavior. */
  19 struct ptcoord ptcoords[MAX_PATTERN_AREA];
  20
  21 /* For each radius, starting index in ptcoords[]. */
  22 unsigned int ptind[MAX_PATTERN_DIST + 2];
  23
  24 /* ptcoords[], ptind[] setup */
  25 static void
  26 ptcoords_init(void)
  27 {
  28         int i = 0; /* Indexing ptcoords[] */
  29
  30         /* First, center point. */
  31         ptind[0] = ptind[1] = 0;
  32         ptcoords[i].x = ptcoords[i].y = 0; i++;
  33
  34         for (int d = 2; d <= MAX_PATTERN_DIST; d++) {
  35                 ptind[d] = i;
  36                 /* For each y, examine all integer solutions
  37                  * of d = |x| + |y| + max(|x|, |y|). */
  38                 /* TODO: (Stern, 2006) uses a hand-modified
  39                  * circles that are finer for small d and more
  40                  * coarse for large d. */
  41                 for (short y = d / 2; y >= 0; y--) {
  42                         short x;
  43                         if (y > d / 3) {
  44                                 /* max(|x|, |y|) = |y|, non-zero x */
  45                                 x = d - y * 2;
  46                                 if (x + y * 2 != d) continue;
  47                         } else {
  48                                 /* max(|x|, |y|) = |x| */
  49                                 /* Or, max(|x|, |y|) = |y| and x is zero */
  50                                 x = (d - y) / 2;
  51                                 if (x * 2 + y != d) continue;
  52                         }
  53
  54                         assert((x > y ? x : y) + x + y == d);
  55
  56                         ptcoords[i].x = x; ptcoords[i].y = y; i++;
  57                         if (x != 0) { ptcoords[i].x = -x; ptcoords[i].y = y; i++; }
  58                         if (y != 0) { ptcoords[i].x = x; ptcoords[i].y = -y; i++; }
  59                         if (x != 0 && y != 0) { ptcoords[i].x = -x; ptcoords[i].y = -y; i++; }
  60                 }
  61         }
  62         ptind[MAX_PATTERN_DIST + 1] = i;
  63
  64 #if 0
  65         for (int d = 0; d <= MAX_PATTERN_DIST; d++) {
  66                 fprintf(stderr, "d=%d (%d) ", d, ptind[d]);
  67                 for (int j = ptind[d]; j < ptind[d + 1]; j++) {
  68                         fprintf(stderr, "%d,%d ", ptcoords[j].x, ptcoords[j].y);
  69                 }
  70                 fprintf(stderr, "\n");
  71         }
  72 #endif
  73 }
  74
  75
  76 /* Zobrist hashes used for points in patterns. */
  77 hash_t pthashes[PTH__ROTATIONS][MAX_PATTERN_AREA][S_MAX];
  78
  79 static void
  80 pthashes_init(void)
  81 {
  82         /* We need fixed hashes for all pattern-relative in
  83          * all pattern users! This is a simple way to generate
  84          * hopefully good ones. Park-Miller powa. :) */
  85
  86         /* We create a virtual board (centered at the sequence start),
  87          * plant the hashes there, then pick them up into the sequence
  88          * with correct coordinates. It would be possible to generate
  89          * the sequence point hashes directly, but the rotations would
  90          * make for enormous headaches. */
  91 #define PATTERN_BOARD_SIZE ((MAX_PATTERN_DIST + 1) * (MAX_PATTERN_DIST + 1))
  92         hash_t pthboard[PATTERN_BOARD_SIZE][4];
  93         int pthbc = PATTERN_BOARD_SIZE / 2; // tengen coord
  94
  95         /* The magic numbers are tuned for minimal collisions. */
  96         hash_t h1 = 0xd6d6d6d1;
  97         hash_t h2 = 0xd6d6d6d2;
  98         hash_t h3 = 0xd6d6d6d3;
  99         hash_t h4 = 0xd6d6d6d4;
 100         for (int i = 0; i < PATTERN_BOARD_SIZE; i++) {
 101                 pthboard[i][S_NONE] = (h1 = h1 * 16787);
 102                 pthboard[i][S_BLACK] = (h2 = h2 * 16823);
 103                 pthboard[i][S_WHITE] = (h3 = h3 * 16811 - 13);
 104                 pthboard[i][S_OFFBOARD] = (h4 = h4 * 16811);
 105         }
 106
 107         /* Virtual board with hashes created, now fill
 108          * pthashes[] with hashes for points in actual
 109          * sequences, also considering various rotations. */
 110 #define PTH_VMIRROR     1
 111 #define PTH_HMIRROR     2
 112 #define PTH_90ROT       4
 113         for (int r = 0; r < PTH__ROTATIONS; r++) {
 114                 for (int i = 0; i < MAX_PATTERN_AREA; i++) {
 115                         /* Rotate appropriately. */
 116                         int rx = ptcoords[i].x;
 117                         int ry = ptcoords[i].y;
 118                         if (r & PTH_VMIRROR) ry = -ry;
 119                         if (r & PTH_HMIRROR) rx = -rx;
 120                         if (r & PTH_90ROT) {
 121                                 int rs = rx; rx = -ry; ry = rs;
 122                         }
 123                         int bi = pthbc + ry * (MAX_PATTERN_DIST + 1) + rx;
 124
 125                         /* Copy info. */
 126                         pthashes[r][i][S_NONE] = pthboard[bi][S_NONE];
 127                         pthashes[r][i][S_BLACK] = pthboard[bi][S_BLACK];
 128                         pthashes[r][i][S_WHITE] = pthboard[bi][S_WHITE];
 129                         pthashes[r][i][S_OFFBOARD] = pthboard[bi][S_OFFBOARD];
 130                 }
 131         }
 132 }
 133
 134 static void __attribute__((constructor))
 135 spatial_init(void)
 136 {
 137         /* Initialization of various static data structures for
 138          * fast pattern processing. */
 139         ptcoords_init();
 140         pthashes_init();
 141 }
 142
 143 inline hash_t
 144 spatial_hash(unsigned int rotation, struct spatial *s)
 145 {
 146         hash_t h = 0;
 147         for (unsigned int i = 0; i < ptind[s->dist + 1]; i++) {
 148                 h ^= pthashes[rotation][i][spatial_point_at(*s, i)];
 149         }
 150         return h & spatial_hash_mask;
 151 }
 152
 153 char *
 154 spatial2str(struct spatial *s)
 155 {
 156         static char buf[1024];
 157         for (unsigned int i = 0; i < ptind[s->dist + 1]; i++) {
 158                 buf[i] = stone2char(spatial_point_at(*s, i));
 159         }
 160         buf[ptind[s->dist + 1]] = 0;
 161         return buf;
 162 }
 163
 164 void
 165 spatial_from_board(struct pattern_config *pc, struct spatial *s,
 166                    struct board *b, struct move *m)
 167 {
 168         assert(pc->spat_min > 0);
 169
 170         /* We record all spatial patterns black-to-play; simply
 171          * reverse all colors if we are white-to-play. */
 172         static enum stone bt_black[4] = { S_NONE, S_BLACK, S_WHITE, S_OFFBOARD };
 173         static enum stone bt_white[4] = { S_NONE, S_WHITE, S_BLACK, S_OFFBOARD };
 174         enum stone (*bt)[4] = m->color == S_WHITE ? &bt_white : &bt_black;
 175
 176         memset(s, 0, sizeof(*s));
 177         for (unsigned int j = 0; j < ptind[pc->spat_max + 1]; j++) {
 178                 ptcoords_at(x, y, m->coord, b, j);
 179                 s->points[j / 4] |= (*bt)[board_atxy(b, x, y)] << ((j % 4) * 2);
 180         }
 181         s->dist = pc->spat_max;
 182 }
 183
 184 /* Compare two spatials, allowing for differences up to isomorphism.
 185  * True means the spatials are equivalent. */
 186 static bool
 187 spatial_cmp(struct spatial *s1, struct spatial *s2)
 188 {
 189         /* Quick preliminary check. */
 190         if (s1->dist != s2->dist)
 191                 return false;
 192
 193         /* We could create complex transposition tables, but it seems most
 194          * foolproof to just check if the sets of rotation hashes are the
 195          * same for both. */
 196         hash_t s1r[PTH__ROTATIONS];
 197         for (unsigned int r = 0; r < PTH__ROTATIONS; r++)
 198                 s1r[r] = spatial_hash(r, s1);
 199         for (unsigned int r = 0; r < PTH__ROTATIONS; r++) {
 200                 hash_t s2r = spatial_hash(r, s2);
 201                 for (unsigned int p = 0; p < PTH__ROTATIONS; p++)
 202                         if (s2r == s1r[p])
 203                                 goto found_rot;
 204                 /* Rotation hash s2r does not correspond to s1r. */
 205                 return false;
 206 found_rot:;
 207         }
 208
 209         /* All rotation hashes of s2 occur in s1. Hopefully that
 210          * indicates something. */
 211         return true;
 212 }
 213
 214
 215 /* Spatial dict manipulation. */
 216
 217 static unsigned int
 218 spatial_dict_addc(struct spatial_dict *dict, struct spatial *s)
 219 {
 220         /* Allocate space in 1024 blocks. */
 221 #define SPATIALS_ALLOC 1024
 222         if (!(dict->nspatials % SPATIALS_ALLOC)) {
 223                 dict->spatials = realloc(dict->spatials,
 224                                 (dict->nspatials + SPATIALS_ALLOC)
 225                                 * sizeof(*dict->spatials));
 226         }
 227         dict->spatials[dict->nspatials] = *s;
 228         return dict->nspatials++;
 229 }
 230
 231 bool
 232 spatial_dict_addh(struct spatial_dict *dict, hash_t hash, unsigned int id)
 233 {
 234         if (dict->hash[hash]) {
 235                 if (dict->hash[hash] != id)
 236                         dict->collisions++;
 237         } else {
 238                 dict->fills++;
 239         }
 240         dict->hash[hash] = id;
 241         return true;
 242 }
 243
 244 /* Spatial dictionary file format:
 245  * /^#/ - comments
 246  * INDEX RADIUS STONES HASH...
 247  * INDEX: index in the spatial table
 248  * RADIUS: @d of the pattern
 249  * STONES: string of ".XO#" chars
 250  * HASH...: space-separated 18bit hash-table indices for the pattern */
 251
 252 static void
 253 spatial_dict_read(struct spatial_dict *dict, char *buf, bool hash)
 254 {
 255         /* XXX: We trust the data. Bad data will crash us. */
 256         char *bufp = buf;
 257
 258         unsigned int index, radius;
 259         index = strtoul(bufp, &bufp, 10);
 260         radius = strtoul(bufp, &bufp, 10);
 261         while (isspace(*bufp)) bufp++;
 262
 263         if (radius > MAX_PATTERN_DIST) {
 264                 /* Too large spatial, skip. */
 265                 struct spatial s = { .dist = 0 };
 266                 unsigned int id = spatial_dict_addc(dict, &s);
 267                 assert(id == index);
 268                 return;
 269         }
 270
 271         /* Load the stone configuration. */
 272         struct spatial s = { .dist = radius };
 273         unsigned int sl = 0;
 274         while (!isspace(*bufp)) {
 275                 s.points[sl / 4] |= char2stone(*bufp++) << ((sl % 4)*2);
 276                 sl++;
 277         }
 278         while (isspace(*bufp)) bufp++;
 279
 280         /* Sanity check. */
 281         if (sl != ptind[s.dist + 1]) {
 282                 fprintf(stderr, "Spatial dictionary: Invalid number of stones (%d != %d) on this line: %s\n",
 283                         sl, ptind[radius + 1] - 1, buf);
 284                 exit(EXIT_FAILURE);
 285         }
 286
 287         /* Add to collection. */
 288         unsigned int id = spatial_dict_addc(dict, &s);
 289         assert(id == index);
 290
 291         /* Add to specified hash places. */
 292         if (hash)
 293                 for (unsigned int r = 0; r < PTH__ROTATIONS; r++)
 294                         spatial_dict_addh(dict, spatial_hash(r, &s), id);
 295 }
 296
 297 void
 298 spatial_write(struct spatial_dict *dict, struct spatial *s, unsigned int id, FILE *f)
 299 {
 300         fprintf(f, "%d %d ", id, s->dist);
 301         fputs(spatial2str(s), f);
 302         for (unsigned int r = 0; r < PTH__ROTATIONS; r++) {
 303                 hash_t rhash = spatial_hash(r, s);
 304                 unsigned int id2 = dict->hash[rhash];
 305                 if (id2 != id) {
 306                         /* This hash does not belong to us. Decide whether
 307                          * we or the current owner is better owner. */
 308                         /* TODO: Compare also # of patternscan encounters? */
 309                         struct spatial *s2 = &dict->spatials[id2];
 310                         if (s2->dist < s->dist)
 311                                 continue;
 312                         if (s2->dist == s->dist && id2 < id)
 313                                 continue;
 314                 }
 315                 fprintf(f, " %"PRIhash"", spatial_hash(r, s));
 316         }
 317         fputc('\n', f);
 318 }
 319
 320 static void
 321 spatial_dict_load(struct spatial_dict *dict, FILE *f, bool hash)
 322 {
 323         char buf[1024];
 324         while (fgets(buf, sizeof(buf), f)) {
 325                 if (buf[0] == '#') continue;
 326                 spatial_dict_read(dict, buf, hash);
 327         }
 328         if (DEBUGL(1)) {
 329                 fprintf(stderr, "Loaded spatial dictionary of %d patterns.\n", dict->nspatials);
 330                 if (hash)
 331                         spatial_dict_hashstats(dict);
 332         }
 333 }
 334
 335 void
 336 spatial_dict_hashstats(struct spatial_dict *dict)
 337 {
 338         /* m hash size, n number of patterns; is zobrist universal hash?
 339          *
 340          * Not so rigorous analysis, but it should give a good approximation:
 341          * Probability of empty bucket is (1-1/m)^n ~ e^(-n/m)
 342          * Probability of non-empty bucket is 1-e^(-n/m)
 343          * Expected number of non-empty buckets is m*(1-e^(-n/m))
 344          * Number of collisions is n-m*(1-e^(-n/m)). */
 345
 346         /* The result: Reality matches these expectations pretty well!
 347          *
 348          * Actual:
 349          *      Loaded spatial dictionary of 1064482 patterns.
 350          *      (Spatial dictionary hash: 513997 collisions (incl. repetitions), 11.88% (7970033/67108864) fill rate).
 351          *
 352          * Theoretical:
 353          *      m = 2^26
 354          *      n <= 8*1064482 (some patterns may have some identical rotations)
 355          *      n = 513997+7970033 = 8484030 should be the correct number
 356          *      n-m*(1-e^(-n/m)) = 514381
 357          *
 358          * To verify, make sure to turn patternprob off (e.g. use
 359          * -e patternscan), since it will insert a pattern multiple times,
 360          * multiplying the reported number of collisions. */
 361
 362         unsigned long buckets = (sizeof(dict->hash) / sizeof(dict->hash[0]));
 363         fprintf(stderr, "\t(Spatial dictionary hash: %d collisions (incl. repetitions), %.2f%% (%d/%lu) fill rate).\n",
 364                         dict->collisions,
 365                         (double) dict->fills * 100 / buckets,
 366                         dict->fills, buckets);
 367 }
 368
 369 void
 370 spatial_dict_writeinfo(struct spatial_dict *dict, FILE *f)
 371 {
 372         /* New file. First, create a comment describing order
 373          * of points in the array. This is just for purposes
 374          * of external tools, Pachi never interprets it itself. */
 375         fprintf(f, "# Pachi spatial patterns dictionary v1.0 maxdist %d\n",
 376                 MAX_PATTERN_DIST);
 377         for (unsigned int d = 0; d <= MAX_PATTERN_DIST; d++) {
 378                 fprintf(f, "# Point order: d=%d ", d);
 379                 for (unsigned int j = ptind[d]; j < ptind[d + 1]; j++) {
 380                         fprintf(f, "%d,%d ", ptcoords[j].x, ptcoords[j].y);
 381                 }
 382                 fprintf(f, "\n");
 383         }
 384 }
 385
 386 /* We try to avoid needlessly reloading spatial dictionary
 387  * since it may take rather long time. */
 388 static struct spatial_dict *cached_dict;
 389
 390 const char * default_spatial_dict_filename = "patterns.spat";
 391
 392 struct spatial_dict *
 393 spatial_dict_init(bool will_append, bool hash, char * spatial_dict_filename)
 394 {
 395         if (cached_dict && !will_append)
 396                 return cached_dict;
 397
 398         FILE *f = fopen(spatial_dict_filename, "r");
 399         if (!f && !will_append) {
 400                 if (DEBUGL(1))
 401                         fprintf(stderr, "No spatial dictionary, will not match spatial pattern features.\n");
 402                 return NULL;
 403         }
 404
 405         struct spatial_dict *dict = calloc2(1, sizeof(*dict));
 406         /* We create a dummy record for index 0 that we will
 407          * never reference. This is so that hash value 0 can
 408          * represent "no value". */
 409         struct spatial dummy = { .dist = 0 };
 410         spatial_dict_addc(dict, &dummy);
 411
 412         if (f) {
 413                 spatial_dict_load(dict, f, hash);
 414                 fclose(f); f = NULL;
 415         } else {
 416                 assert(will_append);
 417         }
 418
 419         cached_dict = dict;
 420         return dict;
 421 }
 422
 423 unsigned int
 424 spatial_dict_put(struct spatial_dict *dict, struct spatial *s, hash_t h)
 425 {
 426         /* We avoid spatial_dict_get() here, since we want to ignore radius
 427          * differences - we have custom collision detection. */
 428         unsigned int id = dict->hash[h];
 429         if (id > 0) {
 430                 /* Is this the same or isomorphous spatial? */
 431                 if (spatial_cmp(s, &dict->spatials[id]))
 432                         return id;
 433
 434                 /* Look a bit harder - perhaps one of our rotations still
 435                  * points at the correct spatial. */
 436                 for (unsigned int r = 0; r < PTH__ROTATIONS; r++) {
 437                         hash_t rhash = spatial_hash(r, s);
 438                         unsigned int rid = dict->hash[rhash];
 439                         /* No match means we definitely aren't stored yet. */
 440                         if (!rid)
 441                                 break;
 442                         if (id != rid && spatial_cmp(s, &dict->spatials[rid])) {
 443                                 /* Yay, this is us! */
 444                                 if (DEBUGL(3))
 445                                         fprintf(stderr, "Repeated collision %d vs %d\n", id, rid);
 446                                 id = rid;
 447                                 /* Point the hashes back to us. */
 448                                 goto hash_store;
 449                         }
 450                 }
 451
 452                 if (DEBUGL(1))
 453                         fprintf(stderr, "Collision %d vs %d\n", id, dict->nspatials);
 454                 id = 0;
 455                 /* dict->collisions++; gets done by addh */
 456         }
 457
 458         /* Add new pattern! */
 459         id = spatial_dict_addc(dict, s);
 460         if (DEBUGL(4)) {
 461                 fprintf(stderr, "new spat %d(%d) %s <%"PRIhash"> ", id, s->dist, spatial2str(s), h);
 462                 for (unsigned int r = 0; r < 8; r++)
 463                         fprintf(stderr,"[%"PRIhash"] ", spatial_hash(r, s));
 464                 fprintf(stderr, "\n");
 465         }
 466
 467         /* Store new pattern in the hash. */
 468 hash_store:
 469         for (unsigned int r = 0; r < PTH__ROTATIONS; r++)
 470                 spatial_dict_addh(dict, spatial_hash(r, s), id);
 471
 472         return id;
 473 }