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 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(int rotation, struct spatial *s)
 145 {
 146         hash_t h = 0;
 147         for (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 (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 (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 (int r = 0; r < PTH__ROTATIONS; r++)
 198                 s1r[r] = spatial_hash(r, s1);
 199         for (int r = 0; r < PTH__ROTATIONS; r++) {
 200                 hash_t s2r = spatial_hash(r, s2);
 201                 for (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         /* Load the stone configuration. */
 264         struct spatial s = { .dist = radius };
 265         int sl = 0;
 266         while (!isspace(*bufp)) {
 267                 s.points[sl / 4] |= char2stone(*bufp++) << ((sl % 4)*2);
 268                 sl++;
 269         }
 270         while (isspace(*bufp)) bufp++;
 271
 272         /* Sanity check. */
 273         if (sl != ptind[s.dist + 1]) {
 274                 fprintf(stderr, "Spatial dictionary: Invalid number of stones (%d != %d) on this line: %s\n",
 275                         sl, ptind[radius + 1] - 1, buf);
 276                 exit(EXIT_FAILURE);
 277         }
 278
 279         /* Add to collection. */
 280         unsigned int id = spatial_dict_addc(dict, &s);
 281         assert(id == index);
 282
 283         /* Add to specified hash places. */
 284         if (hash)
 285                 for (int r = 0; r < PTH__ROTATIONS; r++)
 286                         spatial_dict_addh(dict, spatial_hash(r, &s), id);
 287 }
 288
 289 void
 290 spatial_write(struct spatial_dict *dict, struct spatial *s, int id, FILE *f)
 291 {
 292         fprintf(f, "%d %d ", id, s->dist);
 293         fputs(spatial2str(s), f);
 294         for (int r = 0; r < PTH__ROTATIONS; r++) {
 295                 hash_t rhash = spatial_hash(r, s);
 296                 int id2 = dict->hash[rhash];
 297                 if (id2 != id) {
 298                         /* This hash does not belong to us. Decide whether
 299                          * we or the current owner is better owner. */
 300                         /* TODO: Compare also # of patternscan encounters? */
 301                         struct spatial *s2 = &dict->spatials[id2];
 302                         if (s2->dist < s->dist)
 303                                 continue;
 304                         if (s2->dist == s->dist && id2 < id)
 305                                 continue;
 306                 }
 307                 fprintf(f, " %"PRIhash"", spatial_hash(r, s));
 308         }
 309         fputc('\n', f);
 310 }
 311
 312 static void
 313 spatial_dict_load(struct spatial_dict *dict, FILE *f, bool hash)
 314 {
 315         char buf[1024];
 316         while (fgets(buf, sizeof(buf), f)) {
 317                 if (buf[0] == '#') continue;
 318                 spatial_dict_read(dict, buf, hash);
 319         }
 320         if (DEBUGL(1)) {
 321                 fprintf(stderr, "Loaded spatial dictionary of %d patterns.\n", dict->nspatials);
 322                 if (hash)
 323                         spatial_dict_hashstats(dict);
 324         }
 325 }
 326
 327 void
 328 spatial_dict_hashstats(struct spatial_dict *dict)
 329 {
 330         /* m hash size, n number of patterns; is zobrist universal hash?
 331          *
 332          * Not so rigorous analysis, but it should give a good approximation:
 333          * Probability of empty bucket is (1-1/m)^n ~ e^(-n/m)
 334          * Probability of non-empty bucket is 1-e^(-n/m)
 335          * Expected number of non-empty buckets is m*(1-e^(-n/m))
 336          * Number of collisions is n-m*(1-e^(-n/m)). */
 337
 338         /* The result: Reality matches these expectations pretty well!
 339          *
 340          * Actual:
 341          *      Loaded spatial dictionary of 1064482 patterns.
 342          *      (Spatial dictionary hash: 513997 collisions (incl. repetitions), 11.88% (7970033/67108864) fill rate).
 343          *
 344          * Theoretical:
 345          *      m = 2^26
 346          *      n <= 8*1064482 (some patterns may have some identical rotations)
 347          *      n = 513997+7970033 = 8484030 should be the correct number
 348          *      n-m*(1-e^(-n/m)) = 514381
 349          *
 350          * To verify, make sure to turn patternprob off (e.g. use
 351          * -e patternscan), since it will insert a pattern multiple times,
 352          * multiplying the reported number of collisions. */
 353
 354         unsigned long buckets = (sizeof(dict->hash) / sizeof(dict->hash[0]));
 355         fprintf(stderr, "\t(Spatial dictionary hash: %d collisions (incl. repetitions), %.2f%% (%d/%lu) fill rate).\n",
 356                         dict->collisions,
 357                         (double) dict->fills * 100 / buckets,
 358                         dict->fills, buckets);
 359 }
 360
 361 void
 362 spatial_dict_writeinfo(struct spatial_dict *dict, FILE *f)
 363 {
 364         /* New file. First, create a comment describing order
 365          * of points in the array. This is just for purposes
 366          * of external tools, Pachi never interprets it itself. */
 367         fprintf(f, "# Pachi spatial patterns dictionary v1.0 maxdist %d\n",
 368                 MAX_PATTERN_DIST);
 369         for (int d = 0; d <= MAX_PATTERN_DIST; d++) {
 370                 fprintf(f, "# Point order: d=%d ", d);
 371                 for (int j = ptind[d]; j < ptind[d + 1]; j++) {
 372                         fprintf(f, "%d,%d ", ptcoords[j].x, ptcoords[j].y);
 373                 }
 374                 fprintf(f, "\n");
 375         }
 376 }
 377
 378 /* We try to avoid needlessly reloading spatial dictionary
 379  * since it may take rather long time. */
 380 static struct spatial_dict *cached_dict;
 381
 382 const char *spatial_dict_filename = "patterns.spat";
 383 struct spatial_dict *
 384 spatial_dict_init(bool will_append, bool hash)
 385 {
 386         if (cached_dict && !will_append)
 387                 return cached_dict;
 388
 389         FILE *f = fopen(spatial_dict_filename, "r");
 390         if (!f && !will_append) {
 391                 if (DEBUGL(1))
 392                         fprintf(stderr, "No spatial dictionary, will not match spatial pattern features.\n");
 393                 return NULL;
 394         }
 395
 396         struct spatial_dict *dict = calloc2(1, sizeof(*dict));
 397         /* We create a dummy record for index 0 that we will
 398          * never reference. This is so that hash value 0 can
 399          * represent "no value". */
 400         struct spatial dummy = { .dist = 0 };
 401         spatial_dict_addc(dict, &dummy);
 402
 403         if (f) {
 404                 spatial_dict_load(dict, f, hash);
 405                 fclose(f); f = NULL;
 406         } else {
 407                 assert(will_append);
 408         }
 409
 410         cached_dict = dict;
 411         return dict;
 412 }
 413
 414 int
 415 spatial_dict_put(struct spatial_dict *dict, struct spatial *s, hash_t h)
 416 {
 417         /* We avoid spatial_dict_get() here, since we want to ignore radius
 418          * differences - we have custom collision detection. */
 419         int id = dict->hash[h];
 420         if (id > 0) {
 421                 /* Is this the same or isomorphous spatial? */
 422                 if (spatial_cmp(s, &dict->spatials[id]))
 423                         return id;
 424
 425                 /* Look a bit harder - perhaps one of our rotations still
 426                  * points at the correct spatial. */
 427                 for (int r = 0; r < PTH__ROTATIONS; r++) {
 428                         hash_t rhash = spatial_hash(r, s);
 429                         int rid = dict->hash[rhash];
 430                         /* No match means we definitely aren't stored yet. */
 431                         if (!rid)
 432                                 break;
 433                         if (id != rid && spatial_cmp(s, &dict->spatials[rid])) {
 434                                 /* Yay, this is us! */
 435                                 if (DEBUGL(3))
 436                                         fprintf(stderr, "Repeated collision %d vs %d\n", id, rid);
 437                                 id = rid;
 438                                 /* Point the hashes back to us. */
 439                                 goto hash_store;
 440                         }
 441                 }
 442
 443                 if (DEBUGL(1))
 444                         fprintf(stderr, "Collision %d vs %d\n", id, dict->nspatials);
 445                 id = 0;
 446                 /* dict->collisions++; gets done by addh */
 447         }
 448
 449         /* Add new pattern! */
 450         id = spatial_dict_addc(dict, s);
 451         if (DEBUGL(4)) {
 452                 fprintf(stderr, "new spat %d(%d) %s <%"PRIhash"> ", id, s->dist, spatial2str(s), h);
 453                 for (int r = 0; r < 8; r++)
 454                         fprintf(stderr,"[%"PRIhash"] ", spatial_hash(r, s));
 455                 fprintf(stderr, "\n");
 456         }
 457
 458         /* Store new pattern in the hash. */
 459 hash_store:
 460         for (int r = 0; r < PTH__ROTATIONS; r++)
 461                 spatial_dict_addh(dict, spatial_hash(r, s), id);
 462
 463         return id;
 464 }