トライ（ダブルé…åˆ—ï¼Œç°¡æ½”ãƒ‡ãƒ¼ã‚¿æ§‹é€ ï¼‰ã¨ STL コンテナã®æ¯”較（最新版）
[2011/11/30 æ›´æ–°; std::(unordered_)map でメモリ使用量を見積もる - ny23の日記ã«å¾“ã„，STL コンテナã®ãƒ¡ãƒ¢ãƒªä½¿ç”¨é‡ã‚’計測]
[2011/02/21 æ›´æ–°: marisa-trie 0.1.3; 発表資料（Prefix/Patricia Trie の入れ子による辞書圧縮） - やた＠はてな日記 ã«ã¦ã“ã®è¨˜äº‹ã®å®Ÿé¨“çµæžœã‚’引用ã•ã‚Œã¦ã„ã‚‹ã®ã§ï¼Œä»¥å¾ŒåŽŸå‰‡æ›´æ–°ã—ãªã„ã“ã¨ã¨ã™ã‚‹ï¼ŽãªãŠï¼Œmarisa-trie 㯠検索時間が短くなりました - やた＠はてな日記 ã«ã‚るよã†ã«ï¼Œmarisa-0.2.0-beta3 以é™ã§ã¯ã•ã‚‰ã«æ¤œç´¢ãŒé€Ÿããªã£ã¦ã„ã¾ã™ï¼Ž]
[2011/02/18 æ›´æ–°: marisa-trie ã®ä»•æ§˜å¤‰æ›´ã«ä¼´ã„，追記ã®è¨˜è¿°ã‚’æ•´åˆæ€§ãŒå–れるよã†å¤‰æ›´; 最新版ã§ã¯æœªç¢ºèª]
id:s-yata ã•ã‚“㌠marisa-trie を公開ã•ã‚ŒãŸã®ã§ï¼Œä¾‹ã«ã‚ˆã£ã¦ãƒ™ãƒ³ãƒãƒžãƒ¼ã‚¯ã‚’å–ã£ãŸçµæžœã‚’公開ã—ã¦ãŠãã¾ã™ï¼Žä»¥å‰ï¼Œux-trie ã®ãƒã‚°çµ¡ã¿ã§ã‚³ãƒ¼ãƒ‰ã‚’èªã‚“ã ã¨ã，TAIL 部ã§ãƒˆãƒ©ã‚¤ã®æ®µæ•°ã‚’ã‚‚ã£ã¨å¢—ã‚„ã™ã¨ã©ã†ãªã‚‹ã®ã‹ãªã¨æ€ã£ã¦ã„ãŸã¨ã“ã‚ã ã£ãŸã®ã§ï¼Œéžå¸¸ã«è‰¯ã„タイミングã§ã—ãŸï¼Žmarisa-trie ã§ãƒˆãƒ©ã‚¤ã‚’構築ã™ã‚‹éš›ã®ãƒ‘ラメタã¯ï¼Œãƒˆãƒ©ã‚¤ã®æ•°ï¼ˆbuild ã®ç¬¬ä¸‰å¼•æ•°ï¼‰ã®ã¨ã“ã‚ã—ã‹å¤‰ãˆã¦ã„ã¾ã›ã‚“．Build ã¯æ§‹ç¯‰æ™‚間，Search ã¯è¾žæ›¸é †ï¼ŒSearch* ã¯ãƒ©ãƒ³ãƒ€ãƒ é †ã®æ¤œç´¢æ™‚間．Wikipedia-en 㯠20090929 ã®ã‚‚ã®ï¼Ž
個々ã®ãƒ©ã‚¤ãƒ–ラリã¸ã®ãƒªãƒ³ã‚¯ã‚„ç´°ã‹ã„表記ã®æ„味，marisa-trie 以外ã®çµæžœã«é–¢ã™ã‚‹è€ƒå¯Ÿã¯ä»¥ä¸‹ã®è¨˜äº‹ã‚’å‚照．
個人的ã«ã¯ marisa-trie ã¯åœ§ç¸®çŽ‡ã®é«˜ã•ã‚ˆã‚Šã‚‚，ãã®æ¤œç´¢é€Ÿåº¦ã®é€Ÿã•ã«ã‚ˆã‚Šé…力を感ã˜ã¾ã™ï¼Ž
[追記] marisa-trie 㧠patricia trie -> prefix tree ã«ã™ã‚‹ã¨ï¼ŒWikipedia-en ã§ã¯ãƒˆãƒ©ã‚¤ã®ã‚µã‚¤ã‚ºãŒ1-2割増ãˆï¼Œæ¤œç´¢é€Ÿåº¦ã‚‚1-2割è½ã¡ã¾ã™ï¼ˆéƒµä¾¿ç•ªå·è¾žæ›¸ã®æ–¹ã§ã¯ï¼Œãƒˆãƒ©ã‚¤ã®ã‚µã‚¤ã‚ºã¯ã»ã¼åŒã˜ã§æ¤œç´¢é€Ÿåº¦ã¯ prefix tree ã®æ–¹ãŒé€†ã«åƒ…ã‹ã«é«˜é€Ÿï¼‰ï¼Žæ¬¡ã«ï¼ŒTAIL 部を文å—列化ã—ãªã„ã§ãã®ã¾ã¾ãƒˆãƒ©ã‚¤ã§ä¿æŒã™ã‚‹ã¨ï¼Œï¼ˆWikipedia-en ã§ã¯ï¼‰ãƒˆãƒ©ã‚¤ã®æ•°ãŒä¸€ã¤ã®æ™‚ã®ã¿ã‚µã‚¤ã‚ºãŒå¤§å¹…ã«å¢—ãˆï¼ˆtx ã¨åŒã˜ãらã„），検索速度も特ã«ãƒ©ãƒ³ãƒ€ãƒ æ¤œç´¢é †ã§å¤§ãã（1/2 ãらã„ã¾ã§ï¼‰ä½Žä¸‹ã—ã¾ã™ï¼Žã“ã®è¾ºã‚Šï¼Œã»ã¼äºˆæƒ³é€šã‚Š*1ã®æŒ™å‹•ã§ã™ï¼Ž
====================================================================== Wikipedia-en 記事タイトル | Build | Search | Search* | Size [bytes] #keys=6,890,514 | | | | 136,297,337 ====================================================================== sorted keys Darts 0.32 | 14.283s | 0.527s | 6.694s | 552,547,920 darts-clone 0.32g | 6.575s | 0.625s | 5.946s | 258,735,104 darts-clone 0.32g v:1 | 15.917s | 0.764s | 6.227s | 95,080,448 *darts-clone 0.32e5 | 14.532s | 0.924s | 5.890s | 113,360,140 *darts-clone 0.32e5 v:1 | 13.996s | 0.852s | 5.562s | 87,122,880 darts-clone 0.32e5 huge| 4.128s | 0.554s | 5.920s | 147,491,712 *DASTrie 1.0 -c | NA | NA | NA | NA DASTrie 1.0 | 4.951s | 1.012s | 6.212s | 157,593,742 *doar 0.0.13 static | 8.684s | 0.672s | 5.664s | 123,000,249 -------------------------------------------------------------------- *doar 0.0.13 dynamic | +3.964s | 0.670s | 5.757s | +123,000,249 dda | 3.222s | 0.612s | 7.191s | 517,091,328 --------------------------------------------------------------------- map k:char* | 2.905s | 2.122s | 17.352s | ~515,663,184 map k:string | 10.482s | 6.041s | 25.479s | >515,663,184 unordered_map k:char* | 6.271s | 2.013s | 3.638s | ~460,539,072 unordered_map k:string | 7.303s | 2.182s | 4.540s | >460,539,072 -------------------------------------------------------------------- marisa 0.1.3 #tries=1 | 23.118s | 4.087s | 16.491s | 49,350,092 marisa 0.1.3 #tries=2 | 25.411s | 6.921s | 21.707s | 37,743,122 marisa 0.1.3 #tries=3 | 25.786s | 7.524s | 22.945s | 35,767,947 marisa 0.1.3 #tries=10| 26.206s | 7.841s | 23.633s | 34,802,056 tx 0.18 | 18.611s | 24.969s | 62.604s | 87,520,717 ux 0.1.2 | 46.515s | 45.921s | 67.670s | 49,134,003 ux 0.1.2 isTailUX=false| 16.744s | 30.197s | 49.810s | 105,423,225 ===================================================================== randomized keys doar 0.0.13 dynamic |+24.902s | 1.952s | 5.645s | +123,000,249 dda | 10.017s | 3.647s | 8.589s | 517,091,328 -------------------------------------------------------------------- map k:char* | 18.576s | 3.637s | 21.164s | ~515,663,184 map k:string | 27.897s | 8.417s | 30.713s | >515,663,184 unordered_map k:char* | 6.360s | 3.754s | 3.646s | ~460,539,072 unordered_map k:string | 7.302s | 4.605s | 4.549s | >460,539,072 -------------------------------------------------------------------- marisa 0.1.3 #tries=1 | 32.635s | 5.381s | 16.232s | 49,350,092 marisa 0.1.3 #tries=2 | 34.408s | 8.400s | 21.977s | 37,743,122 marisa 0.1.3 #tries=3 | 34.884s | 9.232s | 22.945s | 35,767,947 marisa 0.1.3 #tries=10| 35.405s | 9.556s | 23.660s | 34,802,056 tx 0.18 | 38.718s | 26.434s | 62.358s | 87,520,717 ux 0.1.2 | 63.228s | 47.239s | 68.075s | 49,134,003 ux 0.1.2 isTailUX=false| 33.159s | 31.547s | 50.461s | 105,423,225 ====================================================================== ====================================================================== 郵便番å·è¾žæ›¸ | Build | Search | Search* | Size [bytes] #keys=118,821 | | | | 950,568 ====================================================================== sorted keys Darts 0.32 | 0.039s | 0.003s | 0.012s | 2,156,576 darts-clone 0.32g | 0.038s | 0.003s | 0.010s | 1,069,056 darts-clone 0.32g v:1 | 0.020s | 0.003s | 0.007s | 106,496 *darts-clone 0.32e5 | 0.171s | 0.004s | 0.013s | 1,154,676 *darts-clone 0.32e5 v:1 | 0.165s | 0.004s | 0.011s | 650,116 darts-clone 0.32e5 huge| 0.141s | 0.004s | 0.013s | 1,363,656 *DASTrie 1.0 -c | 2.226s | 0.010s | 0.018s | 1,248,609 DASTrie 1.0 | 2.354s | 0.010s | 0.019s | 1,411,783 *doar 0.0.13 static | 0.064s | 0.004s | 0.013s | 1,204,561 -------------------------------------------------------------------- *doar 0.0.13 dynamic | +0.091s | 0.004s | 0.013s | +1,204,561 dda | 0.023s | 0.004s | 0.014s | 2,129,920 -------------------------------------------------------------------- map k:char* | 0.043s | 0.028s | 0.074s | ~7,604,544 map k:string | 0.137s | 0.072s | 0.147s | >7,604,544 unordered_map k:char* | 0.023s | 0.010s | 0.027s | ~6,653,976 unordered_map k:string | 0.039s | 0.013s | 0.038s | >6,653,976 -------------------------------------------------------------------- marisa 0.1.3 #tries=1 | 0.083s | 0.033s | 0.059s | 232,729 marisa 0.1.3 #tries=2 | 0.083s | 0.034s | 0.060s | 232,111 marisa 0.1.3 #tries=3 | 0.083s | 0.034s | 0.060s | 232,270 marisa 0.1.3 #tries=10| 0.083s | 0.034s | 0.060s | 233,404 tx 0.18 | 0.096s | 0.090s | 0.129s | 222,888 ux 0.1.2 | 0.092s | 0.158s | 0.187s | 220,209 ux 0.1.2 isTailUX=false| 0.092s | 0.157s | 0.187s | 222,165 ===================================================================== randomized keys *doar 0.0.13 dynamic | +0.104s | 0.009s | 0.013s | +1,204,561 dda | 0.036s | 0.010s | 0.013s | 2,129,920 -------------------------------------------------------------------- map k:char* | 0.068s | 0.035s | 0.085s | ~7,604,544 map k:string | 0.143s | 0.092s | 0.168s | >7,604,544 unordered_map k:char* | 0.029s | 0.023s | 0.027s | ~6,653,976 unordered_map k:string | 0.049s | 0.036s | 0.039s | >6,653,976 -------------------------------------------------------------------- marisa 0.1.3 #tries=1 | 0.121s | 0.036s | 0.059s | 232,729 marisa 0.1.3 #tries=2 | 0.121s | 0.037s | 0.060s | 232,111 marisa 0.1.3 #tries=3 | 0.121s | 0.037s | 0.060s | 232,270 marisa 0.1.3 #tries=10| 0.121s | 0.037s | 0.060s | 233,404 tx 0.18 | 0.168s | 0.097s | 0.129s | 222,888 ux 0.1.2 | 0.174s | 0.165s | 0.187s | 220,209 ux 0.1.2 isTailUX=false| 0.174s | 0.165s | 0.187s | 222,165 ===================================================================== (compiled with gcc 4.6.0 20110212, -O2 -march=core2 -m64)
*1:トライã®æ•°ã‚’一ã¤ã«ã—ã¦ï¼ŒTAIL 部ã®å‡¦ç†ã‚’（速度ã®é€Ÿã„）文å—列ã®ç…§åˆã®ã¿ã§æ¸ˆã¾ã›ã‚‹å ´åˆã«æ¤œç´¢é€Ÿåº¦ã¯æœ€é€Ÿã¨ãªã‚Šï¼ŒTAIL 部ã®ãƒˆãƒ©ã‚¤ã®æ•°ã‚’増やã›ã°å¢—ã‚„ã™ã»ã©ï¼ŒTAIL ã§æ–‡å—列照åˆã™ã‚‹éƒ¨åˆ†ãŒæ¸›ã£ã¦ã„ãã®ã§ï¼ŒTAIL ãªã—ã®å˜ç´”ãªãƒˆãƒ©ã‚¤ã®æ¤œç´¢é€Ÿåº¦ã«è¿‘ã¥ã．
