Knowledge

NL (complexity)

Source đź“ť

910:, according to which NL is closed under complements, the one-sided error in these probabilistic computations can be replaced by zero-sided error. That is, these problems can be solved by probabilistic Turing machines that use logarithmic space and never make errors. The corresponding complexity class that also requires the machine to use only polynomial time is called 747:
This was the strongest deterministic-space inclusion known in 1994 (Papadimitriou 1994 Problem 16.4.10, "Symmetric space"). Since larger space classes are not affected by quadratic increases, the nondeterministic and deterministic classes are known to be equal, so that for example we have
742: 949:
and Abuzer Yakaryılmaz have proven that the deterministic logarithmic-space Turing machine in the statement above can be replaced by a bounded-error probabilistic constant-space Turing machine that is allowed to use only a constant number of random bits.
415: 566: 942:
if and only if such a Turing machine accepts any word in the language for an appropriate choice of certificate in its additional input tape, and rejects any word not in the language regardless of the certificate.
820:, is not limited to polynomial time, because although it has a polynomial number of configurations it can use randomness to escape an infinite loop. If we do limit it to polynomial time, we get the class 62: 637: 163:
Important results in complexity theory allow us to relate this complexity class with other classes, telling us about the relative power of the resources involved. Results in the field of
631:, which tells us that any nondeterministic algorithm can be simulated by a deterministic machine in at most quadratically more space. From Savitch's theorem, we have directly that: 298: 592:, the class of problems solvable by randomized algorithms in logarithmic space and unbounded time, with no error. It is not, however, known or believed to be equal to 286: 903:
to take 2 steps on average before stopping. It only needs to keep a running total of the number of heads in a row it sees, which it can count in log space.
911: 600: 506: 167:, on the other hand, tell us which problems can be solved with this resource. Like much of complexity theory, many important questions about 68: 938:. Consider a deterministic logarithmic-space bounded Turing machine that has an additional read-only read-once input tape. A language is in 1145: 891:
The only problem is that we don't have room in log space for a binary counter that goes up to 2. To get around this we replace it with a
888:, and because there are 2 computation paths in all, we have a good chance of hitting the accepting one (bounded below by a constant). 1113: 1021: 176: 107: 907: 468: 737:{\displaystyle {\mathsf {NL\subseteq SPACE}}(\log ^{2}n)\ \ \ \ {\text{equivalently, }}{\mathsf {NL\subseteq L}}^{2}.} 27: 1507: 1104: 1074: 1138: 127: 99: 75: 979: 787:
that never accept incorrectly but are allowed to reject incorrectly less than 1/3 of the time; this is called
784: 268:
of two literals, if there is a variable assignment that makes the formula true. An example instance, where
123: 1542: 1131: 1523: 1330: 917:
Thus, when we only look at space, it seems that randomization and nondeterminism are equally powerful.
436: 410:{\displaystyle (x_{1}\vee \neg x_{3})\wedge (\neg x_{2}\vee x_{3})\wedge (\neg x_{1}\vee \neg x_{2})} 1512: 929: 476: 1087:(27 June 1997). "Sections 8.4–8.6: The Classes L and NL, NL-completeness, NL equals coNL". 460: 978:
The class NL is closed under the operations complementation, union, and therefore intersection,
1466: 145: 1096: 1089: 899:
coins and stops and rejects if they all land on heads. Since this event has probability 2, we
1461: 1456: 1451: 628: 271: 8: 261: 217: 967: 487: 1036:
A. C. Cem Say, Abuzer Yakaryılmaz, "Finite state verifiers with constant randomness,"
1471: 1100: 1070: 1017: 963: 780: 1435: 1325: 1257: 1247: 1154: 776: 594: 588: 440: 257: 225: 95: 91: 1430: 1420: 1377: 1367: 1360: 1350: 1340: 1298: 1293: 1288: 1272: 1252: 1230: 1225: 1220: 1203: 1198: 1193: 1011: 934: 822: 578: 496: 229: 221: 195: 1059: 1425: 1313: 1235: 1188: 1084: 1054: 900: 816: 431: 241: 118: 853:
If the string is not in the language, both reject along all computation paths.
480: 1536: 1003: 472: 15: 249: 172: 1303: 1213: 983: 561:{\displaystyle {\mathsf {NC_{1}\subseteq L\subseteq NL\subseteq NC_{2}}}} 265: 212: 1415: 1240: 1007: 884:, and execute this 2 times. Because no computation path exceeds length 1410: 1123: 864:
algorithm accepts along at least two-thirds of its computation paths.
164: 103: 1405: 1400: 1345: 1168: 1395: 946: 754: 1502: 1497: 1372: 1355: 750: 144:
can be formally defined in terms of the computational resource
194:
below; however, this name is more frequently used to refer to
1492: 1487: 1308: 860:
algorithm accepts along at least one computation path and a
1320: 1178: 1065:
Papadimitriou, C. (1994). "Chapter 16: Logarithmic Space".
826:, which is contained in but not known or believed to equal 500:
hierarchy. In Papadimitriou 1994, Theorem 16.1, we have:
1335: 1267: 1262: 1183: 1173: 880:
algorithm and choose a random computation path of length
962:: it contains precisely those languages expressible in 640: 509: 301: 274: 30: 126:. Since any deterministic Turing machine is also a 17: 1088: 833:There is a simple algorithm that establishes that 736: 560: 409: 280: 56: 1534: 57:{\displaystyle {\mathsf {L{\overset {?}{=}}NL}}} 1083: 958:There is a simple logical characterization of 1139: 1064: 191: 1114:Introduction to Complexity Theory: Lecture 7 69:(more unsolved problems in computer science) 766: 1146: 1132: 1002: 761: 1120:is what Goldreich calls badRSPACE(log n). 1091:Introduction to the Theory of Computation 953: 920: 1116:. Oded Goldreich. Proposition 6.1. Our 814:, unlike its deterministic counterpart 205: 122:, the class for logspace problems on a 1535: 1153: 783:solvable in logarithmithic space with 720: 714: 711: 664: 661: 658: 655: 652: 646: 643: 604:, the polynomial-time restrictions of 551: 547: 543: 537: 534: 528: 520: 516: 512: 49: 46: 41: 38: 33: 1127: 973: 928:can equivalently be characterised by 856:If the string is in the language, an 791:. The constant 1/3 is arbitrary; any 177:Unsolved problems in computer science 1013:Complexity Theory: A Modern Approach 264:formula of which each clause is the 18:Unsolved problem in computer science 1038:Logical Methods in Computer Science 471:) was independently discovered by 13: 391: 375: 340: 318: 275: 14: 1554: 1508:Probabilistically checkable proof 612:, which some authors refer to as 210:Several problems are known to be 479:in 1987; they received the 1995 459:is the class of languages whose 932:, analogous to classes such as 420: 128:nondeterministic Turing machine 100:nondeterministic Turing machine 76:computational complexity theory 1030: 1016:. Cambridge University Press. 996: 688: 669: 443:, but it is not known whether 404: 372: 366: 337: 331: 302: 198:, which is not known to equal 1: 1047: 1040:, Vol. 10(3:6)2014, pp. 1-17. 908:Immerman–SzelepcsĂ©nyi theorem 785:probabilistic Turing machines 627:to deterministic space using 469:Immerman–SzelepcsĂ©nyi theorem 895:counter, which simply flips 196:randomized logarithmic space 124:deterministic Turing machine 7: 1095:. PWS Publishing. pp.  1010:(2009). "Definition 4.19". 10: 1559: 1524:List of complexity classes 1521: 1480: 1444: 1388: 1281: 1161: 494:can be placed within the 437:polynomial-time algorithm 90:ogarithmic-space) is the 1513:Interactive proof system 1067:Computational Complexity 989: 799:< 1/2 would suffice. 767:Probabilistic definition 192:probabilistic definition 98:that can be solved by a 762:Alternative definitions 116:is a generalization of 1467:Arithmetical hierarchy 954:Descriptive complexity 921:Certificate definition 738: 562: 411: 282: 146:nondeterministic space 58: 1462:Grzegorczyk hierarchy 1457:Exponential hierarchy 1389:Considered infeasible 739: 563: 412: 283: 281:{\displaystyle \neg } 59: 1452:Polynomial hierarchy 1282:Suspected infeasible 876:, we simply take an 638: 507: 299: 272: 218:log-space reductions 206:NL-complete problems 28: 1481:Families of classes 1162:Considered feasible 705:equivalently,  582:. It is known that 477:RĂłbert SzelepcsĂ©nyi 467:. This result (the 451:. It is known that 435:, since there is a 1543:Complexity classes 1155:Complexity classes 1069:. Addison-Wesley. 974:Closure properties 968:transitive closure 802:It turns out that 734: 558: 488:circuit complexity 407: 278: 186:is referred to as 54: 1530: 1529: 1472:Boolean hierarchy 1445:Class hierarchies 1023:978-0-521-42426-4 964:first-order logic 781:decision problems 706: 702: 699: 696: 693: 629:Savitch's theorem 425:It is known that 96:decision problems 44: 1550: 1148: 1141: 1134: 1125: 1124: 1110: 1094: 1080: 1041: 1034: 1028: 1027: 1000: 872:is contained in 845:is contained in 777:complexity class 757: 743: 741: 740: 735: 730: 729: 724: 723: 707: 704: 700: 697: 694: 691: 681: 680: 668: 667: 626: 619: 615: 611: 607: 603: 597: 591: 585: 581: 576:is contained in 575: 572:More precisely, 567: 565: 564: 559: 557: 556: 555: 554: 524: 523: 499: 493: 466: 458: 454: 450: 446: 441:2-satisfiability 434: 429:is contained in 428: 416: 414: 413: 408: 403: 402: 387: 386: 365: 364: 352: 351: 330: 329: 314: 313: 287: 285: 284: 279: 258:2-satisfiability 232:asks, for nodes 226:2-satisfiability 134:is contained in 92:complexity class 86:ondeterministic 65: 63: 61: 60: 55: 53: 52: 45: 37: 19: 1558: 1557: 1553: 1552: 1551: 1549: 1548: 1547: 1533: 1532: 1531: 1526: 1517: 1476: 1440: 1384: 1378:PSPACE-complete 1277: 1157: 1152: 1107: 1077: 1050: 1045: 1044: 1035: 1031: 1024: 1001: 997: 992: 976: 956: 923: 906:Because of the 789:one-sided error 769: 764: 749: 725: 710: 709: 708: 703: 676: 672: 642: 641: 639: 636: 635: 624: 617: 613: 609: 605: 599: 593: 587: 583: 577: 573: 550: 546: 519: 515: 511: 510: 508: 505: 504: 495: 491: 483:for this work. 464: 456: 452: 448: 444: 430: 426: 423: 398: 394: 382: 378: 360: 356: 347: 343: 325: 321: 309: 305: 300: 297: 296: 273: 270: 269: 230:ST-connectivity 222:ST-connectivity 208: 148:(or NSPACE) as 130:, we have that 72: 71: 66: 36: 32: 31: 29: 26: 25: 23: 21: 12: 11: 5: 1556: 1546: 1545: 1528: 1527: 1522: 1519: 1518: 1516: 1515: 1510: 1505: 1500: 1495: 1490: 1484: 1482: 1478: 1477: 1475: 1474: 1469: 1464: 1459: 1454: 1448: 1446: 1442: 1441: 1439: 1438: 1433: 1428: 1423: 1418: 1413: 1408: 1403: 1398: 1392: 1390: 1386: 1385: 1383: 1382: 1381: 1380: 1370: 1365: 1364: 1363: 1353: 1348: 1343: 1338: 1333: 1328: 1323: 1318: 1317: 1316: 1314:co-NP-complete 1311: 1306: 1301: 1291: 1285: 1283: 1279: 1278: 1276: 1275: 1270: 1265: 1260: 1255: 1250: 1245: 1244: 1243: 1233: 1228: 1223: 1218: 1217: 1216: 1206: 1201: 1196: 1191: 1186: 1181: 1176: 1171: 1165: 1163: 1159: 1158: 1151: 1150: 1143: 1136: 1128: 1122: 1121: 1111: 1105: 1085:Michael Sipser 1081: 1075: 1062: 1055:Complexity Zoo 1049: 1046: 1043: 1042: 1029: 1022: 1004:Arora, Sanjeev 994: 993: 991: 988: 975: 972: 966:with an added 955: 952: 922: 919: 866: 865: 854: 810:. Notice that 768: 765: 763: 760: 745: 744: 733: 728: 722: 719: 716: 713: 690: 687: 684: 679: 675: 671: 666: 663: 660: 657: 654: 651: 648: 645: 623:We can relate 570: 569: 553: 549: 545: 542: 539: 536: 533: 530: 527: 522: 518: 514: 422: 419: 418: 417: 406: 401: 397: 393: 390: 385: 381: 377: 374: 371: 368: 363: 359: 355: 350: 346: 342: 339: 336: 333: 328: 324: 320: 317: 312: 308: 304: 277: 260:asks, given a 242:directed graph 207: 204: 67: 51: 48: 43: 40: 35: 22: 16: 9: 6: 4: 3: 2: 1555: 1544: 1541: 1540: 1538: 1525: 1520: 1514: 1511: 1509: 1506: 1504: 1501: 1499: 1496: 1494: 1491: 1489: 1486: 1485: 1483: 1479: 1473: 1470: 1468: 1465: 1463: 1460: 1458: 1455: 1453: 1450: 1449: 1447: 1443: 1437: 1434: 1432: 1429: 1427: 1424: 1422: 1419: 1417: 1414: 1412: 1409: 1407: 1404: 1402: 1399: 1397: 1394: 1393: 1391: 1387: 1379: 1376: 1375: 1374: 1371: 1369: 1366: 1362: 1359: 1358: 1357: 1354: 1352: 1349: 1347: 1344: 1342: 1339: 1337: 1334: 1332: 1329: 1327: 1324: 1322: 1319: 1315: 1312: 1310: 1307: 1305: 1302: 1300: 1297: 1296: 1295: 1292: 1290: 1287: 1286: 1284: 1280: 1274: 1271: 1269: 1266: 1264: 1261: 1259: 1256: 1254: 1251: 1249: 1246: 1242: 1239: 1238: 1237: 1234: 1232: 1229: 1227: 1224: 1222: 1219: 1215: 1212: 1211: 1210: 1207: 1205: 1202: 1200: 1197: 1195: 1192: 1190: 1187: 1185: 1182: 1180: 1177: 1175: 1172: 1170: 1167: 1166: 1164: 1160: 1156: 1149: 1144: 1142: 1137: 1135: 1130: 1129: 1126: 1119: 1115: 1112: 1108: 1106:0-534-94728-X 1102: 1098: 1093: 1092: 1086: 1082: 1078: 1076:0-201-53082-1 1072: 1068: 1063: 1061: 1057: 1056: 1052: 1051: 1039: 1033: 1025: 1019: 1015: 1014: 1009: 1005: 999: 995: 987: 985: 981: 980:concatenation 971: 969: 965: 961: 951: 948: 944: 941: 937: 936: 931: 927: 918: 915: 913: 909: 904: 902: 898: 894: 889: 887: 883: 879: 875: 871: 868:To show that 863: 859: 855: 852: 851: 850: 848: 844: 840: 836: 831: 829: 825: 824: 819: 818: 813: 809: 805: 800: 798: 794: 790: 786: 782: 778: 774: 759: 756: 752: 731: 726: 717: 685: 682: 677: 673: 649: 634: 633: 632: 630: 621: 602: 596: 590: 580: 540: 531: 525: 503: 502: 501: 498: 489: 484: 482: 478: 474: 473:Neil Immerman 470: 462: 442: 438: 433: 399: 395: 388: 383: 379: 369: 361: 357: 353: 348: 344: 334: 326: 322: 315: 310: 306: 295: 294: 293: 291: 267: 263: 262:propositional 259: 255: 251: 247: 243: 239: 235: 231: 227: 223: 219: 215: 214: 203: 201: 197: 193: 189: 185: 182:Occasionally 180: 178: 174: 170: 166: 161: 159: 155: 151: 147: 143: 139: 137: 133: 129: 125: 121: 120: 115: 111: 109: 105: 101: 97: 93: 89: 85: 81: 77: 70: 1208: 1117: 1090: 1066: 1053: 1037: 1032: 1012: 998: 977: 959: 957: 945: 939: 933: 930:certificates 925: 924: 916: 905: 896: 892: 890: 885: 881: 877: 873: 869: 867: 861: 857: 846: 842: 838: 834: 832: 827: 821: 815: 811: 807: 803: 801: 796: 792: 788: 772: 770: 746: 622: 586:is equal to 571: 485: 424: 421:Containments 292:, might be: 289: 253: 245: 237: 233: 220:, including 211: 209: 199: 187: 183: 181: 168: 162: 157: 153: 149: 141: 140: 135: 131: 117: 113: 112: 108:memory space 87: 83: 79: 73: 1361:#P-complete 1299:NP-complete 1214:NL-complete 1008:Barak, Boaz 984:Kleene star 481:Gödel Prize 461:complements 447:or whether 266:disjunction 213:NL-complete 190:due to its 104:logarithmic 94:containing 1416:ELEMENTARY 1241:P-complete 1048:References 970:operator. 893:randomized 841:. Clearly 453:NL = co-NL 288:indicates 244:, whether 171:are still 165:algorithms 106:amount of 1411:2-EXPTIME 849:, since: 795:with 0 ≤ 718:⊆ 683:⁡ 650:⊆ 541:⊆ 532:⊆ 526:⊆ 392:¬ 389:∨ 376:¬ 370:∧ 354:∨ 341:¬ 335:∧ 319:¬ 316:∨ 276:¬ 250:reachable 1537:Category 1406:EXPSPACE 1401:NEXPTIME 1169:DLOGTIME 771:Suppose 455:, where 102:using a 1396:EXPTIME 1304:NP-hard 1097:294–302 947:Cem Say 775:is the 755:NPSPACE 463:are in 64:⁠ 24:⁠ 1503:NSPACE 1498:DSPACE 1373:PSPACE 1103:  1073:  1020:  982:, and 901:expect 751:PSPACE 701:  698:  695:  692:  449:L = NL 445:NL = P 216:under 154:NSPACE 1493:NTIME 1488:DTIME 1309:co-NP 990:Notes 457:co-NL 252:from 240:in a 175:(see 156:(log 1321:TFNP 1101:ISBN 1071:ISBN 1018:ISBN 912:ZPLP 616:and 608:and 601:ZPLP 475:and 439:for 236:and 224:and 173:open 1436:ALL 1336:QMA 1326:FNP 1268:APX 1263:BQP 1258:BPP 1248:ZPP 1179:ACC 779:of 674:log 618:ZPL 610:ZPL 598:or 595:RLP 589:ZPL 486:In 290:not 248:is 228:. 179:). 160:). 74:In 1539:: 1431:RE 1421:PR 1368:IP 1356:#P 1351:PP 1346:⊕P 1341:PH 1331:AM 1294:NP 1289:UP 1273:FP 1253:RP 1231:CC 1226:SC 1221:NC 1209:NL 1204:FL 1199:RL 1194:SL 1184:TC 1174:AC 1099:. 1060:NL 1058:: 1006:; 986:. 960:NL 940:NL 935:NP 926:NL 914:. 878:NL 870:NL 858:NL 847:NL 839:NL 837:= 830:. 828:NL 823:RL 808:NL 806:= 758:. 753:= 625:NL 620:. 614:RL 606:RL 584:NL 579:AC 574:NL 497:NC 492:NL 490:, 465:NL 427:NL 256:. 202:. 200:NL 188:RL 184:NL 169:NL 152:= 150:NL 142:NL 138:. 136:NL 114:NL 110:. 80:NL 78:, 1426:R 1236:P 1189:L 1147:e 1140:t 1133:v 1118:C 1109:. 1079:. 1026:. 897:n 886:n 882:n 874:C 862:C 843:C 835:C 817:L 812:C 804:C 797:x 793:x 773:C 732:. 727:2 721:L 715:L 712:N 689:) 686:n 678:2 670:( 665:E 662:C 659:A 656:P 653:S 647:L 644:N 568:. 552:2 548:C 544:N 538:L 535:N 529:L 521:1 517:C 513:N 432:P 405:) 400:2 396:x 384:1 380:x 373:( 367:) 362:3 358:x 349:2 345:x 338:( 332:) 327:3 323:x 311:1 307:x 303:( 254:S 246:T 238:T 234:S 158:n 132:L 119:L 88:L 84:N 82:( 50:L 47:N 42:? 39:= 34:L 20::

Index

(more unsolved problems in computer science)
computational complexity theory
complexity class
decision problems
nondeterministic Turing machine
logarithmic
memory space
L
deterministic Turing machine
nondeterministic Turing machine
nondeterministic space
algorithms
open
Unsolved problems in computer science
probabilistic definition
randomized logarithmic space
NL-complete
log-space reductions
ST-connectivity
2-satisfiability
ST-connectivity
directed graph
reachable
2-satisfiability
propositional
disjunction
P
polynomial-time algorithm
2-satisfiability
complements

Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.

↑