Knowledge

Gelman-Rubin statistic

Source đź“ť

22: 620: 461: 347: 260: 716: 468: 168: 354: 644: 96: 267: 734: 176: 652: 615:{\displaystyle W={\frac {1}{J}}\sum _{j=1}^{J}\left({\frac {1}{L-1}}\sum _{i=1}^{L}(x_{i}^{(j)}-{\overline {x}}_{j})^{2}\right)} 99: 43: 105: 737:
compares whether the mean of the first x percent of a chain and the mean of the last y percent of a chain match.
98:
Monte Carlo simulations (chains) are started with different initial values. The samples from the respective
892: 778:
Gelman, Andrew; Rubin, Donald B. (1992). "Inference from Iterative Simulation Using Multiple Sequences".
170:(of the j-th simulation), the variance between the chains and the variance in the chains is estimated: 887: 829: 68: 456:{\displaystyle B={\frac {L}{J-1}}\sum _{j=1}^{J}({\overline {x}}_{j}-{\overline {x}}_{*})^{2}} 787: 8: 791: 854: 803: 756: 629: 81: 845:
Vats, Dootika; Knudson, Christina (2021). "Revisiting the Gelman–Rubin Diagnostic".
747:
Vats, Dootika; Knudson, Christina (2021). "Revisiting the Gelman–Rubin Diagnostic".
342:{\displaystyle {\overline {x}}_{*}={\frac {1}{J}}\sum _{j=1}^{J}{\overline {x}}_{j}} 864: 795: 766: 881: 799: 255:{\displaystyle {\overline {x}}_{j}={\frac {1}{L}}\sum _{i=1}^{L}x_{i}^{(j)}} 807: 868: 770: 859: 761: 41:
parameter to this template to explain the issue with the article.
831:
7.4 Monitoring Convergence | Advanced Statistical Computing
711:{\displaystyle R={\frac {{\frac {L-1}{L}}W+{\frac {1}{L}}B}{W}}} 622:
Averaged variances of the individual chains across all chains
722:
When L tends to infinity and B tends to zero, R tends to 1.
655: 632: 471: 357: 270: 179: 108: 84: 725:
A different formula is given by Vats & Knudson.
710: 638: 614: 455: 341: 254: 162: 90: 879: 163:{\displaystyle x_{1}^{(j)},\dots ,x_{L}^{(j)}} 31:needs attention from an expert in Mathematics 67:allows a statement about the convergence of 844: 777: 746: 626:An estimate of the Gelman-Rubin statistic 858: 760: 880: 46:may be able to help recruit an expert. 15: 463:Variance of the means of the chains 13: 14: 904: 827: 102:are discarded. From the samples 20: 728: 349:Mean of the means of all chains 838: 821: 598: 572: 566: 553: 444: 403: 247: 241: 155: 149: 125: 119: 1: 815: 740: 74: 586: 432: 412: 328: 277: 186: 7: 10: 909: 834:– via bookdown.org. 69:Monte Carlo simulations 44:WikiProject Mathematics 712: 640: 616: 552: 508: 457: 402: 343: 321: 256: 230: 164: 92: 65:Gelman-Rubin statistic 800:10.1214/ss/1177011136 713: 641: 617: 532: 488: 458: 382: 344: 301: 262:Mean value of chain j 257: 210: 165: 93: 653: 630: 469: 355: 268: 177: 106: 82: 893:Monte Carlo methods 847:Statistical Science 792:1992StaSc...7..457G 780:Statistical Science 749:Statistical Science 576: 251: 159: 129: 708: 636: 612: 556: 453: 339: 252: 231: 160: 139: 109: 88: 888:Estimation theory 869:10.1214/20-STS812 771:10.1214/20-STS812 735:Geweke Diagnostic 706: 697: 681: 646:then results as 639:{\displaystyle R} 589: 530: 486: 435: 415: 380: 331: 299: 280: 208: 189: 91:{\displaystyle J} 61: 60: 900: 873: 872: 862: 842: 836: 835: 825: 811: 774: 764: 717: 715: 714: 709: 707: 702: 698: 690: 682: 677: 666: 663: 645: 643: 642: 637: 621: 619: 618: 613: 611: 607: 606: 605: 596: 595: 590: 582: 575: 564: 551: 546: 531: 529: 515: 507: 502: 487: 479: 462: 460: 459: 454: 452: 451: 442: 441: 436: 428: 422: 421: 416: 408: 401: 396: 381: 379: 365: 348: 346: 345: 340: 338: 337: 332: 324: 320: 315: 300: 292: 287: 286: 281: 273: 261: 259: 258: 253: 250: 239: 229: 224: 209: 201: 196: 195: 190: 182: 169: 167: 166: 161: 158: 147: 128: 117: 97: 95: 94: 89: 56: 53: 47: 33:. Please add a 24: 23: 16: 908: 907: 903: 902: 901: 899: 898: 897: 878: 877: 876: 843: 839: 828:Peng, Roger D. 826: 822: 818: 743: 731: 689: 667: 665: 664: 662: 654: 651: 650: 631: 628: 627: 601: 597: 591: 581: 580: 565: 560: 547: 536: 519: 514: 513: 509: 503: 492: 478: 470: 467: 466: 447: 443: 437: 427: 426: 417: 407: 406: 397: 386: 369: 364: 356: 353: 352: 333: 323: 322: 316: 305: 291: 282: 272: 271: 269: 266: 265: 240: 235: 225: 214: 200: 191: 181: 180: 178: 175: 174: 148: 143: 118: 113: 107: 104: 103: 83: 80: 79: 77: 57: 51: 48: 42: 25: 21: 12: 11: 5: 906: 896: 895: 890: 875: 874: 837: 819: 817: 814: 813: 812: 786:(4): 457–472. 775: 742: 739: 730: 727: 720: 719: 705: 701: 696: 693: 688: 685: 680: 676: 673: 670: 661: 658: 635: 624: 623: 610: 604: 600: 594: 588: 585: 579: 574: 571: 568: 563: 559: 555: 550: 545: 542: 539: 535: 528: 525: 522: 518: 512: 506: 501: 498: 495: 491: 485: 482: 477: 474: 464: 450: 446: 440: 434: 431: 425: 420: 414: 411: 405: 400: 395: 392: 389: 385: 378: 375: 372: 368: 363: 360: 350: 336: 330: 327: 319: 314: 311: 308: 304: 298: 295: 290: 285: 279: 276: 263: 249: 246: 243: 238: 234: 228: 223: 220: 217: 213: 207: 204: 199: 194: 188: 185: 157: 154: 151: 146: 142: 138: 135: 132: 127: 124: 121: 116: 112: 100:burn-in phases 87: 76: 73: 59: 58: 28: 26: 19: 9: 6: 4: 3: 2: 905: 894: 891: 889: 886: 885: 883: 870: 866: 861: 856: 852: 848: 841: 833: 832: 824: 820: 809: 805: 801: 797: 793: 789: 785: 781: 776: 772: 768: 763: 758: 754: 750: 745: 744: 738: 736: 726: 723: 703: 699: 694: 691: 686: 683: 678: 674: 671: 668: 659: 656: 649: 648: 647: 633: 608: 602: 592: 583: 577: 569: 561: 557: 548: 543: 540: 537: 533: 526: 523: 520: 516: 510: 504: 499: 496: 493: 489: 483: 480: 475: 472: 465: 448: 438: 429: 423: 418: 409: 398: 393: 390: 387: 383: 376: 373: 370: 366: 361: 358: 351: 334: 325: 317: 312: 309: 306: 302: 296: 293: 288: 283: 274: 264: 244: 236: 232: 226: 221: 218: 215: 211: 205: 202: 197: 192: 183: 173: 172: 171: 152: 144: 140: 136: 133: 130: 122: 114: 110: 101: 85: 72: 70: 66: 55: 45: 40: 36: 32: 29:This article 27: 18: 17: 850: 846: 840: 830: 823: 783: 779: 752: 748: 732: 729:Alternatives 724: 721: 625: 78: 64: 62: 52:January 2024 49: 38: 34: 30: 882:Categories 860:1812.09384 816:References 762:1812.09384 741:Literature 75:Definition 672:− 587:¯ 578:− 534:∑ 524:− 490:∑ 439:∗ 433:¯ 424:− 413:¯ 384:∑ 374:− 329:¯ 303:∑ 284:∗ 278:¯ 212:∑ 187:¯ 134:… 808:2246093 788:Bibcode 806:  35:reason 855:arXiv 853:(4). 804:JSTOR 757:arXiv 755:(4). 37:or a 733:The 63:The 39:talk 865:doi 796:doi 767:doi 884:: 863:. 851:36 849:. 802:. 794:. 782:. 765:. 753:36 751:. 71:. 871:. 867:: 857:: 810:. 798:: 790:: 784:7 773:. 769:: 759:: 718:. 704:W 700:B 695:L 692:1 687:+ 684:W 679:L 675:1 669:L 660:= 657:R 634:R 609:) 603:2 599:) 593:j 584:x 573:) 570:j 567:( 562:i 558:x 554:( 549:L 544:1 541:= 538:i 527:1 521:L 517:1 511:( 505:J 500:1 497:= 494:j 484:J 481:1 476:= 473:W 449:2 445:) 430:x 419:j 410:x 404:( 399:J 394:1 391:= 388:j 377:1 371:J 367:L 362:= 359:B 335:j 326:x 318:J 313:1 310:= 307:j 297:J 294:1 289:= 275:x 248:) 245:j 242:( 237:i 233:x 227:L 222:1 219:= 216:i 206:L 203:1 198:= 193:j 184:x 156:) 153:j 150:( 145:L 141:x 137:, 131:, 126:) 123:j 120:( 115:1 111:x 86:J 54:) 50:(

Index

WikiProject Mathematics
Monte Carlo simulations
burn-in phases
Geweke Diagnostic
arXiv
1812.09384
doi
10.1214/20-STS812
Bibcode
1992StaSc...7..457G
doi
10.1214/ss/1177011136
JSTOR
2246093
7.4 Monitoring Convergence | Advanced Statistical Computing
arXiv
1812.09384
doi
10.1214/20-STS812
Categories
Estimation theory
Monte Carlo methods

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

↑