Knowledge

Freidlin–Wentzell theorem

Source 📝

228: 746: 612: 430: 94: 623: 492: 313: 223:{\displaystyle {\begin{cases}dX_{t}^{\varepsilon }=b(X_{t}^{\varepsilon })\,dt+{\sqrt {\varepsilon }}\,dB_{t},\\X_{0}^{\varepsilon }=0,\end{cases}}} 741:{\displaystyle \liminf _{\varepsilon \downarrow 0}{\big (}\varepsilon \log \mathbf {P} {\big }{\big )}\geq -\inf _{\omega \in G}I(\omega ).} 607:{\displaystyle \limsup _{\varepsilon \downarrow 0}{\big (}\varepsilon \log \mathbf {P} {\big }{\big )}\leq -\inf _{\omega \in F}I(\omega )} 830: 40:. Roughly speaking, the Freidlin–Wentzell theorem gives an estimate for the probability that a (scaled-down) sample path of an 799: 774: 769:. Grundlehren der Mathematischen Wissenschaften 260 (Second ed.). New York: Springer-Verlag. pp. xii+430. 840: 85: 835: 845: 794:. Applications of Mathematics (New York) 38 (Second ed.). New York: Springer-Verlag. pp. xvi+396. 825: 103: 425:{\displaystyle I(\omega )={\frac {1}{2}}\int _{0}^{T}|{\dot {\omega }}_{t}-b(\omega _{t})|^{2}\,dt} 33: 762: 49: 29: 809: 784: 249: 8: 37: 795: 770: 41: 806: 781: 53: 819: 758: 440: 274: 45: 25: 253: 234: 44:
will stray far from the mean path. This statement is made precise using
17: 473: 459: 292:
satisfies the large deviations principle with good rate function
757: 458:) = +∞ otherwise. In other words, for every 216: 626: 495: 316: 97: 789: 740: 606: 424: 222: 817: 708: 628: 577: 497: 696: 689: 666: 645: 565: 558: 535: 514: 48:. The Freidlin–Wentzell theorem generalizes 792:Large deviations techniques and applications 72:starting at the origin, 0 ∈  767:Random perturbations of dynamical systems 415: 168: 151: 307: ∪ {+∞} given by 818: 84:-valued Itō diffusion solving an Itō 790:Dembo, Amir; Zeitouni, Ofer (1998). 13: 14: 857: 831:Stochastic differential equations 68:be a standard Brownian motion on 660: 529: 86:stochastic differential equation 732: 726: 635: 601: 595: 504: 405: 400: 387: 358: 326: 320: 250:uniformly Lipschitz continuous 148: 130: 1: 751: 59: 7: 281:, the family of processes ( 10: 862: 22:Freidlin–Wentzell theorem 841:Large deviations theory 34:large deviations theory 836:Theorems in statistics 763:Wentzell, Alexander D. 742: 608: 426: 224: 743: 609: 427: 225: 32:) is a result in the 30:Alexander D. Wentzell 846:Probability theorems 624: 493: 314: 273:) equipped with the 95: 38:stochastic processes 826:Asymptotic analysis 479: ⊆  465: ⊆  356: 303: →  244: →  203: 147: 123: 738: 722: 642: 604: 591: 511: 422: 342: 220: 215: 189: 133: 109: 50:Schilder's theorem 812:(See chapter 5.6) 759:Freidlin, Mark I. 707: 627: 576: 496: 372: 340: 166: 853: 805: 780: 747: 745: 744: 739: 721: 700: 699: 693: 692: 680: 679: 670: 669: 663: 649: 648: 641: 613: 611: 610: 605: 590: 569: 568: 562: 561: 549: 548: 539: 538: 532: 518: 517: 510: 431: 429: 428: 423: 414: 413: 408: 399: 398: 380: 379: 374: 373: 365: 361: 355: 350: 341: 333: 252:. Then, on the 233:where the drift 229: 227: 226: 221: 219: 218: 202: 197: 181: 180: 167: 162: 146: 141: 122: 117: 861: 860: 856: 855: 854: 852: 851: 850: 816: 815: 802: 777: 754: 711: 695: 694: 688: 687: 675: 671: 665: 664: 659: 644: 643: 631: 625: 622: 621: 580: 564: 563: 557: 556: 544: 540: 534: 533: 528: 513: 512: 500: 494: 491: 490: 485: 471: 409: 404: 403: 394: 390: 375: 364: 363: 362: 357: 351: 346: 332: 315: 312: 311: 302: 291: 280: 268: 261: 214: 213: 198: 193: 186: 185: 176: 172: 161: 142: 137: 118: 113: 99: 98: 96: 93: 92: 62: 54:Brownian motion 12: 11: 5: 859: 849: 848: 843: 838: 833: 828: 814: 813: 800: 787: 775: 753: 750: 749: 748: 737: 734: 731: 728: 725: 720: 717: 714: 710: 706: 703: 698: 691: 686: 683: 678: 674: 668: 662: 658: 655: 652: 647: 640: 637: 634: 630: 629:lim inf 615: 614: 603: 600: 597: 594: 589: 586: 583: 579: 575: 572: 567: 560: 555: 552: 547: 543: 537: 531: 527: 524: 521: 516: 509: 506: 503: 499: 498:lim sup 483: 469: 433: 432: 421: 418: 412: 407: 402: 397: 393: 389: 386: 383: 378: 371: 368: 360: 354: 349: 345: 339: 336: 331: 328: 325: 322: 319: 300: 286: 278: 266: 259: 231: 230: 217: 212: 209: 206: 201: 196: 192: 188: 187: 184: 179: 175: 171: 165: 160: 157: 154: 150: 145: 140: 136: 132: 129: 126: 121: 116: 112: 108: 105: 104: 102: 61: 58: 46:rate functions 9: 6: 4: 3: 2: 858: 847: 844: 842: 839: 837: 834: 832: 829: 827: 824: 823: 821: 811: 808: 803: 801:0-387-98406-2 797: 793: 788: 786: 783: 778: 776:0-387-98362-7 772: 768: 764: 760: 756: 755: 735: 729: 723: 718: 715: 712: 704: 701: 684: 681: 676: 672: 656: 653: 650: 638: 632: 620: 619: 618: 598: 592: 587: 584: 581: 573: 570: 553: 550: 545: 541: 525: 522: 519: 507: 501: 489: 488: 487: 482: 478: 475: 468: 464: 461: 457: 453: 449: 445: 442: 441:Sobolev space 438: 419: 416: 410: 395: 391: 384: 381: 376: 369: 366: 352: 347: 343: 337: 334: 329: 323: 317: 310: 309: 308: 306: 299: 296: :  295: 289: 284: 276: 275:supremum norm 272: 265: 262: =  258: 255: 251: 247: 243: 240: :  239: 236: 210: 207: 204: 199: 194: 190: 182: 177: 173: 169: 163: 158: 155: 152: 143: 138: 134: 127: 124: 119: 114: 110: 106: 100: 91: 90: 89: 87: 83: 79: 75: 71: 67: 57: 55: 52:for standard 51: 47: 43: 42:Itō diffusion 39: 35: 31: 27: 26:Mark Freidlin 23: 19: 791: 766: 616: 480: 476: 466: 462: 455: 451: 447: 443: 439:lies in the 436: 434: 304: 297: 293: 287: 282: 270: 263: 256: 254:Banach space 245: 241: 237: 235:vector field 232: 88:of the form 81: 77: 73: 69: 65: 63: 21: 15: 277:||⋅|| 18:mathematics 820:Categories 752:References 474:closed set 472:and every 76:, and let 730:ω 716:∈ 713:ω 705:− 702:≥ 682:∈ 677:ε 657:⁡ 651:ε 636:↓ 633:ε 599:ω 585:∈ 582:ω 574:− 571:≤ 551:∈ 546:ε 526:⁡ 520:ε 505:↓ 502:ε 392:ω 382:− 370:˙ 367:ω 344:∫ 324:ω 200:ε 164:ε 144:ε 120:ε 60:Statement 765:(1998). 460:open set 446:(;  269:(;  24:(due to 810:1619036 785:1652127 450:), and 279:∞ 798:  773:  456:ω 437:ω 288:ε 80:be an 20:, the 290:>0 796:ISBN 771:ISBN 617:and 64:Let 28:and 709:inf 654:log 578:inf 523:log 435:if 248:is 36:of 16:In 822:: 807:MR 782:MR 761:; 486:, 56:. 804:. 779:. 736:. 733:) 727:( 724:I 719:G 697:) 690:] 685:G 673:X 667:[ 661:P 646:( 639:0 602:) 596:( 593:I 588:F 566:) 559:] 554:F 542:X 536:[ 530:P 515:( 508:0 484:0 481:C 477:F 470:0 467:C 463:G 454:( 452:I 448:R 444:H 420:t 417:d 411:2 406:| 401:) 396:t 388:( 385:b 377:t 359:| 353:T 348:0 338:2 335:1 330:= 327:) 321:( 318:I 305:R 301:0 298:C 294:I 285:) 283:X 271:R 267:0 264:C 260:0 257:C 246:R 242:R 238:b 211:, 208:0 205:= 195:0 191:X 183:, 178:t 174:B 170:d 159:+ 156:t 153:d 149:) 139:t 135:X 131:( 128:b 125:= 115:t 111:X 107:d 101:{ 82:R 78:X 74:R 70:R 66:B

Index

mathematics
Mark Freidlin
Alexander D. Wentzell
large deviations theory
stochastic processes
Itō diffusion
rate functions
Schilder's theorem
Brownian motion
stochastic differential equation
vector field
uniformly Lipschitz continuous
Banach space
supremum norm
Sobolev space
open set
closed set
Freidlin, Mark I.
Wentzell, Alexander D.
ISBN
0-387-98362-7
MR
1652127
ISBN
0-387-98406-2
MR
1619036
Categories
Asymptotic analysis
Stochastic differential equations

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