, 17(r))r.to,4 est à trajectoires continues et à accroissements indépendants, Leprocessus

?. Pour-tous-f,

I. Le-sous-espace, Qà(V) muni du produit scalaire 1u,a )o:( Q-Èu,Q-*u >, est un espace de Hilbert et l'espace L3: Lz(Vo,11) des opérateurs de Hilbert- Schmidt de Ve dans un espace de Hilbert .Ff est, muni de la norme llvll?s : llvgà ll2 : ?r(ùgv

. Un-espace-de-hilbert-séparable,

, L!, nous définissons les normes /ftr llldlll, : (n lo" llô,ll',sl,)u, t e fo,rl et nous notons N?r(0,7;LB) l'espace de Hilbert des processus, p.1

L. Probleme-etudie,

F. Soit, U. , and V. Trois-espaces-de-hilbert-séparables, Wt(t))re1o,4 un Qtprocessus de Wiener à valeurs lans U et (W2(t))te1o,rj un Q2-processus de

, Wiener à valeurs dans I/, où Çr et Qz sont des opéïateurs positifs, symétriques et traçables définis sur U et V respectivement De plus, nous supposons que les processus de Wiener (lfr(t))r.to, (t))teto,rl sont indépendants, pp.1-2

. Soit, 4 le processus stochastique à valeurs dans I/ x V, solution du problème de filtrage non linéaire : (5. 1) y: Io' G(x")d,s +w2(t) ou: (H1) Xs est donné dans, I

, H2) L'application, p.0

I. /-est-lipschitzienne and . Bornée,

G. H3-)-l-'application and . Est-continue, bornée et admet des dérivées du premier et deuxième ordre continues et bornées, notées G' et G" respectivement . De plus, nous supposons que pour tout r ? .F/, la fonction G" (r) est dans L@3; V) (l'espace des fonctions linéaires de .[! dans V) et I'application iL F+ G

R. H4-)-l-'opérateur,

, Afin de définir le filtre associé au problème de filtrage non linéaire (5.1), rappelons Ia notation suivante

I. Pour-tout-k-e and . Uct, H,lR) I'espace linéaire des applications $ de H dans IR admettant des dérivées de F!échet jusqu'à l'ordre ,k, bornées et uniformément continues

A. , et pour toute fonction $ de UCI nous défrnissons le filtre zr1 associé au système (5.1) par : rt(ù: Etô6ùl!

, où )r : o{Yr,O ( s < t} est la tribu engendrée par les trajectoires de I'observation jusqu'au temps f

L. Probabilite and D. Reference,

. Dans-ce-paragraphe, nous utilisons la méthode de "la probabilité de référence" pour définir une nouvelle probabitité sous laquelle le processus (Y1)x?[0

. Soit, r1 le processus stochastique défini pour tout t ? [0 dy, )v -; lr'llc

D. Cette-définition, par utilisation de la formule de Kallianpur-Striebel, on en déduit que les processus n1(S) et rt, pp.1-1

, En tenant compte de cette dernière remarque, nous pouvons démontrer le résultat principal de ce chapitre

, Ies appli'cati'ons y à ot(A,) ety r-+ rt(A,) sont conti,nues deC(l},fl,V) dans M+(H)' l'espace des mesures

. De-ia-conaergence-Étroi-'te,

. De-plus, Q. Pour-toute-fonction, and U. Ir-),

. Et-y-t-+-rt, A, ô) sont localement lipschi'tzi'ennes de C(l},Tl,V) dans IF

. Preuve, Soient (An)n>o une suite de C([0, T1,V) convergeant vers 3r' et

, 5, il nous suffit de prouver que la suite (o1(gr, >o converge verc o{9

D. , Définition 5.4.4, nous obtenons : ot(un,ô):E(ô(xr) zr(a^)): [ ô6r) z{y,)d,P

L. Or and . Suite, Zr(A^))^>s converge presque sûrement vers le processus ô(Xr)Zr(a) et, il existe une constante positive K telle que ld(Xr)Zr(A)l 3 K presque sûrernent. Donc, d'après le théorème de Lebesgue, nous avons : ,ITt lrofx,l z,(a*) d,P : trotx,t z{v) d,P c'est-à-dire : 'IT-ot(an' Ô) : o'

I. Ainsi,

, Prouvons maintenant la deuxième partie du Théorème 5.4.5, c'est à dire que I'application y r-+ ot(y,ô) est lipschitzienne

$. La-fonction and . Étant-bornée, il existe une constante positive C1 telle que ll,owazao, pp.5-7

C. Pour-tous-y,-y-'dans,

D. , A) : exp(%(g)). nous montrons facilement qu'il existe une constante positive C2 telle que : lzr(a) -zr

, avons: tv(a) -v(a')t 3 ti f ,".'rr'"1'-oor!1,' ,',ir,l"r^*")d,w1(s) >v I rt */. | < s(s) -a'G),G'(x,)F(x") >u ldt *; l,' | < s, G"(x") (*r*"lqÏ) (oro,oÏ)

. De, R(x")d,wr(s) tu l) _.^ * est une (Ft)tep,rlmartingale locale, nous obtenons, en prenant l'espérance J;Inn la probabilité P dans les deux membres de cette inégalité : *; E (1,' t.s(s) -u'(s)

. Finalement, comme toutes les fonctions apparaissant dans l'expression précédente sont bornées, I'inégalité de Cauchy-Schwarz entraîne que lrtut l -v(a')ld,P < crlla -u'll* où C3 est une constante positive

, 7) et (5.8) nous avon lrr(a, ô) -or(a' , ô)l < cr cz lrtu@ -v@')ldP er, en utilisant la majoration obrenue pour lrt\t l -V(a)ldP dans (b.9), nous en déduisons qu'il existe une constante positive C (C : CtCzCù telle que lor(a, ô) -or(y', ùl s c lla -a

, ce qui implique que I'application U è ot

. De, de manière analogue qu'en dimension finie, etre o1(!,1)-1 est borné, nous déduisons de l'inégalité (5.10) que l'applicatiorL A â rt(A,ô) est localement, IR. tr L04 Bibliographie [1] N.U.Ahmed, Nonlinear filtering for stochastic differential equations on Hilbert spaces. Actes de 14th IMACS World Congress, Georgia Tech, pp.5-8

N. U. Ahmed, M. Fuhrman, and J. Zabczyk, On Filtering Equations in Infinite Dimensions, Journal of Functional Analysis, vol.143, issue.1, pp.180-204, 1997.
DOI : 10.1006/jfan.1996.2970

URL : https://doi.org/10.1006/jfan.1996.2970

N. U. Ahmed and J. Zabczyk, Nonlinear filtering for semilinear stochastic differential equations on Hilbert spaces, Preprint 522 of the Inst'itute of Mathemati,cs of the Poli,sh Academy of Sci,enees, 1994.

J. Baras, G. Blankenship, and W. Hopkins, Existence, uniqueness and asymp totic behaviour of solutions of a class of. Zakai equation with unbounded coefficients, IEEE Tfansacti,ons on Automati, pp.28-203, 1983.

V. E. Benés, Exact finite dimensional filters for certain diffusions with nonlinear drift. ,Stochastics and Stochast, ics Reports, vol.5, pp.65-92, 1981.

A. Bensoussan, Filtrage optimal des systèmes linéaires, Dunod, issue.1e71

A. Bensoussan, On some approximation techniques in nonlinear filtering Stochastic Differenti,al Systems, Stochastic Control and Appli'cati,ons, 1986.

A. Bensoussan, R. Glowinski, and A. Rascanu, Approximation of Zaku equation by the splitting-up method, Stochastic Systems and Optimization Lecture Notes in Control and Infomnat'ion Sci, p.136
DOI : 10.1007/bfb0002686

A. Bhatt, G. Kallianpur, and R. L. Karandikar, Uniqueness and robustness of solution of measure valuqd equations of nonlinear filtering, p.Chapel Hill, 1994.

J. M. Bismut, D. Michel, and D. Conditionnel\es, onal Analysi,s Part I44 (198i) L74-2I1, Part II 4 (L982), pp.274-292

V. Bogachev, M. Rôckner, and B. Schmuland, Generalized Mehler semigroups and applications. Probabiti,ty Theory and, Related Fi, pp.93-225, 1996.

P. Cannarsa and V. Vespri, Existence and uniqueness of solutions to a class of stochastic partial differential equations. Stochasti,c Analgs'is and, pp.315-339, 1985.

M. Chaleyat-maurel and D. Michel, Hypoeliipticity theorems and conditional laws. Z ei,t. W ahr, Venn. G eb, vol.65, pp.573-597, 1984.

M. Chaleyat-maurel and D. Michel, Des résultats de non existence de filtre de dimension frnie, Stochastics anil Stochastics Reports, vol.3, pp.83-102, 1984.

M. Chaleyat-maurel and D. Michel, Une propriete de continuite en filtrage non lineaire, Stochastics, vol.13, issue.1-2, pp.11-40, 1986.
DOI : 10.1080/17442508408833320

J. M. Clark, The Design of Robust Approximations to the Stochastic Differential Equations of Nonlinear Filtering, Communi ,cati,on Systems and, Ranilom Process Theory Sijthoff and Noordhoof (1e78)
DOI : 10.1007/978-94-011-7577-7_41

R. F. Curtain and A. J. Pritchard, Infinite dimensional linear systems theory. Lecture Notes in Control and Informat'ion Sciences

G. Da-prato and J. Zabczyk, Stochastic equations in infrnite dimensions. Encyclopedia of Mathematics and its Applicati,ons 44, 1992.

G. Da-prato and J. Zabczyk, Regular Densities of Invariant Measures in Hilbert Spaces, Journal of Functional Analysis, vol.130, issue.2, pp.427-449, 1995.
DOI : 10.1006/jfan.1995.1076

M. H. Davis, Pathwise nonlinear filtering, Stochastic Systems the Mathematiis of Filtering and Identification and Applications, 1981.

M. , H. A. Davis, and S. I. Marcus, An introduction to nonlinear filtering, Stochastic Systems the Mathematics of Filtering and ldentification and Applications

M. H. Davis and M. Spathopoulos, Pathwise Nonlinear Filtering for Nondegenerate Diffusions with Noise Correlation, SIAM Journal on Control and Optimization, vol.25, issue.2, pp.260-278, 1987.
DOI : 10.1137/0325016

G. , D. Masi, W. , and R. , Continuous-time approximations for the nonlinear filtering problem, Mathemati,cs and, Opti'mizati'on7, pp.233-245, 1931.

G. , D. Masi, M. Pratelli, and W. Runggaldier, An approximation for the nonlinear filtering problem with error bound, Stochast'ics and Stochasti, pp.247-271, 1985.

M. N. Dietsch, Continuity with respect to the trajectory of the observation of the filter associated to systems in infinite dimension s. Proceed,i,ngs Ôf Prague Stochasti, pp.103-106, 1998.

M. N. Dietsch, Filtering of a pa"rtially observed process in the case of a high signal-to-noise ratio for correlated systems, Stochastic Analysi,s and Applicatiorzs, vol.18, pp.3-347, 2000.

M. N. Dietsch, A continuity property for the filter associated to Hilbertspace valued systems, Systems and Control Letters, vol.40, pp.3-165, 2000.

M. N. Dietsch, Existence of a density for the filter associated to Hilbertspace valued systems, Proceedi,ngs of the 39th IEEE Conference on Decision and Control, 2000.

M. N. Dietsch, Filtering equations for nonlinear filtering in infinite dimensions

M. and N. Dietsch, Nonlinear filtering for Hilbert-space valued systems

R. Elliott and R. Glowinski, Approximations to solutions of the Zal<aifiltering equation.'Stochasti,c Analys'is and Appli, pp.145-168, 1988.

R. J. Elliott and M. Kohlmann, Robust filtering for correlated multidimensional observations, Mathematische Zeitschrift, vol.178, issue.4, pp.78-559, 1981.
DOI : 10.1007/BF01174775

R. J. Elliott and J. B. Moore, Zakai equation for Hilbert spâce valued processes. Stochasti,c Analysi,s and, Appli,catiene, pp.597-605, 1998.

G. S. Ferreyra, SmoothneSs of the unnormalized conditional rneasures of stochastic nonlinear filtering, Proceedings of 23rd IEEE Conference on Deci,si,on and Control, Las Vegas (NV), 1984.

W. H. Flerning, D. Ji, and E. Pardoux, Piecewise linear filtering with small ob' servation noise Proceedings 8th Internati,onal Conference on Analys,is and Opti,mi,zati,on of Systems, Lecture Notes in Control and Information Sciences, pp.725-739, 1988.

W. H. Fleming, D. Ji-;-p, Q. Salame, and . Zhang, Discrete time piecewise linear filtering with small observation noise. Brown University Providence RI 02912, Division of Applied Mathematics, pp.88-115, 1988.

W. H. Fleming and S. K. Mitter, Optimal control and pathwise nonlinear filtering of non degenerate diffusions, Stochast'ics and Stochastics Reports, pp.63-77

W. H. Fleming and E. Pardoux, Piecewise monotone filtering with Small observation noise, SIAM Journal on Control and Opti, pp.27-115, 1989.

P. Florchinger, Continuité pax rapport à la trajectoire de l'observation du filtre associé à des systèmes corrélés à coefficients de l'observation non bornés. Stochasti,cs anil, Stochastics Reports, vol.28, pp.2-64, 1989.

]. P. Florchinger, Malliavin calculus with time depending coefficients and application to nonlinear filtering. Probability Theory and Related Fi, pp.86-203, 1990.

P. Florchinger, Zakas equation of nonlinear filtering with unbounded coefficients . The case of dependent noises. Systems and Control Letters 2I (1ee3), pp.413-422
DOI : 10.1016/0167-6911(93)90097-p

P. Florchinger, Nonlinear filtering in infinite dimensional spaces. Proceed,ings of the 37th IEEE Conference on Dec'is'i'on and Control
DOI : 10.1109/cdc.1997.649669

P. Florchinger, F. Le, and . Gland, Time-discretization of the zakai equation for diffusion processes observed in correlated noise, Stochastics and Stochastic Reports, vol.11, issue.3, pp.233-256, 1991.
DOI : 10.1007/BF00536382

URL : https://hal.archives-ouvertes.fr/hal-00912048

M. F\rhrman, Analyticity of transition semigroups and closability of bilinear forms in Hilbert spaces, pp.5-1995

M. Fujisaki, G. Kallianpur, and H. Kunita, Stochastic differential equations for the nonlinear filtering problem, Osaka Journal of Mathematics I, issue.L972, pp.19-40

, Gegout-Petit, Filtrage d'un processus partiellement observé et équations différentielles stochastiques rétrogrades réfléchies, Thèse, 1995.

A. Gegout-petit, Approximate Filter for the Conditional Law of a Partially Observed Process in Nonlinear Filtering, SIAM Journal on Control and Optimization, vol.36, issue.4, pp.1423-1447, 1998.
DOI : 10.1137/S0363012995287040

W. and H. , Nonlinear filtering of non degenerate diffusions with unbounded coefficients, Thèse

K. Ito, Approximation of the Zakai equation for nonlinear filtering. SIAM ' Joumal on Control and Opti,mi,zati, pp.34-620, 1996.

K. Ito and B. Rozovkii, Approximation of the Kushner equation for nonlinear filtering. SIAM Joumal an Control and Opti,mizati, pp.3-893, 2000.

G. Kallianpur, Stochastic filtering theory, 1980.

R. Kalman, A:new approach to linear filtering and prediction problems, J. Basi,c Ens ASME, vol.82, pp.33-45, 1960.

R. Kalman and R. Bucy, New Results in Linear Filtering and Prediction Theory, Journal of Basic Engineering, vol.83, issue.1, pp.95-108, 1961.
DOI : 10.1115/1.3658902

]. R. Katzur, B. Bobrovski, and Z. Schuss, Asymptotic Analysis of the Optimal Filtering Problem for One-Dimensional Diffusions Measured in a Low Noise Channel, Part I, SIAM Journal on Applied Mathematics, vol.44, issue.3, pp.591-604, 1984.
DOI : 10.1137/0144041

H. Korezlioglu and G. Mazziotto, Approximations of the nonlinear filter by periodic sampling and quantization Analysis and Optimization of Systems Lecture Notes i,n Control and, pp.62-553, 1984.

N. Krylov and B. Rozovsky, ON THE CAUCHY PROBLEM FOR LINEAR STOCHASTIC PARTIAL DIFFERENTIAL EQUATIONS, Mathematics of the USSR-Izvestiya, vol.11, issue.6, pp.1267-1284, 1977.
DOI : 10.1070/IM1977v011n06ABEH001768

H. Kunita, Densities of a measure???valued process governed by a stochastic partial differential equation, Systems & Control Letters, vol.1, issue.2, pp.100-104, 1981.
DOI : 10.1016/S0167-6911(81)80044-8

H. Kunita, Stochastic flows and stochastic difierential equations, 1990.

H. J. Kushner, Dynamical equations for optimal nonlinear filtering, Journal of Differential Equations, vol.3, issue.2, pp.179-190, 1967.
DOI : 10.1016/0022-0396(67)90023-X

H. J. Kushner, Filtering for Linear Distributed Parameter Systems, SIAM Journal on Control, vol.8, issue.3, pp.346-359, 1970.
DOI : 10.1137/0308024

H. J. Kushner, Probability methods for approximations in stochastic control and for elliptic equations, 1977.

F. and L. Gland, Time discretization of nonlinear filtering equations. Proceedi ,ngs of the2Sth IEEE Conference on Deci,s'ion and Control, pp.260-2606, 1994.
URL : https://hal.archives-ouvertes.fr/hal-00912070

J. Levine, Finite Dimensional Filters for a Class of Nonlinear Systems and Immersion in a Linear System, SIAM Journal on Control and Optimization, vol.25, issue.6, pp.25-1430, 1987.
DOI : 10.1137/0325079

R. S. Tipster and A. N. Shiryayev, Statistics of random processes, 1978.

S. Lototsky, R. Mikulevicius, and B. Rozovskii, Nonlinear Filtering Revisited: A Spectral Approach, SIAM Journal on Control and Optimization, vol.35, issue.2
DOI : 10.1137/S0363012993248918

URL : http://www.usc.edu/dept/LAS/CAMS/usr/facmemb/boris/preprints/lot02.ps

Z. M. Ma and M. Rôckner, Introduction to the theory of non-symmetric Dirichlet forms, 1992.
DOI : 10.1007/978-3-642-77739-4

M. Métivier and S. D. Gruyter, , 1974.

D. Michel, R??gularit?? des lois conditionnelles en th??orie du filtrage non-lin??aire et calcul des variations stochastique, Journal of Functional Analysis, vol.41, issue.1, pp.3-36, 1981.
DOI : 10.1016/0022-1236(81)90059-8

URL : https://doi.org/10.1016/0022-1236(81)90059-8

P. Milheiro and . Oliveira, Etudes asymptotiques en filtrage non linéaire avec petit bruit d'observation. Thèse, 1990.

J. Y. Ouward, Martingales locales et théorème de Girsanov dans les espaces de Hilbert réels séparables Annales de l'Institut Henri Poincaré B e4 (1973) 351-368

E. Pardoux, Equations aux dérivées partielles stochastiques non linéaires monotones, Thèse, 1975.

E. Pardoux, Stochastic partial differential equatiôns and filtering of diffusion processes. ^9tochasti,cs and Stochasti, cs Repor-tt, vol.3, issue.2, pp.127-167, 1979.

E. Pardoux, ??quations du filtrage non lin??aire de la pr??diction et du lissage, Stochastics, vol.11, issue.3-4, pp.193-231, 1982.
DOI : 10.1007/BF00536382

E. Pa, rdoux, Filtrage non linéaire et équations aux dérivées partielles stochastiques associées Ecole d'Eté de Probabilités de Saint-Flour XIX, Lectures Notes, Mathemati,cs L464, pp.67-163, 1991.

J. Picard, Filtrage de diffusions vectorielles faiblement bruitées Analysis and Optimisation of Systems, Lecture Notes i, Control and Informati,on Sci, vol.83, 1986.

J. Picard, Nonlinear Filtering of One-Dimensional Diffusions in the Case of a High Signal-to-Noise Ratio, SIAM Journal on Applied Mathematics, vol.46, issue.6, pp.46-1098, 1986.
DOI : 10.1137/0146065

URL : https://hal.archives-ouvertes.fr/inria-00076154

J. Picard, Asymptotic study of estimation problems with small observation noise. Stochasti,c Modelli,ng and Filtering, Lecture Notes in Control and Information Sciences, vol.91, 1984.

J. Picard, Efficiency of the Extended Kalman Filter for Nonlinear Systems with Small Noise, SIAM Journal on Applied Mathematics, vol.51, issue.3, pp.843-885, 1991.
DOI : 10.1137/0151042

H. J. Sussmann, On the Gap Between Deterministic and Stochastic Ordinary Differential Equations, The Annals of Probability, vol.6, issue.1, pp.6-978
DOI : 10.1214/aop/1176995608

H. J. Sussmann, Rigorous Results on the Cubic Sensor Problem, Stochastic Systems the Mathematics of Filtering and Identification and Applications, 1981.
DOI : 10.1007/978-94-009-8546-9_36

L. ,

H. J. Sussmann, Les équations différentielles du filtrage non linéaire.,Outils et Mod,èles Mathémati,ques pour I'Automat'ique, l'Analyse des Systèmes et le Tfaitement du Si,gnal S (1983) 639-648, On the gap between deterministic and stochastic ordinary differential equations. Annals of Probabi,li, pp.6-19, 1978.

W. S. Wong, On a new class of finite dimensional estimation algebras, Systems & Control Letters, vol.9, issue.1, pp.78-83, 1987.
DOI : 10.1016/0167-6911(87)90012-0

I. Yaesh, B. Z. Bobrovsky, and Z. Schuss, Asymptotic analysis of the optimal filtering problem for two dimensional diffusions measured in a low noise chanel, SIAM Journal of Appli, pp.50-1134, 1990.

K. Yoshida and F. Analysis, , 1965.

M. Zal<ai, On the optimal filtering of diffusion processes. Zei,t

. Geb and . Tl, , pp.230-243

O. Zeitouni and A. Dembo, On the maximal achievable accuracy in nonlinear filtering problems, IEEE Transactions on Automatic Control, vol.33, issue.10, pp.965-967, 1988.
DOI : 10.1109/9.7256