Rough McKean-Vlasov dynamics for robust ensemble Kalman filtering

Motivated by the challenge of incorporating data into misspecified and multiscale dynamical models, we study a McKean-Vlasov equation that contains the data stream as a common driving rough path. This setting allows us to prove well-posedness as well as continuity with respect to the driver in an appropriate rough-path topology. The latter property is key in our subsequent development of a robust data assimilation methodology: We establish propagation of chaos for the associated interacting particle system, which in turn is suggestive of a numerical scheme that can be viewed as an extension of the ensemble Kalman filter to a rough-path framework. Finally, we discuss a data-driven method based on subsampling to construct suitable rough path lifts and demonstrate the robustness of our scheme in a number of numerical experiments related to parameter estimation problems in multiscale contexts.

[1]  D. Crisan,et al.  Approximate McKean–Vlasov representations for a class of SPDEs , 2005, math/0510668.

[2]  Terry Lyons,et al.  Physical Brownian motion in a magnetic field as a rough path , 2015, Transactions of the American Mathematical Society.

[3]  Qin Li,et al.  Ensemble Kalman inversion: mean-field limit and convergence analysis , 2019, Statistics and Computing.

[4]  Gabriel Stoltz,et al.  Partial differential equations and stochastic methods in molecular dynamics* , 2016, Acta Numerica.

[5]  Pathwise nonlinear filtering with correlated noise , 2009 .

[6]  Samuel N. Cohen,et al.  Pathwise stochastic control with applications to robust filtering , 2019, The Annals of Applied Probability.

[7]  Terry Lyons,et al.  Evolving communities with individual preferences , 2013, 1303.4243.

[8]  Grigorios A. Pavliotis,et al.  Multiscale Methods: Averaging and Homogenization , 2008 .

[9]  T. Kurtz,et al.  Particle representations for a class of nonlinear SPDEs , 1999 .

[10]  Sebastian Reich,et al.  Data assimilation: The Schrödinger perspective , 2018, Acta Numerica.

[11]  Andreas Neuenkirch,et al.  A Milstein-type scheme without Lévy area terms for SDEs driven by fractional Brownian motion , 2010, 1001.3344.

[12]  Shirley Dex,et al.  JR 旅客販売総合システム（マルス）における運用及び管理について , 1991 .

[13]  Ryne Beeson,et al.  Reduced Order Nonlinear Filters for Multi-Scale Systems with Correlated Sensor Noise , 2018, 2018 21st International Conference on Information Fusion (FUSION).

[14]  F. Flandoli,et al.  Propagation of chaos for interacting particles subject to environmental noise. , 2014, 1403.1981.

[15]  W. Stannat,et al.  Mean field limit of Ensemble Square Root filters - discrete and continuous time , 2020, Foundations of Data Science.

[16]  J. Diehl,et al.  Pathwise stability of likelihood estimators for diffusions via rough paths , 2013, 1311.1061.

[17]  G. A. Pavliotis,et al.  Maximum likelihood drift estimation for multiscale diffusions , 2008, 0806.3248.

[18]  R. Carmona,et al.  Probabilistic Theory of Mean Field Games with Applications II: Mean Field Games with Common Noise and Master Equations , 2018 .

[19]  A. Shiryayev,et al.  Statistics of Random Processes Ii: Applications , 2000 .

[20]  N. Namachchivaya,et al.  Optimal nudging in particle filters , 2014 .

[21]  P. Imkeller,et al.  Dimensional reduction in nonlinear filtering: A homogenization approach , 2011, 1112.2986.

[22]  Theresa Lange,et al.  On the continuous time limit of the ensemble Kalman filter , 2019, Math. Comput..

[23]  G. Evensen,et al.  Analysis Scheme in the Ensemble Kalman Filter , 1998 .

[24]  Mark H. A. Davis,et al.  Pathwise nonlinear filtering for nondegenerate diffusions with noise correlation , 1987 .

[25]  L. Szpruch,et al.  Weak existence and uniqueness for McKean–Vlasov SDEs with common noise , 2019, The Annals of Probability.

[26]  Grigorios A. Pavliotis,et al.  Parameter estimation for multiscale diffusions : an overview , 2012 .

[27]  Terry Lyons,et al.  Discretely sampled signals and the rough Hoff process , 2013, 1310.4054.

[28]  Zakai equation of nonlinear filtering with unbounded coefficients: The case of dependent noises , 1993 .

[29]  Mark H. A. Davis.  On a multiplicative functional transformation arising in nonlinear filtering theory , 1980 .

[30]  Sean P. Meyn,et al.  Feedback Particle Filter , 2013, IEEE Transactions on Automatic Control.

[31]  Rough nonlocal diffusions. , 2019, 1905.07270.

[32]  Sean P. Meyn,et al.  Diffusion Map-based Algorithm for Gain Function Approximation in the Feedback Particle Filter , 2020, SIAM/ASA J. Uncertain. Quantification.

[33]  C. Chou.  The Vlasov equations , 1965 .

[34]  Andrew L. Allan.  Robust filtering and propagation of uncertainty in hidden Markov models , 2020, 2005.04982.

[35]  Andrey Kormilitzin,et al.  A Primer on the Signature Method in Machine Learning , 2016, ArXiv.

[36]  I. Bailleul,et al.  Propagation of chaos for mean field rough differential equations , 2019, 1907.00578.

[37]  J. M. Clark.  The Design of Robust Approximations to the Stochastic Differential Equations of Nonlinear Filtering , 1978 .

[38]  Nikolas Nüsken,et al.  State and Parameter Estimation from Observed Signal Increments , 2019, Entropy.

[39]  P. Imkeller,et al.  A homogenization approach to multiscale filtering , 2012 .

[40]  Qin Li,et al.  Ensemble Kalman Inversion for nonlinear problems: weights, consistency, and variance bounds , 2020, ArXiv.

[41]  Brian Jefferies.  Feynman-Kac Formulae , 1996 .

[42]  Tim Sauer,et al.  Correlation between System and Observation Errors in Data Assimilation , 2018, Monthly Weather Review.

[43]  Mark H. A. Davis.  A Pathwise Solution of the Equations of Nonlinear Filtering , 1982 .

[44]  Zhou Zhou,et al.  “A Tale of Two Time Scales: Determining Integrated Volatility with Noisy High-Frequency Data” , 2005 .

[45]  Geir Evensen,et al.  The Ensemble Kalman Filter: theoretical formulation and practical implementation , 2003 .

[46]  Pierre Del Moral,et al.  On the stability and the uniform propagation of chaos of Extended Ensemble Kalman-Bucy filters , 2016 .

[47]  A. Duncan,et al.  On the geometry of Stein variational gradient descent , 2019, ArXiv.

[48]  G. A. Pavliotis,et al.  Parameter Estimation for Multiscale Diffusions , 2007 .

[49]  Sebastian Reich,et al.  Affine-Invariant Ensemble Transform Methods for Logistic Regression , 2021, Foundations of Computational Mathematics.

[50]  G. Pavliotis,et al.  Data-driven coarse graining in action: Modeling and prediction of complex systems. , 2015, Physical review. E, Statistical, nonlinear, and soft matter physics.

[51]  Siragan Gailus,et al.  Discrete-Time Statistical Inference for Multiscale Diffusions , 2018, Multiscale Model. Simul..

[52]  Ryne T. Beeson,et al.  Approximation of the Filter Equation for Multiple Timescale, Correlated, Nonlinear Systems , 2020, SIAM J. Math. Anal..

[53]  A. Stuart,et al.  Well-posedness and accuracy of the ensemble Kalman filter in discrete and continuous time , 2013, 1310.3167.

[54]  Grigorios A. Pavliotis,et al.  Frequency Domain Estimation of Integrated Volatility for Itô Processes in the Presence of Market-Microstructure Noise , 2009, Multiscale Model. Simul..

[55]  Robert Azencott,et al.  Adaptive Sub-sampling for Parametric Estimation of Gaussian Diffusions , 2010 .

[56]  A. Lejay,et al.  Semi-martingales and rough paths theory , 2005 .

[57]  A. Stuart,et al.  Extracting macroscopic dynamics: model problems and algorithms , 2004 .

[58]  J. R. Maddison,et al.  Bayesian inference of ocean diffusivity from Lagrangian trajectory data , 2018, Ocean Modelling.

[59]  Terry Lyons,et al.  On the Importance of the Levy Area for Studying the Limits of Functions of Converging Stochastic Processes. Application to Homogenization , 2003 .

[60]  Grigorios A. Pavliotis,et al.  A new framework for extracting coarse-grained models from time series with multiscale structure , 2014, J. Comput. Phys..

[61]  Martin Hairer,et al.  A Course on Rough Paths , 2020, Universitext.

[62]  Ali M. Mosammam.  The Oxford handbook of nonlinear filtering , 2012 .

[63]  Stochastic partial differential equations: a rough paths view on weak solutions via Feynman–Kac , 2017 .

[64]  Peter K. Friz,et al.  Multidimensional Stochastic Processes as Rough Paths: Theory and Applications , 2010 .

[65]  P. Cochat,et al.  Et al , 2008, Archives de pediatrie : organe officiel de la Societe francaise de pediatrie.

[66]  Wilhelm Stannat,et al.  McKean-Vlasov SDEs in nonlinear filtering , 2021, SIAM J. Control. Optim..

[67]  Wilhelm Stannat,et al.  Analysis of the feedback particle filter with diffusion map based approximation of the gain , 2021, Foundations of Data Science.

[68]  Dan Crisan,et al.  On a robust version of the integral representation formula of nonlinear filtering , 2005 .

[69]  D. Crisan,et al.  Fundamentals of Stochastic Filtering , 2008 .

[70]  Peter K. Friz,et al.  Rough stochastic differential equations , 2021 .

[71]  A. Guillin,et al.  On the rate of convergence in Wasserstein distance of the empirical measure , 2013, 1312.2128.

[72]  I. Bailleul,et al.  Solving mean field rough differential equations , 2018 .

[73]  Terry Lyons,et al.  Learning from the past, predicting the statistics for the future, learning an evolving system , 2013, 1309.0260.

[74]  Grigorios A. Pavliotis,et al.  Drift Estimation of Multiscale Diffusions Based on Filtered Data , 2021, Foundations of Computational Mathematics.

[75]  Colin J. Cotter,et al.  Probabilistic Forecasting and Bayesian Data Assimilation , 2015 .

[76]  Wilhelm Stannat,et al.  Long-Time Stability and Accuracy of the Ensemble Kalman-Bucy Filter for Fully Observed Processes and Small Measurement Noise , 2016, SIAM J. Appl. Dyn. Syst..

[77]  Grigorios A. Pavliotis,et al.  Semiparametric Drift and Diffusion Estimation for Multiscale Diffusions , 2013, Multiscale Model. Simul..

[78]  Rupert Klein,et al.  Scale-Dependent Models for Atmospheric Flows , 2010 .

[79]  Yacine Ait-Sahalia,et al.  How Often to Sample a Continuous-Time Process in the Presence of Market Microstructure Noise , 2003 .

[80]  J. Deuschel,et al.  Pathwise McKean–Vlasov theory with additive noise , 2020 .

[81]  Sebastian Reich.  Frequentist perspective on robust parameter estimation using the ensemble Kalman filter , 2022, ArXiv.

[82]  Xin Tong,et al.  Analysis of a localised nonlinear ensemble Kalman Bucy filter with complete and accurate observations , 2019, Nonlinearity.

[83]  H. McKean,et al.  A CLASS OF MARKOV PROCESSES ASSOCIATED WITH NONLINEAR PARABOLIC EQUATIONS , 1966, Proceedings of the National Academy of Sciences of the United States of America.

[84]  D. Crisan,et al.  Robust filtering: Correlated noise and multidimensional observation , 2012, 1201.1858.

[85]  A. Doucet,et al.  A Tutorial on Particle Filtering and Smoothing: Fifteen years later , 2008 .

[86]  Quantitative Convergence of the Filter Solution for Multiple Timescale Nonlinear Systems with Coarse-Grain Correlated Noise , 2020, 2011.12801.

[87]  Peter W. Sauer,et al.  Dynamic Data-Driven Adaptive Observations in Data Assimilation for Multi-scale Systems , 2018, Handbook of Dynamic Data Driven Applications Systems.

[88]  An energy method for rough partial differential equations , 2017, Journal of Differential Equations.

[89]  E. Lorenz.  Deterministic nonperiodic flow , 1963 .

[90]  Gail C. Murphy,et al.  Learning from the past , 2003, 10th Working Conference on Reverse Engineering, 2003. WCRE 2003. Proceedings..

[91]  Nikolas Nüsken,et al.  Stein Variational Gradient Descent: many-particle and long-time asymptotics , 2021, Foundations of Data Science.

[92]  Hiroshi Tanaka.  Limit Theorems for Certain Diffusion Processes with Interaction , 1984 .

[93]  Hui-Yu Tsai,et al.  Lorenz Equations 之研究 , 1998 .

[94]  Mark Holland,et al.  Central limit theorems and invariance principles for Lorenz attractors , 2007 .

[95]  Amirhossein Taghvaei,et al.  Kalman Filter and its Modern Extensions for the Continuous-time Nonlinear Filtering Problem , 2017, ArXiv.

[96]  A. Sznitman.  Topics in propagation of chaos , 1991 .

[97]  Ryne Beeson,et al.  Particle Filters with Nudging in Multiscale Chaotic Systems: With Application to the Lorenz ’96 Atmospheric Model , 2020, Journal of Nonlinear Science.

[98]  B. Gess,et al.  Stochastic nonlinear Fokker–Planck equations , 2019, Nonlinear Analysis.

[99]  Theresa Lange,et al.  On the continuous time limit of ensemble square root filters , 2021, Communications in Mathematical Sciences.

[100]  Statistical Properties for Flows with Unbounded Roof Function, Including the Lorenz Attractor , 2018, Journal of Statistical Physics.

[101]  Lai-Sang Young,et al.  What Are SRB Measures, and Which Dynamical Systems Have Them? , 2002 .

[102]  S. Shreve,et al.  Stochastic differential equations , 1955, Mathematical Proceedings of the Cambridge Philosophical Society.

[103]  A. M. Davie,et al.  Differential Equations Driven by Rough Paths: An Approach via Discrete Approximation , 2007, 0710.0772.

[104]  Harold J. Kushner,et al.  A Robust Discrete State Approximation to the Optimal Nonlinear Filter for a Diffusion. , 1980 .

[105]  Sebastian Reich,et al.  Supervised learning from noisy observations: Combining machine-learning techniques with data assimilation , 2020, ArXiv.

[106]  Dan Crisan,et al.  Pathwise approximations for the solution of the non-linear filtering problem , 2021, ArXiv.

[107]  I. Gyöngy.  On the approximation of stochastic partial differential equations II , 1988 .

[108]  N Sri Namachchivaya,et al.  Particle filtering in high-dimensional chaotic systems. , 2012, Chaos.

[109]  T. Kurtz,et al.  Numerical Solutions for a Class of SPDEs with Application to Filtering , 2001 .

[110]  G. A. Pavliotis,et al.  Multiscale modelling and inverse problems , 2010, 1009.2943.

[111]  P. Moral,et al.  On the Mathematical Theory of Ensemble (Linear-Gaussian) Kalman-Bucy Filtering , 2020, 2006.08843.

[112]  Siragan Gailus,et al.  Statistical Inference for Perturbed Multiscale Dynamical Systems , 2015 .

[113]  Peter K. Friz,et al.  Multiscale Systems, Homogenization, and Rough Paths , 2016, Probability and Analysis in Interacting Physical Systems.

[114]  D. Simon.  Optimal State Estimation: Kalman, H Infinity, and Nonlinear Approaches , 2006 .

[115]  C. J. Cotter,et al.  Estimating eddy diffusivities from noisy Lagrangian observations , 2009, 0904.4817.

[116]  G. Evensen.  Data Assimilation: The Ensemble Kalman Filter , 2006 .

[117]  Ismaël F. Bailleul,et al.  The Inverse Problem for Rough Controlled Differential Equations , 2015, SIAM J. Control. Optim..

[118]  Georg A. Gottwald,et al.  Stochastic Model Reduction for Slow-Fast Systems with Moderate Time Scale Separation , 2019, Multiscale Model. Simul..

[119]  Peter C. Kiessler,et al.  Statistical Inference for Ergodic Diffusion Processes , 2006 .

[120]  Gorjan Alagic,et al.  #p , 2019, Quantum information & computation.

[121]  Sebastian Riedel,et al.  Convergence rates for the full Gaussian rough paths , 2011 .

[122]  Robert Azencott,et al.  SUB-SAMPLING AND PARAMETRIC ESTIMATION FOR MULTISCALE DYNAMICS ∗ , 2013 .