1: \begin{abstract}

2:   In this work we marry multi-index Monte Carlo with ensemble Kalman

3:   filtering (EnKF) to produce the multi-index EnKF method

4:   (MIEnKF). The MIEnKF method is based on independent samples of

5:   four-coupled EnKF estimators on a multi-index hierarchy of

6:   resolution levels, and it may be viewed as an extension of the

7:   multilevel EnKF (MLEnKF) method developed by the same authors in

8:   2020.  Multi-index here refers to a two-index method, consisting of

9:   a hierarchy of EnKF estimators that are coupled in two degrees of

10:   freedom: time discretization and ensemble size.  Under certain

11:   assumptions, the MIEnKF method is proven to be more tractable than

12:   EnKF and MLEnKF, and this is also verified in numerical examples.

13:   %% In this work, we combined the multi-index Monte Carlo method and

14:   %% ensemble Kalman filtering (EnKF) to develop the multi-index EnKF (MIEnKF) method. This method is based on independent samples of

15:   %% four-coupled EnKF estimators on a multi-index hierarchy of

16:   %% resolution levels, and it may be regarded as an extension of the

17:   %% multilevel EnKF (MLEnKF) method previously developed by the

18:   %% same authors. The MIEnKF method consists of a

19:   %% hierarchy of EnKF estimators coupled in two degrees of

20:   %% freedom: time discretization and ensemble size.

21:   %% Under certain assumptions, the MIEnKF method was proven to be

22:   %% more tractable than EnKF and MLEnKF; this tractability was also verified

23:   %% using numerical examples. 

24:   \bigskip

25:   \noindent

26:   

27:   \textbf{Key words}: Monte Carlo, multilevel, multi-index, convergence

28:   rates, Kalman filter, ensemble Kalman filter

29:   %   \bigskip

30:   

31:   \noindent \textbf{AMS subject classification}: 65C30, 65Y20. 

32:   

33: \end{abstract}

34: