Open access peer-reviewed chapter

On Optimal and Simultaneous Stochastic Perturbations with Application to Estimation of High-Dimensional Matrix and Data Assimilation in High-Dimensional Systems

By Hong Son Hoang and Remy Baraille

Submitted: November 21st 2017Reviewed: April 16th 2018Published: November 5th 2018

DOI: 10.5772/intechopen.77273

Downloaded: 354


This chapter is devoted to different types of optimal perturbations (OP), deterministic, stochastic, OP in an invariant subspace, and simultaneous stochastic perturbations (SSP). The definitions of OPs are given. It will be shown how the OPs are important for the study on the predictability of behavior of system dynamics, generating ensemble forecasts as well as in the design of a stable filter. A variety of algorithm-based SSP methodology for estimation and decomposition of very high-dimensional (Hd) matrices are presented. Numerical experiments will be presented to illustrate the efficiency and benefice of the perturbation technique.


  • predictability
  • optimal perturbation
  • invariant subspace
  • simultaneous stochastic perturbation
  • dynamical system
  • filter stability
  • estimation of high-dimensional matrix

1. Introduction

Study in high-dimensional systems (HdS) today constitutes one of the most important research subjects thanks to the exponential increase of the power and speed of computers: after Moore’s law, the number of transistors in a dense integrated circuit doubles approximately every 2 years (see Myhrvold [1]). However, this exponential increase is still far from being sufficient for responding to great demand on computational and memory resources in implementing the optimal data assimilation algorithms (like Kalman filter (KF) [2], for example) for operational forecasting systems (OFS).

This chapter is devoted to the role of perturbations as an efficient tool for predictability of dynamical system, ensemble forecasting and for overcoming the difficulties in the design of data assimilation algorithms, in particular, of the optimal adaptive filtering for extremely HdS. In [3, 4], Lorenz has studied the problem of predictability of the atmosphere. It is found that the atmosphere is a chaotic system and a predictability limit to numerical forecast is of about 2 weeks. The barrier of predictability has to be overcome in order to increase the time period of a forecast further. The fact that estimates of the current state are inaccurate and that numerical models have inadequacies leads to forecast errors that grow with increasing forecast lead time. Ensemble forecasting aims at quantifying this flow-dependent forecast uncertainty. Today, a medium-range forecast has become a standard product. In the 1990s, the ensemble forecasting (EnF) technique was introduced in operational centers such as the European Centre for Medium-range Weather Forecast (ECMWF) (see Palmer et al. [5]), the NCEP (US National Center for Environmental Prediction) (Toth and Kalnay [6]). It is found that a single forecast can depart rapidly from the real atmosphere. The idea of the ensemble forecasting is to add the perturbations around the control forecast to produce a collection of forecasts that try to better simulate the possible uncertainties in a numerical forecast. The ensemble mean can then act as a nonlinear filter such that its skill is higher than that of individual members in a statistical sense (Toth and Kalnay [6]).

The chapter is organized as follows. Section 2 outlines first the optimal perturbation (OP) theory, on how the OP plays the important role for seeking the most growing direction of prediction error (PE). The predictability theory of the dynamical system as well as a stability of the filtering algorithm all are developed on the basis of OP. The definition of the optimal deterministic perturbation (ODP) and some theoretical results on the ODP are introduced. It is found that the ODP is associated with the right singular vector (SV) of the system dynamics. In Section 3, the two other classes of ODPs are presented: the leading eigenvector (EV) and real Schur vector (SchV) of the system dynamics. Mention that the first EV is the ODS in the eigen invariant subspace (EI-InS) of the system dynamics. As to the leading SchV, it is ODS in the Schur invariant subspace (Sch-InS) which is closely related to the EI-InS in the sense that the subspace of the leading SchVs, generated by the sampling procedure (Sampling-P, Section 3), converges to the EI-InS. In Section 4, we present the other type of OP called as optimal stochastic perturbation (OSP). Mention that the OSP is a natural extension of the ODP which gives insight into understanding of what represents the most growing PE and how one can produce it by stochastically perturbing the initial state. One important class of perturbations (known as simultaneous stochastic perturbation—SSP) is presented in Section 5. It will be shown that the SSP is very efficient for solving optimization problems in high-dimensional (Hd) setting. The different algorithms for estimating, decomposing … Hd matrices are also presented here. Numerical examples are presented in Section 6 for illustrating the theoretical results and efficiency of the OPs in solving data assimilation problems. The experiment on data assimilation in the Hd ocean model MICOM by the filters constructed on the basis of the Schur ODSs and SSPs is presented in Section 7. The concluding remarks are presented in Section 8.

2. Optimal perturbations: predictability and filter stability

2.1. Stability of filter

The behavior of atmosphere or ocean is recognized as highly sensitive to initial conditions. It means that a small change in an initial condition can alter strongly the trajectory of the system. It is therefore important to be able to know about the directions of rapid growth of the system state. The research on OP is namely aimed at finding the methods to better capture these rapidly growing directions of the system dynamics, to optimize the predictability of the physical process under consideration.

To explain this phenomenon more clearly, consider a standard linear filtering problem


where ΦRn×nis the state transition matrix, HRp×nis an observation matrix. Under standard conditions related to the model and observation noises wk,vk, the minimum mean squared (MMS) estimate x̂kcan be obtained by the well-known KF [2].


where ζk+1=zk+1Hx̂k+1/kis the innovation vector, x̂k+1is the filtered (or analysis) estimate, x̂k+1/kis the one-step ahead prediction for xk+1. The KF gain Kis given by


From Eq. (2), it can be shown that the transition matrix for the filtered estimate equation is expressed by L=IKHΦ.

For HdS, the KF gain (3) is impossible to compute. In a study by Hoang et al. [7], it is suggested to find the gain with the structure


with PrRn×ne—an operator projecting a vector from the reduced space Rneto the full system space Rn, KeRne×pis the gain for the reduced filter. One of very important questions arising here is how one can choose a subspace of projection and structure of Keto make Lto be stable? It is found in the work done by Hoang et al. [8] that detectability of the input-output system (1) is sufficient for the existence of a stabilizing gain Kand this gain can be constructed with Prconsisting from all unstable EVs (or unstable SVs, SchVs. See Section 3) of the system dynamics.

2.2. Singular value decomposition and optimal perturbations

Consider the singular value decomposition (SVD) of Φ[9],


where U1,V1R×n1, Dn1Rn1×n1, n1is the number of all unstable and neutral SVs of Φ. In the future, for simplicity, unless otherwise stated, we say on the set of all unstable SVs as that including all unstable and neutral SVs.

Suppose the system is s-detectable (detectability of all the columns of U1or V1). From the research by Hoang et al. [10], there exists a stabilizing gain Ks(sufficient but not necessary condition) with Pr=U1(see Eq. (4)) such that the transition matrix L=IKsHΦis stable. It signifies that the filter is stable and the estimation error is bounded. The columns of U1, i.e., the left unstable SVs of Φ, serve as a basis for seeking appropriate correction in the filtering algorithm.

On the other hand, in practice, for extreme HdS, one cannot compute all elements of U1but only some of its subset U1U1. Using U1instead of U1cannot guarantee a filter stability. The ensemble forecasting has been proposed as an approach to prevent a possible large error in the forecast and requires a knowledge on the rapidly growing directions of the PE. In this context, the OPs appear to be important which allow to search the rapid growing directions of the PE by model integration of OPs. They (i.e., OPs) are infact the unstable right SVs (RSV).

2.3. Optimal perturbation

Let δxbe a given perturbation, representing an error (uncertainty, deterministic, or stochastic) around the true system state x, i.e., x̂f=x+δx, (at some instant k). The prediction of the system state x̂pcan be obtained by forwarding the numerical model (1) on the basis of x̂f—filtered estimate, i.e., x̂p=Φx̂f. We have then


One sees the perturbation δxin the initial system state “grows” into Φδxwhich represents uncertainty in the forecast.

In general, the perturbation δxmay be any element in the n-dimensional space, Rn, i.e., δxRn. For eflδxfl—a sample of the filtered error (FE) ef, integrating the model by efresults in eplΦδxfl—a sample for the PE ep. By generating the ensemble of perturbations EfLefll=1Laccording to the distribution of ef, one can produce the ensemble of PE samples EpLepll=1Land use them to estimate the distribution of the PE ep. This serves as a basis for the particle filtering [11].

The ensemble-based filtering (EnBF) algorithm is simplified for the standard linear filtering problems with efbeing of zero mean and the error covariance matrix (ECM) P. This technique is aimed to approximate the ECM without solving the matrix Ricatti equation (Ensemble KF—EnKF, [12]. Mention that at the present, one can generate only about O(100) samples at each assimilation instant. This ensemble size is too small compared to the system dimension. That is why it is important to have a good strategy for selecting the “optimal” samples (perturbations) to better approximate the ECM in the filtering algorithm.

2.3.1. Optimal deterministic perturbation



where .2denotes the Euclidean vector norm (here <.,.>denotes the dot product). Let Φhave the SVD (5).

Definition 2.1. The ODP δxois the solution of the extremal problem


under the constraint (7). One can prove

Lemma 2.1. The optimal perturbation in the sense (8) and (7) is


where v1is the first right SV of Φ.

2.3.2. Subspaces of ODPs



Consider the optimization problem


under the constraint (7). Similar to the proof of Lemma 2.1, one can prove

Lemma 2.2. The optimal perturbation in the sense (9) and (7) is


where v2is the second right SV of Φ.

By iteration, for


applying Lemma 2.2 with slight modifications, one finds that the OPs for Φi,i=0,1,,n1are +/vi,i=1,2,n..

Theorem 2.1. The optimal perturbation for the matrix Φi1,i=1,,nis +/viwhere viis the ith leading right SV of Φi1,i=1,,n.

The OP for Φi1will be called the ithOP for Φ(or the ithSOP—singular OP).

Comment 2.2. The OPs, presented above, are optimal in the sense of the Euclidean norm .2. In practice, there is a need to normalize the state vector (using the inverse of the covariance matrix M). The normalization is done by changing δx=M1/2δx, and all the results presented above remain valid s.t. the new δx,


The weighted norm δxM1is known as the Mahanalobis norm.

As to the PE, a normalization is also applied in order to have a possibility to compare different variables like density, temperature, velocity … In this situation, the norm for y=Φδxmay be seminorm [5].

2.4. Ensemble forecasting

The idea of ensemble forecasting is that instead of performing “deterministic” forecasts, stochastic forecasts should be made: several model forecasts are performed by introducing perturbations in the filtered estimate or in the models.

Since 1994, NCEP (National Centers for Environmental Prediction, USA) has been running 17 global forecasts per day, with the perturbations obtained using the method of breeding growing perturbations. This ensures that the perturbations contain growing dynamical perturbations. The length of the forecasts allows the generation of outlook for the second week. At the ECMWF, the perturbation method is based on the use of SVs, which grow even faster than the bred or Lyapunov vector perturbations. The ECMWF ensemble contains 50 members [13].

3. Perturbations based on leading EVs and SchVs

3.1. Adaptive filter (AF)

The idea underlying the AF is to construct a filter which uses feedback in the form of the PE signal (innovation) to adjust the free parameters in the gain to optimize the filter performance. If in the KF, the optimality is defined as a minimum mean squared error (MMS), in the AF, optimality is understood in the sense of MMS for the prediction output error (innovation). This definition allows to define the optimality of the filter in the realization space, but not in the probability space as done in the KF.

The optimal gain thus can be determined from solving the optimization problem by adjusting all elements of the filter gain. There are two major difficulties:

  1. Instability: As the filter gain is estimated stochastically during the optimization process, the filter may become unstable due to the stochastic character of the filter gain.

  2. Reduction of tuning parameters: For extreme HdS, the number of elements in the filter gain is still very high. Reduction of the number of tunning gain elements is necessary.

3.2. Leading EVs and SchVs as optimal perturbations

Interest on stability of the AF arises soon after the AF has been introduced. The study on the filter stability shows that it is possible to provide a filter stability when the system is detectable [8]. For the different parameterized stabilizing gain structures based on a subspace of unstable and neutral EVs, see [8]. As the EVs may be complex and their computation is unstable (Lanczos [14]), in [8], it is proved that one can also ensure a stability of the filter if the space of projection is constructed from a set of unstable and neutral SchVs of the system dynamics. The unstable and neutral real SchVs are referred to as SchVs associated with the unstable and neutral eigenvalues of the system dynamics. The advantage of the real SchVs is that they are real, orthonormal, and their computation is stable. Moreover, the algorithm for estimating dominant SchVs is simple which is based on the power iteration procedure (Sampling-P, see [15]). As to the unstable SVs, although they are real and orthonormal, their computation requires an adjoint operator (the transpose matrix ΦT). Construction of adjoint code (AC) is a time-consuming and tedious process. Approximating leading SVs without (AC) can be done on the basis of Algorithms 5.2.

3.3. EVs as optimal perturbations in the invariant subspace

Let Φbe diagonalizable. Introduce the set


The subspace of xCnsatisfying Φx=λxfor some λC1is known as an invariant subspace of Φ: the matrix Φxacts on to stretch the vector xbut conserves the direction of x. Consider the optimization problem


It is seen that the optimal solution is the first EV xei1of Φwith the largest magnitude equal to λ1. We will call λ1a first optimal EV perturbation (denoted as EI-OP).

For a symmetric matrix, the EI-OP coincides with the SOP. The EI-OP is not unique.

By solving the optimization problem (8) s.t.


one finds the second EI-OP xei2. In a similar way, by defining


for i=1,2,..,n1, we obtain a sequence of EI-OPs xeii, i=1,2,..,n. The first neEI-OPs are unstable SVs.

In general, for a defective case (not diagonalizable), Φdoes not have nlinearly independent EVs and the independent generalized EVs can serve as “optimal” perturbations to construct a subspace of projection in the AF.

To summarize, let the EV decomposition be


where Jis a matrix of Jordan canonical form, Xei1is the matrix inverse of Xei(see Golub and Van Loan [9]). The columns of Xeiare the EVs of Φ, Jis a block diagonal with the diagonal blocks of 1 or 2 dimensions. The rank kdecomposition is Xei,1J1X˜ei,1where


with Xei,1Rn×k, Xei,2Rn×nk. Multiplying the right of EVkby Xei,1yields Xei,1J1, i.e., we obtain the klargest (in modulus) perturbations in the eigen (invariant) space of Φ. The perturbations being the column vectors of Xei,1(i.e., the kfirst EVs of Φ) are the first kOPs of Φin the eigen-invariant subspace (EI-InS).

3.4. Dominant SchVs as OPs in the Schur invariant subspace

The study of Hoang et al. [8] shows that the subspace of projection of the stable filter can be constructed on the basis of all unstable EVs or SchVs of Φ.

Compared to the EVs, the approach based on real Schur decomposition is of preference in practice since the SchVs are real and orthonormal. Moreover, there exists a simple, power iterative algorithm for approaching the set of real leading SchVs. According to Theorem 7.3.1 in Golub and Van Loan [9], the subspace RXs,1spanned by the nuleading SchVs converges to the unique invariant subspace DnuΦ(called a dominant invariant subspace) associated with the eigenvalues λ1,,λnuif λnu>λnu+1. In this sense, we consider the leading SchVs as OPs (denoted as Sch-OP) in the Schur invariant subspace (Sch-InS).

4. Optimal stochastic perturbation (OSP)

In Section 2, the perturbation δxis deterministic (see Definition 2.1). In practice, it happens that δxis of stochastic nature. For example, the priori information on the FE is an zero mean random vector (RV) with the ECM P. The question arising here is how one determine the OP in such situation and how to find it.

We will consider now δxas an element of the Hilbert space Hof RVs. This space His a complete normed linear vector space equipped with the inner product


where E.denotes the mathematical expectation. The norm in His defined as


All elements of Hare of finite variance and for simplicity, we assume they all have zero mean value.

Introduce the set of RVS


Definition 4.1. The optimal stochastic perturbation (OSP) δxois the solution of the extremal problem


under the constraint (19). One can prove

Lemma 4.1. For δxSsδx, there exists δy=δy1δynTSsδxsuch that δx=k=1nvkδyk.

Lemma 4.2. The optimal perturbation in the sense (20) and (19) is


where ψis a RV with zero mean and unit variance, v1is the first right SV of Φ.

Comment 4.1. Comparing the ODP with OSP shows that if the ODS is the first right SV (defined up to the sign), the OSP is an ensemble of vectors lying in the subspace of the first right SV with the lengths being the samples of the RV of zero mean and unit variance.



and consider the objective function


Lemma 4.3. The optimal perturbation in the sense (21) and (19) is δxo=ψv2, where ψis an RV with zero mean and unit variance, v2is the first right SV of Φ.

By iteration, for


applying Lemma 4.3 with slight modifications, one finds that the OSP for Φk,k=0,1,,n1are ψvk,k=1,2,n., where ψis an RV with zero mean and unit variance, vkis the kthright SV of Φ.

Theorem 4.1. The optimal perturbation for the matrix Φi1,i=1,,nis ψvi, where ψis an RV with zero mean and unit variance, vkis the kthright SV of Φ.

5. Simultaneous stochastic perturbations (SSP)

In [16], Spall proposes a simultaneous perturbation stochastic approximation (SPSA) algorithm for finding optimal unknown parameters by minimizing some objective function. The main feature of the simultaneous perturbation gradient approximation (SPGA) resides in the way to approximate the gradient vector (in average): a sample gradient vector is estimated by perturbing simultaneously all components of the unknown vector in a stochastic way. This method requires only two or three measurements of the objective function, regardless of the dimension of the vector of unknown parameters. In a study by Hoang and Baraille [17], the idea of the SPGA is described in detail, with a wide variety of applications in engineering domains. The application to estimation of ECM in the filtering problem is given in the work done by Hoang and Baraille [18]. In the research by Hoang and Baraille [19], a simple algorithm for estimating the elements of an unknown matrix as well as the way to decompose the estimated matrix into a product of two matrices, under the condition that only the matrix-vector product is accessible, has been proposed.

5.1. Theoretical background of SPGA: gradient approximation

The component-wise perturbation is a method for numerical computation of the cost function with respect to the vector of unknown parameters. It is based on the idea to perturb separately each component of the vector of parameters. For very HdS, this technique is impossible to implement. An alternative to the component-wise perturbation is the SSP approach.



denote the gradient of Jθcomputed at θ=θ0. Suppose Δj,j=1,,nare RVs independent identically distributed (i.i.d) according to the Bernoulli law which assumes two values +1 or −1 with equal probabilities 1/2. It implies that


Suppose Jθis infinitely differentiable at θ=θ0. Using a Taylor series expansion,


where D2Jθ0is the Hessian matrix computed at θθ0. For the choice


cis a small positive value, from Eq. (24)


Dividing both sides of the last equality by δθk=cΔkimplies


Taking the mathematical expectation for both sides of the last equation yields


One sees that from the assumptions on Δ¯, EΔ¯k=010Tit follows DJu0TEΔ¯k=Jθ0/θk. Moreover, as all the moments of the Bernoulli variables Δiand Δi1are finite, EΔ¯TD2Jθ0Δ¯k=0since there exists a finite D2Jθ0, one concludes that


The result expressed by Eq. (28) constitutes a basis for approximating the gradient vector by simultaneous perturbation. The left of Eq. (28) can be easily approximated by noticing that for an ensemble of Li.i.d samples Δ¯1Δ¯L, we can generate the corresponding ensemble of Li.i.d sample estimates for the gradient vector at θ=θ0,


where Δklis the kthcomponent of the lthsample Δ¯l. The left of Eq. (28) is then well approximated by averaging Lsample gradients in Eq. (29),


Introduce the notations


Theorem 1 (Hoang and Baraille [19]) states that the estimate mLm1LmnθLTconverges to the gradient vector DJθ0as Land c0with the order O1/Lwhere


5.2. Algorithm for estimation of an unknown matrix

Let Δ¯Δ1ΔnT, Δi,i=1,,nbe Bernoulli independent and identically distributed (i.i.d.) variables assuming two values ±1 with equal probability 1/2. Introduce Δ¯11/Δ1,,1/ΔnT, Δ¯ccΔ¯,c>0is a small positive value.

Algorithm 5.1. Suppose it is possible to compute the product Φx=bxfor a given x. At the beginning let l=1. Let the value ube assigned to the vector x, i.e., xu, Lbe a (large) fixed integer number.

Step 1. Generate Δ¯lwhose components are lthsamples of the Bernoulli i.i.d. variables assuming two values +/− 1 with equal probabilities 1/2;

Step 2. Compute δbl=Φu+Δ¯clΦu, Δ¯cl=cΔ¯l.

Step 3. Compute gil=δbilΔ¯cl1, δbiis the ithcomponent of δb, gilis the column vector consisting of derivative of biuw.r.t. u, i=1,,m.

Step 4. Go to Step 1 if l<L. Otherwise, go to Step 5.

Step 5. Compute

ĝi=1Ll=1Lgil,i=1,,m, Φ̂LDxb=ĝ1ĝmT.

5.3. Operations with Φand its transpose

Algorithm 5.1 allows to store Φ̂Las composed from the two ensembles of vectors elements:


The product z=Φ̂Ly, yRn, can be performed as zi=k=1nϕ̂ikyk= k=1n1Ll=1Lδbilδxklyk, or in a more compact form


Eq. (33) allows to perform z=Φ̂Lywith Lm+2n+1elementary operations.

Similarly, computation of ziof z=Φ̂TLy,yRmis performed as


5.4. Estimation of decomposition of Φ

Let Φbe a matrix of dimensions m×n. For definiteness, let mnwith rankΦ=m. We want to find the best approximation for Φamong members of the class of matrices


under the constraint

Condition (C) A,Bare matrices of dimension m×r,r×n, rm, rankABT=r.

Under the condition (C), the optimization problem is formulated as


where .Fdenotes the Frobenius matrix norm. Consider Φand let UΣVTbe SVD of Φ(5). Let Φ˜=Φ+ΔΦ,Φ˜=U˜Σ˜V˜Tand σ˜1σ˜2σ˜m, σ˜kbe the kthsingular value of Φ˜. Then, we have


where AoBoTis a solution to the problem (36) s.t Condition (C) (Theorem 3.1 of Hoang and Baraille [19]).

Theorem 3.1. Hoang and Baraille [19] implies that ΦeoAoBoTis equal to the matrix formed by truncating the SVD of Φto its first rSVs and singular values. It allows to avoid storing elements of the estimate Φ̂Lof Φ(their number is of order O10m×n).

5.4.1. Decomposition algorithms

Let the elements of Φ(or Φ̂) be available (may be in algorithmic form). By perturbing stochastically simultaneously all the elements of Aand B, one can write out the iterative algorithm for estimating the elements of Aand B. For more detail, see Hoang and Baraille [19].

5.4.2. Iterative decomposition algorithm

Another way to decompose the matrix Φis to solve iteratively the following optimization problems

Algorithm 5.2

At the beginning let i=1.

Step 1. For i=1, solve the minimization problem


Its solution is denoted as âi,b̂i.

Step 2. For i<r, put ii+1and solve the problem


Step 3. If i=r, compute

Φ̂=ÂrB̂Tr, Âr=â1âr, B̂r=b̂1b̂r.

and stop. Otherwise, go to Step 2.

From Theorem 3.2 of Hoang and Baraille [19], the couple Âr,B̂ris a solution for the problem (36)(C).

6. Numerical example

Consider the matrix ΦR2×2


The singular values and the right SVs for Φare displayed in Table 1 which are obtained by solving the classical equations for eigenvalues of ΦTΦ.


Table 1.

Estimates of Φobtained by Algorithm 5.2.

First, we apply Algorithm 5.1 to estimate the matrix Φ. Figure 1 shows the estimates produced by Algorithm 5.1. It is seen that the estimates are converging quickly to the true elements of Φ.

Figure 1.

Estimates of elements of the matrix Φ.

Next Algorithm 5.2 (Iterative Decomposition Algorithm) has been applied to estimate the decomposition of the matrix Φ. After each ithiteration, the algorithm yieds b̂i,ĉi.

The different OPs are shown in Table 2 where xsvri, xeii, and xschiare the theoretical SV-POs, EI-POs, Sch-POs. The vectors x̂schiare the components of Xtcomputed by Algorithm 3.1 (Sampling-P). As to ĉni, they are the results of normalization (with the unit Euclidean norm) of ĉi.


Table 2.

Different OPs.

Looking at the first OPs, one sees that xsvr1, xsch1, x̂sch1, and ĉn1produce almost the same amplification. The first xei1has the amplification three times less than those of xsvr1, xsch1. The second xsch2is much less opimal than xsvr2and ĉn2. By comparing xsvriwith ĉnifor i=1,2, one concludes that the obtained results justify the correctness of Theorem 3.1 of Hoang and Baraille [19]. Mention that only x̂schiand ĉnican be calculated for HdS.

In Table 1, we show the results obtained by Algorithm 5.2 after two consecutive iterations (matrix estimation in R1subspace). The elements of the true Φ=ϕijare displayed in the second column, whereas their estimates—in the third column,


The estimates, resulting from the first iteration, are the elements of Φ̂1(Table 1, column 4). After the first iteration, Φ2ΦΦ̂1=b2c2,Tand the optimization yields the estimates ϕ̂ij2displayed in the column 5. From the columns 4–5, one sees that the first iteration allows to well estimate the two biggest elements Φ11=5, Φ12=7. In the similar way, the second iteration captures the two biggest elements of Φ2.

7. Assimilation in high-dimensional ocean model MICOM

7.1. Ocean model MICOM

To see the impact of optimal SchVs in the design of filtering algorithm for HdS, in this section, we present the results of the experiment on the Hd ocean model MICOM (Miami Isopycnal Ocean Model). This numerical experiment is identical to that described in Hoang and Baraille [15]. The model configuration is a domain situated in the North Atlantic from 30°N to 60°N and 80°W to 44°W; for the exact model domain and some main features of the ocean current produced by the model, see and Baraille [15]. The system state x=huvwhere h=hijkis a layer thickness and u=uijk,v=vijkare two velocity components. Mention that after discretization, the dimension of the system state is n=302400. The observations available at each assimilation instant are the sea surface height (SSH) with dimension p=221.

7.1.1. Data matrix based on dominant Sch-Ops

The filter is a reduced-order filter (ROF) with the variable has a reduced state and u,vare calculated on the basis of the geostrophy hypothesis. To obtain the gain in the ROF, first the Algorithm 3.1 has been implemented to generate an ensemble of dominant SchVs (totally 72 SchVs, denoted as EnSCH). The sample ECM MdSCHis computed on the basis of the EnSCH. Due to rank deficiency, the sample MdSCHis considered only as a data matrix. The optimization procedure is applied to minimize the distance between the data matrix MdSCHand the structured parametrized ECM MSCH=MvMhwhich is written in the form of the Schur product of two matrices MSCH=MvθMhθ. Here, Mvis the vertical ECM, Mhis the horizontal ECM [18]), θis a vector of unknown parameters. Mention that the hypothesis on separability of the vertical and horizontal variables in the ECM is not new in the meteorology [20]. The gain is computed according to Eq. (3) with R=αI, α>0is a small positive value. The ROF is denoted as PEF (SSP).

7.1.2. Data matrix based on SSP approach

The second data matrix MdSSPis obtained by perturbing the system state according to the SSP method. The SSP samples are simulated in the way similar to that described above for generating EnSCH, with the difference that the perturbation components δhlijkare the i.i.d. random Bernoulli variables assuming two values ±1 with the equal probability 1/2. The same optimization procedure has been applied to estimate MSSP. The obtained ROF is denoted as PEF (SSP).

Figure 2 shows the evolution of estimates for the gain coefficients k1computed from the estimated coefficients of ĉklof MdSCHand MdSSPon the basis of EnSCH(curve” schur”) and EnSSP(curve “random”), during model integration. It is seen that two coefficients are evolved in nearly the same manner, of nearly the same magnitude as that of k1in the CHF (Cooper-Haines filter, Cooper and Haines [21]). Mention that the CHF is a filter widely used in the oceanic data assimilation, which projects the PE of the surface height data by lifting or lowering of water columns.

Figure 2.

Evolution of estimates for the gain coefficients k1 computed from ĉkl on the basis of EnSCH (curve “Schur”) and EnSSP (curve “random”), during model integration. It is seen that two coefficients are evolved in nearly the same manner, of nearly the same magnitude as that of k1 in the CHF. The same picture is obtained for other ck,k=2,3,4.

The same pictures are obtained for the estimates ĉk,k=2,3,4. Mention that in the CHF, c2=c3=0.

7.2. Performance of different filters

In Table 3, the performances of the three filters are displayed. The errors are the averaged (spatially and temporally) rms of PE for the SSH and for the two velocity components uand v.

rmsCHF (cm)PEF (SCH) (cm/s)PEF (RAN) (cm/s)

Table 3.

rms of PE for ssh, and u, vvelocity components.

The results in Table 3 show that two filters PEF (SCH) and PEF (SSP) are practically of the same performance, and their estimates are much better compared to those of the CHF, with a slightly better performance for the PEF (SSP). We note that as the PEF (SCH) is constructed on the basis of an ensemble of samples tending to the first dominant SchV, its performance must be theoretically better than that of the PEF (SSP). The slightly better performance of PEF (SSP) (compared to that of PEF (SCH)) may be explained by the fact that the best theoretical performance of PEF (SCH) can be obtained only if the model is linear, stationary, and the number of PE samples in EnSCHat each iteration must be large enough. The ensemble size of EnSCHin the present experiment is too small compared with the dimension of the MICOM model.

To illustrate the efficiency of adaptation, in Figure 3, we show the cost functions (variances of innovation) resulting from the three filters PEF (SCH), PEF (SSP), and AF (i.e., APEF based on PEF (SCH); the same performance is observed for the AF based on PEF (SSP)). Undoubtedly, the adaptation allows to improve considerably the performances of nonadaptive filters.

Figure 3.

Variance of PE resulting from three filters PEF (SCH), PEF (SSP), and AF. It is seen that the AF yields much better performance compared to the two other nonadaptive filters PEF (SCH) and PEF (SSP).

8. Conclusion remarks

We have presented in this chapter the different types of OPs—deterministic, stochastic, or optimal, the invariant subspaces of the system dynamics. The ODPs and OSPs play an important role in the study on the predictability of the system dynamics as well as in construction of optimal OFS for environmental geophysical systems.

One other class of perturbation known as SSP is found to be a very efficient tool for solving optimization and estimation problems, especially with Hd matrices and in computing the optimal perturbations.

The numerical experiments presented in this chapter confirm the important role of the different types of OPs in the numerical study of Hd assimilation systems.

© 2018 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution 3.0 License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

How to cite and reference

Link to this chapter Copy to clipboard

Cite this chapter Copy to clipboard

Hong Son Hoang and Remy Baraille (November 5th 2018). On Optimal and Simultaneous Stochastic Perturbations with Application to Estimation of High-Dimensional Matrix and Data Assimilation in High-Dimensional Systems, Perturbation Methods with Applications in Science and Engineering, İlkay Bakırtaş, IntechOpen, DOI: 10.5772/intechopen.77273. Available from:

chapter statistics

354total chapter downloads

More statistics for editors and authors

Login to your personal dashboard for more detailed statistics on your publications.

Access personal reporting

Related Content

This Book

Next chapter

Periodic Perturbations: Parametric Systems

By Albert Morozov

Related Book

First chapter

Recent Advances in Fragment Molecular Orbital-Based Molecular Dynamics (FMO-MD) Simulations

By Yuto Komeiji, Yuji Mochizuki, Tatsuya Nakano and Hirotoshi Mori

We are IntechOpen, the world's leading publisher of Open Access books. Built by scientists, for scientists. Our readership spans scientists, professors, researchers, librarians, and students, as well as business professionals. We share our knowledge and peer-reveiwed research papers with libraries, scientific and engineering societies, and also work with corporate R&D departments and government entities.

More About Us