Bayesian methods for estimation, optimization and experimental design

Mathematical models often contain parameters that need to be calibrated from measured data. The emergence of efficient Markov Chain Monte Carlo (MCMC) methods has made the Bayesian approach a standard tool in quantifying the uncertainty in the parameters. With MCMC, the parameter estimation problem can be solved in a fully statistical manner, and the whole distribution of the parameters can be explored, instead of obtaining point estimates and using, e.g., Gaussian approximations. In this thesis, MCMC methods are applied to parameter estimation problems in chemical reaction engineering, population ecology, and climate modeling. Motivated by the climate model experiments, the methods are developed further to make them more suitable for problems where the model is computationally intensive. After the parameters are estimated, one can start to use the model for various tasks. Two such tasks are studied in this thesis: optimal design of experiments, where the task is to design the next measurements so that the parameter uncertainty is minimized, and model-based optimization, where a model-based quantity, such as the product yield in a chemical reaction model, is optimized. In this thesis, novel ways to perform these tasks are developed, based on the output of MCMC parameter estimation. A separate topic is dynamical state estimation, where the task is to estimate the dynamically changing model state, instead of static parameters. For example, in numerical weather prediction, an estimate of the state of the atmosphere must constantly be updated based on the recently obtained measurements. In this thesis, a novel hybrid state estimation method is developed, which combines elements from deterministic and random sampling methods.

On Variational Ensemble Kalman Filtering

The Extended Kalman Filter (EKF) and four dimensional assimilation variational method (4D-VAR) are both advanced data assimilation methods. The EKF is impractical in large scale problems and 4D-VAR needs much effort in building the adjoint model. In this work we have formulated a data assimilation method that will tackle the above difficulties. The method will be later called the Variational Ensemble Kalman Filter (VEnKF). The method has been tested with the Lorenz95 model. Data has been simulated from the solution of the Lorenz95 equation with normally distributed noise. Two experiments have been conducted, first with full observations and the other one with partial observations. In each experiment we assimilate data with three-hour and six-hour time windows. Different ensemble sizes have been tested to examine the method. There is no strong difference between the results shown by the two time windows in either experiment. Experiment I gave similar results for all ensemble sizes tested while in experiment II, higher ensembles produce better results. In experiment I, a small ensemble size was enough to produce nice results while in experiment II the size had to be larger. Computational speed is not as good as we would want. The use of the Limited memory BFGS method instead of the current BFGS method might improve this. The method has proven succesful. Even if, it is unable to match the quality of analyses of EKF, it attains significant skill in forecasts ensuing from the analysis it has produced. It has two advantages over EKF; VEnKF does not require an adjoint model and it can be easily parallelized.

Variational Ensemble Kalman Filtering in Hydrology

The current thesis manuscript studies the suitability of a recent data assimilation method, the Variational Ensemble Kalman Filter (VEnKF), to real-life fluid dynamic problems in hydrology. VEnKF combines a variational formulation of the data assimilation problem based on minimizing an energy functional with an Ensemble Kalman filter approximation to the Hessian matrix that also serves as an approximation to the inverse of the error covariance matrix. One of the significant features of VEnKF is the very frequent re-sampling of the ensemble: resampling is done at every observation step. This unusual feature is further exacerbated by observation interpolation that is seen beneficial for numerical stability. In this case the ensemble is resampled every time step of the numerical model. VEnKF is implemented in several configurations to data from a real laboratory-scale dam break problem modelled with the shallow water equations. It is also tried in a two-layer Quasi- Geostrophic atmospheric flow problem. In both cases VEnKF proves to be an efficient and accurate data assimilation method that renders the analysis more realistic than the numerical model alone. It also proves to be robust against filter instability by its adaptive nature.

Variational ensemble Kalman filtering applied to data assimilation problems in computational fluid dynamics

One challenge on data assimilation (DA) methods is how the error covariance for the model state is computed. Ensemble methods have been proposed for producing error covariance estimates, as error is propagated in time using the non-linear model. Variational methods, on the other hand, use the concepts of control theory, whereby the state estimate is optimized from both the background and the measurements. Numerical optimization schemes are applied which solve the problem of memory storage and huge matrix inversion needed by classical Kalman filter methods. Variational Ensemble Kalman filter (VEnKF), as a method inspired the Variational Kalman Filter (VKF), enjoys the benefits from both ensemble methods and variational methods. It avoids filter inbreeding problems which emerge when the ensemble spread underestimates the true error covariance. In VEnKF this is tackled by resampling the ensemble every time measurements are available. One advantage of VEnKF over VKF is that it needs neither tangent linear code nor adjoint code. In this thesis, VEnKF has been applied to a two-dimensional shallow water model simulating a dam-break experiment. The model is a public code with water height measurements recorded in seven stations along the 21:2 m long 1:4 m wide flume’s mid-line. Because the data were too sparse to assimilate the 30 171 model state vector, we chose to interpolate the data both in time and in space. The results of the assimilation were compared with that of a pure simulation. We have found that the results revealed by the VEnKF were more realistic, without numerical artifacts present in the pure simulation. Creating a wrapper code for a model and DA scheme might be challenging, especially when the two were designed independently or are poorly documented. In this thesis we have presented a non-intrusive approach of coupling the model and a DA scheme. An external program is used to send and receive information between the model and DA procedure using files. The advantage of this method is that the model code changes needed are minimal, only a few lines which facilitate input and output. Apart from being simple to coupling, the approach can be employed even if the two were written in different programming languages, because the communication is not through code. The non-intrusive approach is made to accommodate parallel computing by just telling the control program to wait until all the processes have ended before the DA procedure is invoked. It is worth mentioning the overhead increase caused by the approach, as at every assimilation cycle both the model and the DA procedure have to be initialized. Nonetheless, the method can be an ideal approach for a benchmark platform in testing DA methods. The non-intrusive VEnKF has been applied to a multi-purpose hydrodynamic model COHERENS to assimilate Total Suspended Matter (TSM) in lake Säkylän Pyhäjärvi. The lake has an area of 154 km2 with an average depth of 5:4 m. Turbidity and chlorophyll-a concentrations from MERIS satellite images for 7 days between May 16 and July 6 2009 were available. The effect of the organic matter has been computationally eliminated to obtain TSM data. Because of computational demands from both COHERENS and VEnKF, we have chosen to use 1 km grid resolution. The results of the VEnKF have been compared with the measurements recorded at an automatic station located at the North-Western part of the lake. However, due to TSM data sparsity in both time and space, it could not be well matched. The use of multiple automatic stations with real time data is important to elude the time sparsity problem. With DA, this will help in better understanding the environmental hazard variables for instance. We have found that using a very high ensemble size does not necessarily improve the results, because there is a limit whereby additional ensemble members add very little to the performance. Successful implementation of the non-intrusive VEnKF and the ensemble size limit for performance leads to an emerging area of Reduced Order Modeling (ROM). To save computational resources, running full-blown model in ROM is avoided. When the ROM is applied with the non-intrusive DA approach, it might result in a cheaper algorithm that will relax computation challenges existing in the field of modelling and DA.

Advances in sequential data assimilation and numerical weather forecasting: an Ensemble Transform Kalman-Bucy Filter, a study on clustering in deterministic Ensemble Square Root Filters, and a test of a new time stepping scheme in an atmospheric model

This dissertation deals with aspects of sequential data assimilation (in particular ensemble Kalman filtering) and numerical weather forecasting. In the first part, the recently formulated Ensemble Kalman-Bucy (EnKBF) filter is revisited. It is shown that the previously used numerical integration scheme fails when the magnitude of the background error covariance grows beyond that of the observational error covariance in the forecast window. Therefore, we present a suitable integration scheme that handles the stiffening of the differential equations involved and doesn’t represent further computational expense. Moreover, a transform-based alternative to the EnKBF is developed: under this scheme, the operations are performed in the ensemble space instead of in the state space. Advantages of this formulation are explained. For the first time, the EnKBF is implemented in an atmospheric model. The second part of this work deals with ensemble clustering, a phenomenon that arises when performing data assimilation using of deterministic ensemble square root filters in highly nonlinear forecast models. Namely, an M-member ensemble detaches into an outlier and a cluster of M-1 members. Previous works may suggest that this issue represents a failure of EnSRFs; this work dispels that notion. It is shown that ensemble clustering can be reverted also due to nonlinear processes, in particular the alternation between nonlinear expansion and compression of the ensemble for different regions of the attractor. Some EnSRFs that use random rotations have been developed to overcome this issue; these formulations are analyzed and their advantages and disadvantages with respect to common EnSRFs are discussed. The third and last part contains the implementation of the Robert-Asselin-Williams (RAW) filter in an atmospheric model. The RAW filter is an improvement to the widely popular Robert-Asselin filter that successfully suppresses spurious computational waves while avoiding any distortion in the mean value of the function. Using statistical significance tests both at the local and field level, it is shown that the climatology of the SPEEDY model is not modified by the changed time stepping scheme; hence, no retuning of the parameterizations is required. It is found the accuracy of the medium-term forecasts is increased by using the RAW filter.

Microblogging Temporal Summarization: Filtering Important Twitter Updates for Breaking News

While news stories are an important traditional medium to broadcast and consume news, microblogging has recently emerged as a place where people can dis- cuss, disseminate, collect or report information about news. However, the massive information in the microblogosphere makes it hard for readers to keep up with these real-time updates. This is especially a problem when it comes to breaking news, where people are more eager to know “what is happening”. Therefore, this dis- sertation is intended as an exploratory effort to investigate computational methods to augment human effort when monitoring the development of breaking news on a given topic from a microblog stream by extractively summarizing the updates in a timely manner. More specifically, given an interest in a topic, either entered as a query or presented as an initial news report, a microblog temporal summarization system is proposed to filter microblog posts from a stream with three primary concerns: topical relevance, novelty, and salience. Considering the relatively high arrival rate of microblog streams, a cascade framework consisting of three stages is proposed to progressively reduce quantity of posts. For each step in the cascade, this dissertation studies methods that improve over current baselines. In the relevance filtering stage, query and document expansion techniques are applied to mitigate sparsity and vocabulary mismatch issues. The use of word embedding as a basis for filtering is also explored, using unsupervised and supervised modeling to characterize lexical and semantic similarity. In the novelty filtering stage, several statistical ways of characterizing novelty are investigated and ensemble learning techniques are used to integrate results from these diverse techniques. These results are compared with a baseline clustering approach using both standard and delay-discounted measures. In the salience filtering stage, because of the real-time prediction requirement a method of learning verb phrase usage from past relevant news reports is used in conjunction with some standard measures for characterizing writing quality. Following a Cranfield-like evaluation paradigm, this dissertation includes a se- ries of experiments to evaluate the proposed methods for each step, and for the end- to-end system. New microblog novelty and salience judgments are created, building on existing relevance judgments from the TREC Microblog track. The results point to future research directions at the intersection of social media, computational jour- nalism, information retrieval, automatic summarization, and machine learning.

FILTERING TECHNIQUES APPLIED ON RAW CARRIER PHASE FOR GPS DETECTING SMALL DYNAMIC DISPLACEMENTS

This work is part of a research under construction since 2000, in which the main objective is to measure small dynamic displacements by using L1 GPS receivers. A very sensible way to detect millimetric periodic displacements is based on the Phase Residual Method (PRM). This method is based on the frequency domain analysis of the phase residuals resulted from the L1 double difference static data processing of two satellites in almost orthogonal elevation angle. In this article, it is proposed to obtain the phase residuals directly from the raw phase observable collected in a short baseline during a limited time span, in lieu of obtaining the residual data file from regular GPS processing programs which not always allow the choice of the aimed satellites. In order to improve the ability to detect millimetric oscillations, two filtering techniques are introduced. One is auto-correlation which reduces the phase noise with random time behavior. The other is the running mean to separate low frequency from the high frequency phase sources. Two trials have been carried out to verify the proposed method and filtering techniques. One simulates a 2.5 millimeter vertical antenna displacement and the second uses the GPS data collected during a bridge load test. The results have shown a good consistency to detect millimetric oscillations.

Deformed Gaussian-orthogonal-ensemble description of small-world networks

The study of spectral behavior of networks has gained enthusiasm over the last few years. In particular, random matrix theory (RMT) concepts have proven to be useful. In discussing transition from regular behavior to fully chaotic behavior it has been found that an extrapolation formula of the Brody type can be used. In the present paper we analyze the regular to chaotic behavior of small world (SW) networks using an extension of the Gaussian orthogonal ensemble. This RMT ensemble, coined the deformed Gaussian orthogonal ensemble (DGOE), supplies a natural foundation of the Brody formula. SW networks follow GOE statistics until a certain range of eigenvalue correlations depending upon the strength of random connections. We show that for these regimes of SW networks where spectral correlations do not follow GOE beyond a certain range, DGOE statistics models the correlations very well. The analysis performed in this paper proves the utility of the DGOE in network physics, as much as it has been useful in other physical systems.

Spin filtering and disorder-induced magnetoresistance in carbon nanotubes: Ab initio calculations

Nitrogen-doped carbon nanotubes can provide reactive sites on the porphyrin-like defects. It is well known that many porphyrins have transition-metal atoms, and we have explored transition-metal atoms bonded to those porphyrin-like defects inN-doped carbon nanotubes. The electronic structure and transport are analyzed by means of a combination of density functional theory and recursive Green's function methods. The results determined the heme B-like defect (an iron atom bonded to four nitrogens) is the most stable and has a higher polarization current for a single defect. With randomly positioned heme B defects in nanotubes a few hundred nanometers long, the polarization reaches near 100%, meaning they are effective spin filters. A disorder-induced magnetoresistance effect is also observed in those long nanotubes, and values as high as 20 000% are calculated with nonmagnectic eletrodes.

H(infinity) filtering for rectangular discrete-time descriptor systems

This paper deals with the H(infinity) recursive estimation problem for general rectangular time-variant descriptor systems in discrete time. Riccati-equation based recursions for filtered and predicted estimates are developed based on a data fitting approach and game theory. In this approach, the nature determines a state sequence seeking to maximize the estimation cost, whereas the estimator tries to find an estimate that brings the estimation cost to a minimum. A solution exists for a specified gamma-level if the resulting cost is positive. In order to present some computational alternatives to the H(infinity) filters developed, they are rewritten in information form along with the respective array algorithms. (C) 2009 Elsevier Ltd. All rights reserved.

Noise reduction in biomedical speech signal processing based on time and frequency Kalman filtering combined with spectral subtraction

The goal of this paper is to study and propose a new technique for noise reduction used during the reconstruction of speech signals, particularly for biomedical applications. The proposed method is based on Kalman filtering in the time domain combined with spectral subtraction. Comparison with discrete Kalman filter in the frequency domain shows better performance of the proposed technique. The performance is evaluated by using the segmental signal-to-noise ratio and the Itakura-Saito`s distance. Results have shown that Kalman`s filter in time combined with spectral subtraction is more robust and efficient, improving the Itakura-Saito`s distance by up to four times. (C) 2007 Elsevier Ltd. All rights reserved.

Information Filtering and Array Algorithms for Discrete-Time Markovian Jump Linear Systems

This technical note develops information filter and array algorithms for a linear minimum mean square error estimator of discrete-time Markovian jump linear systems. A numerical example for a two-mode Markovian jump linear system, to show the advantage of using array algorithms to filter this class of systems, is provided.

Modeling and Filtering Double-Frequency Jitter in One-Way Master-Slave Chain Networks

One-way master-slave (OWMS) chain networks are widely used in clock distribution systems due to their reliability and low cost. As the network nodes are phase-locked loops (PLLs), double-frequency jitter (DFJ) caused by their phase detectors appears as an impairment to the performance of the clock recovering process found in communication systems and instrumentation applications. A nonlinear model for OWMS chain networks with P + 1 order PLLs as slave nodes is presented, considering the DFJ. Since higher order filters are more effective in filtering DFJ, the synchronous state stability conditions for an OWMS chain network with third-order nodes are derived, relating the loop gain and the filter coefficients. By using these conditions, design examples are discussed.

On the Filtering Problem for Continuous-Time Markov Jump Linear Systems with no Observation of the Markov Chain

We consider in this paper the optimal stationary dynamic linear filtering problem for continuous-time linear systems subject to Markovian jumps in the parameters (LSMJP) and additive noise (Wiener process). It is assumed that only an output of the system is available and therefore the values of the jump parameter are not accessible. It is a well known fact that in this setting the optimal nonlinear filter is infinite dimensional, which makes the linear filtering a natural numerically, treatable choice. The goal is to design a dynamic linear filter such that the closed loop system is mean square stable and minimizes the stationary expected value of the mean square estimation error. It is shown that an explicit analytical solution to this optimal filtering problem is obtained from the stationary solution associated to a certain Riccati equation. It is also shown that the problem can be formulated using a linear matrix inequalities (LMI) approach, which can be extended to consider convex polytopic uncertainties on the parameters of the possible modes of operation of the system and on the transition rate matrix of the Markov process. As far as the authors are aware of this is the first time that this stationary filtering problem (exact and robust versions) for LSMJP with no knowledge of the Markov jump parameters is considered in the literature. Finally, we illustrate the results with an example.