Linear system identification from ensemble snapshot observations

E-print/Working paper (2019)

Developments in transcriptomics techniques have caused a large demand in tailored computational methods for modelling gene expression dynamics from experimental data. Recently, so-called single-cell ... [more ▼]

Developments in transcriptomics techniques have caused a large demand in tailored computational methods for modelling gene expression dynamics from experimental data. Recently, so-called single-cell experiments have revolutionised genetic studies. These experiments yield gene expression data in single cell resolution for a large number of cells at a time. However, the cells are destroyed in the measurement process, and so the data consist of snapshots of an ensemble evolving over time, instead of time series. The problem studied in this article is how such data can be used in modelling gene regulatory dynamics. Two different paradigms are studied for linear system identification. The first is based on tracking the evolution of the distribution of cells over time. The second is based on the so-called pseudotime concept, identifying a common trajectory through the state space, along which cells propagate with different rates. Therefore, at any given time, the population contains cells in different stages of the trajectory. Resulting methods are compared in numerical experiments. [less ▲]

Spatial discretization error in Kalman filtering for discrete-time infinite dimensional systems

in IMA Journal of Mathematical Control and Information (2018), 35(suppl_1), 51-72

We derive a reduced-order state estimator for discrete-time infinite dimensional linear systems with finite dimensional Gaussian input and output noise. This state estimator is the optimal one-step ... [more ▼]

We derive a reduced-order state estimator for discrete-time infinite dimensional linear systems with finite dimensional Gaussian input and output noise. This state estimator is the optimal one-step estimate that takes values in a fixed finite dimensional subspace of the system’s state space — consider, for example, a Finite Element space. The structure of the obtained state estimator is like the Kalman filter, but with an additional optimal embedding operator mapping from the reduced space to the original state space. We derive a Riccati difference equation for the error covariance and use sensitivity analysis to obtain a bound for the error of the state estimate due to the state space discretization. [less ▲]

Bayesian variable selection in linear dynamical systems

E-print/Working paper (2018)

We develop a method for reconstructing regulatory interconnection networks between variables evolving according to a linear dynamical system. The work is motivated by the problem of gene regulatory ... [more ▼]

We develop a method for reconstructing regulatory interconnection networks between variables evolving according to a linear dynamical system. The work is motivated by the problem of gene regulatory network inference, that is, finding causal effects between genes from gene expression time series data. In biological applications, the typical problem is that the sampling frequency is low, and consequentially the system identification problem is ill-posed. The low sampling frequency also makes it impossible to estimate derivatives directly from the data. We take a Bayesian approach to the problem, as it offers a natural way to incorporate prior information to deal with the ill-posedness, through the introduction of sparsity promoting prior for the underlying dynamics matrix. It also provides a framework for modelling both the process and measurement noises. We develop Markov Chain Monte Carlo samplers for the discrete-valued zero-structure of the dynamics matrix, and for the continuous-time trajectory of the system. [less ▲]

Iterative observer-based state and parameter estimation for linear systems

in ESAIM: Control, Optimisation and Calculus of Variations (2018), 24(1), 265-288

We propose an iterative method for joint state and parameter estimation using measurements on a time interval [0,T] for systems that are backward output stabilizable. Since this time interval is fixed ... [more ▼]

We propose an iterative method for joint state and parameter estimation using measurements on a time interval [0,T] for systems that are backward output stabilizable. Since this time interval is fixed, errors in initial state may have a big impact on the parameter estimate. We propose to use the back and forth nudging (BFN) method for estimating the system’s initial state and a Gauss–Newton step between BFN iterations for estimating the system parameters. Taking advantage of results on the optimality of the BFN method, we show that for systems with skew-adjoint generators, the initial state and parameter estimate minimizing an output error cost functional is an attractive fixed point for the proposed method. We treat both linear source estimation and bilinear parameter estimation problems. [less ▲]

Convergence of discrete time Kalman filter estimate to continuous time estimate

in International Journal of Control (2016), 89(4), 668-679

Composition of passive boundary control systems

in Mathematical Control and Related Fields (2013), 3(1), 1-19

Interaction of vocal fold and vocal tract oscillations

in Proceedings of the 24th Nordic Seminar on Computational Mechanics (2011)

We study the mechanical feedback coupling between the human vocal folds and vocal tract (VT) by simulating fundamental frequency glides over the lowest VT resonance. In the classical source–filter theory ... [more ▼]

We study the mechanical feedback coupling between the human vocal folds and vocal tract (VT) by simulating fundamental frequency glides over the lowest VT resonance. In the classical source–filter theory of speech production, the vocal folds produce a signal which is filtered by the resonator, vocal tract without any feedback. We have developed a computational model of the vocal folds and the VT that also includes a counter pressure from the VT to the vocal folds. This coupling gives rise to new computational observations (such as modal locking) that can be established experimentally. [less ▲]