arXiv subjects

Håkon Tjelmeland

Publications and source records attributed to Håkon Tjelmeland.

13 recordsLinked to original sources

arXiv2026

Multiproposal Elliptical Slice Sampling

We introduce Multiproposal Elliptical Slice Sampling, a self-tuning multiproposal Markov chain Monte Carlo method for Bayesian inference with Gaussian priors. Our method generalizes the Elliptical Slice Sampling algorithm by 1) allowing multiple candidate proposals to be sampled in parallel at each self-tuning step, and 2) basing the acceptance step on a distance-informed transition matrix that can favor proposals far from the current state. This allows larger moves in state space and faster self-tuning, at essentially no additional wall clock time for expensive likelihoods, and results in improved mixing. We additionally provide theoretical arguments and experimental results suggesting dimension-robust mixing behavior, making the algorithm particularly well suited for Bayesian PDE inverse problems.

stat.CO↗

arXiv2026

Bayesian Semi-Blind Deconvolution at Scale

Blind image deconvolution refers to the problem of simultaneously estimating the blur kernel and the true image from a set of observations when both the blur kernel and the true image are unknown. Sometimes, additional image and/or blur information is available and the term semi-blind deconvolution (SBD) is used. We consider a recently introduced Bayesian conjugate hierarchical model for SBD, formulated on an extended cyclic lattice to allow a computationally scalable Gibbs sampler. In this article, we extend this model to the general SBD problem, rewrite the previously proposed Gibbs sampler so that operations are performed in the Fourier domain whenever possible, and introduce a new marginal Hamiltonian Monte Carlo (HMC) blur update, obtained by analytically integrating the blur-image joint conditional over the image. The cyclic formulation combined with non-trivial linear algebra manipulations allows a Fourier-based, scalable HMC update, otherwise complicated by the rigid constraints of the SBD problem. Having determined the padding size in the cyclic embedding through a numerical experiment, we compare the mixing and exploration behaviour of the Gibbs and HMC blur updates on simulated data and on a real geophysical seismic imaging problem where we invert a grid with $300\times50$ nodes, corresponding to a posterior with approximately $80,000$ parameters.

stat.CO↗

arXiv2026

Sparsifying transform priors in Gaussian graphical models

Bayesian methods constitute a popular approach for estimating the conditional independence structure in Gaussian graphical models, since they can quantify the uncertainty through the posterior distribution. Inference in this framework is typically carried out with Markov chain Monte Carlo (MCMC). However, the most widely used choice of prior distribution for the precision matrix, the so called G-Wishart distribution, suffers from an intractable normalizing constant, which gives rise to the problem of double intractability in the updating steps of the MCMC algorithm. In this article, we propose a new class of prior distributions for the precision matrix, termed ST priors, that allow for the construction of MCMC algorithms that do not suffer from double intractability issues. A realization from an ST prior distribution is obtained by applying a sparsifying transform on a matrix from a distribution with support in the set of all positive definite matrices. We carefully present the theory behind the construction of our proposed class of priors and also perform some numerical experiments, where we apply our methods on a human gene expression dataset. The results suggest that our proposed MCMC algorithm is able to converge and achieve acceptable mixing when applied on the real data.

stat.ME↗

arXiv2025

An MCMC hypothesis test to check a claimed sampler: applied to a claimed sampler for the G-Wishart distribution

Suppose we have a distribution of interest, with density $p(x),x\in {\cal X}$ say, and an algorithm claimed to generate samples from $p(x)$. Moreover, assume we have available a Metropolis--Hastings transition kernel fulfilling detail balance with respect to $p(x)$. In such a situation we formulate a hypothesis test where $H_0$ is that the claimed sampler really generates correct samples from $p(x)$. We use that if initialising the Metropolis--Hastings algorithm with a sample generated by the claimed sampler and run the chain for a fixed number of updates, the initial and final states are exchangeable if $H_0$ is true. Combining this idea with the permutation strategy we define a natural test statistic and a valid p-value. Our motivation for considering the hypothesis test situation is a proposed sampler in the literature, claimed to generate samples from G-Wishart distribution. As no proper proof for the validity of this sampler seems to be available, we are exactly in the hypothesis test situation discussed above. We therefore apply the defined hypothesis test to the claimed sampler. For comparison we also apply the hypothesis test to a known exact sampler for a subset of G-Wishart distributions. The obtained p-values clearly show that the sampler claimed to be able to generate samples from any G-Wishart distribution is in fact not sampling from the specified distribution. In contrast, and as one should expect, the p-values obtained when using the known exact algorithm does not indicate any problems.

stat.CO↗

arXiv2023

A sparse matrix formulation of model-based ensemble Kalman filter

We introduce a computationally efficient variant of the model-based ensemble Kalman filter (EnKF). We propose two changes to the original formulation. First, we phrase the setup in terms of precision matrices instead of covariance matrices, and introduce a new prior for the precision matrix which ensures it to be sparse. Second, we propose to split the state vector into several blocks and formulate an approximate updating procedure for each of these blocks. We study in a simulation example the computational speedup and the approximation error resulting from using the proposed approach. The speedup is substantial for high dimensional state vectors, allowing the proposed filter to be run on much larger problems than can be done with the original formulation. In the simulation example the approximation error resulting from using the introduced block updating is negligible compared to the Monte Carlo variability inherent in both the original and the proposed procedures.

stat.ME↗

arXiv2021

A generalised and fully Bayesian framework for ensemble updating

We propose a generalised framework for the updating of a prior ensemble to a posterior ensemble, an essential yet challenging part in ensemble-based filtering methods. The proposed framework is based on a generalised and fully Bayesian view on the traditional ensemble Kalman filter (EnKF). In the EnKF, the updating of the ensemble is based on Gaussian assumptions, whereas in our general setup the updating may be based on another parametric family. In addition, we propose to formulate an optimality criterion and to find the optimal update with respect to this criterion. The framework is fully Bayesian in the sense that the parameters of the assumed forecast model are treated as random variables. As a consequence, a parameter vector is simulated, for each ensemble member, prior to the updating. In contrast to existing fully Bayesian approaches, where the parameters are simulated conditionally on all the forecast samples, the parameters are in our framework simulated conditionally on both the data and all the forecast samples, except the forecast sample which is to be updated. The proposed framework is studied in detail for two parametric families: the linear-Gaussian model and the finite state-space hidden Markov model. For both cases, we present simulation examples and compare the results with existing ensemble-based filtering methods. The results of the proposed approach indicate a promising performance. In particular, the filter based on the linear-Gaussian model gives a more realistic representation of the uncertainty than the traditional EnKF, and the effect of not conditioning on the forecast sample which is to be updated when simulating the parameters is remarkable.

stat.ME↗

arXiv2019

Ensemble updating of binary state vectors by maximising the expected number of unchanged components

In recent years, several ensemble-based filtering methods have been proposed and studied. The main challenge in such procedures is the updating of a prior ensemble to a posterior ensemble at every step of the filtering recursions. In the famous ensemble Kalman filter, the assumption of a linear-Gaussian state space model is introduced in order to overcome this issue, and the prior ensemble is updated with a linear shift closely related to the traditional Kalman filter equations. In the current article, we consider how the ideas underlying the ensemble Kalman filter can be applied when the components of the state vectors are binary variables. While the ensemble Kalman filter relies on Gaussian approximations of the forecast and filtering distributions, we instead use first order Markov chains. To update the prior ensemble, we simulate samples from a distribution constructed such that the expected number of equal components in a prior and posterior state vector is maximised. We demonstrate the performance of our approach in a simulation example inspired by the movement of oil and water in a petroleum reservoir, where also a more naïve updating approach is applied for comparison. Here, we observe that the Frobenius norm of the difference between the estimated and the true marginal filtering probabilities is reduced to the half with our method compared to the naïve approach, indicating that our method is superior. Finally, we discuss how our methodology can be generalised from the binary setting to more complicated situations.

stat.ME↗

arXiv2018

A Bayesian model for lithology/fluid class prediction using a Markov mesh prior fitted from a training image

We consider a Bayesian model for inversion of observed amplitude variation with offset (AVO) data into lithology/fluid classes, and study in particular how the choice of prior distribution for the lithology/fluid classes influences the inversion results. Two distinct prior distributions are considered, a simple manually specified Markov random field prior with a first order neighborhood and a Markov mesh model with a much larger neighborhood estimated from a training image. They are chosen to model both horisontal connectivity and vertical thickness distribution of the lithology/fluid classes, and are compared on an offshore clastic oil reservoir in the North Sea. We combine both priors with the same linearised Gaussian likelihood function based on a convolved linearised Zoeppritz relation and estimate properties of the resulting two posterior distributions by simulating from these distributions with the Metropolis-Hastings algorithm. The influence of the prior on the marginal posterior probabilities for the lithology/fluid classes is clearly observable, but modest. The importance of the prior on the connectivity properties in the posterior realisations, however, is much stronger. The larger neighborhood of the Markov mesh prior enables it to identify and model connectivity and curvature much better than what can be done by the first order neighborhood Markov random field prior. As a result, we conclude that the posterior realisations based on the Markov mesh prior appear with much higher lateral connectivity, which is geologically plausible.

stat.AP↗

arXiv2018

A multiple-try Metropolis-Hastings algorithm with tailored proposals

We present a new multiple-try Metropolis-Hastings algorithm designed to be especially beneficial when a tailored proposal distribution is available. The algorithm is based on a given acyclic graph $G$, where one of the nodes in $G$, $k$ say, contains the current state of the Markov chain and the remaining nodes contain proposed states generated by applying the tailored proposal distribution. The Metropolis-Hastings algorithm alternates between two types of updates. The first update type is using the tailored proposal distribution to generate new states in all nodes in $G$ except in node $k$. The second update type is generating a new value for $k$, thereby changing the value of the current state. We evaluate the effectiveness of the proposed scheme in an example with previously defined target and proposal distributions.

stat.CO↗

arXiv2017

Prior specification for binary Markov mesh models

We propose prior distributions for all parts of the specification of a Markov mesh model. In the formulation we define priors for the sequential neighborhood, for the parametric form of the conditional distributions and for the parameter values. By simulating from the resulting posterior distribution when conditioning on an observed scene, we thereby obtain an automatic model selection procedure for Markov mesh models. To sample from such a posterior distribution, we construct a reversible jump Markov chain Monte Carlo algorithm (RJMCMC). We demonstrate the usefulness of our prior formulation and the limitations of our RJMCMC algorithm in two examples.

stat.ME↗

arXiv2015

Approximations and bounds for binary Markov random fields

Discrete Markov random fields form a natural class of models to represent images and spatial data sets. The use of such models is, however, hampered by a computationally intractable normalising constant. This makes parameter estimation and a fully Bayesian treatment of discrete Markov random fields difficult. We apply approximation theory for pseudo-Boolean functions to binary Markov random fields and construct approximations and upper and lower bounds for the associated computationally intractable normalising constant. As a by-product of this process we also get a partially ordered Markov model approximation of the binary Markov random field. We present numerical examples with both the pairwise interaction Ising model and with higher-order interaction models, showing the quality of our approximations and bounds. We also present simulation examples and one real data example demonstrating how the approximations and bounds can be applied for parameter estimation and to handle a fully-Bayesian model computationally.

stat.CO↗

arXiv2015

Prior specification of neighbourhood and interaction structure in binary Markov random fields

In this paper we propose a prior distribution for the clique set and dependence structure of binary Markov random fields (MRFs). In the formulation we allow both pairwise and higher order interactions. We construct the prior by first defining a prior for the neighbourhood system of the MRF, and conditioned on this we define a prior for the appearance of higher order interactions. Finally, for the parameter values we adopt a prior that allows for parameter values to equal, and in this way we reduce the effective number of free parameters. To sample from a resulting posterior distribution conditioned on an observed scene we construct a reversible jump Markov chain Monte Carlo (RJMCMC) algorithm. We circumvent evaluations of the intractable normalising constant of the MRF when running this algorithm by adopting a previously defined approximate auxiliary variable algorithm. We demonstrate the usefulness of our prior in two simulation examples and one real data example.

stat.ME↗

arXiv2015

Fully Bayesian binary Markov random field models: Prior specification and posterior simulation

We propose a flexible prior model for the parameters of binary Markov random fields (MRF) defined on rectangular lattices and with maximal cliques defined from a template maximal clique. The prior model allows higher-order interactions to be included. We also define a reversible jump Markov chain Monte Carlo (RJMCMC) algorithm to sample from the associated posterior distribution. The number of possible parameters for an MRF with for instance k x l maximal cliques becomes high even for small values of k and l. To get a flexible model which may adapt to the structure of a particular observed image we do not put any absolute restrictions on the parametrisation. Instead we define a parametric form for the MRF where the parameters have interpretation as potentials for the various clique configurations, and limit the effective number of parameters by assigning apriori discrete probabilities for events where groups of parameter values are equal. To run our RJMCMC algorithm we have to cope with the computationally intractable normalising constant of MRFs. For this we adopt a previously defined approximation for binary MRFs, but we also briefly discuss other alternatives. We demonstrate the flexibility of our prior formulation with simulated and real data examples.

stat.ME↗