arXiv subjects

Torsten A. Enßlin

Publications and source records attributed to Torsten A. Enßlin.

At least 37 records · Page 2Linked to original sources

arXiv2023

Bayesian radio interferometric imaging with direction-dependent calibration

Context: Radio interferometers measure frequency components of the sky brightness, modulated by the gains of the individual radio antennas. Due to atmospheric turbulence and variations in the operational conditions of the antennas these gains fluctuate. Thereby the gains do not only depend on time but also on the spatial direction on the sky. To recover high quality radio maps an accurate reconstruction of the direction and time-dependent individual antenna gains is required. Aims: This paper aims to improve the reconstruction of radio images, by introducing a novel joint imaging and calibration algorithm including direction-dependent antenna gains. Methods: Building on the \texttt{resolve} framework, we designed a Bayesian imaging and calibration algorithm utilizing the image domain gridding method for numerically efficient application of direction-dependent antenna gains. Furthermore by approximating the posterior probability distribution with variational inference, our algorithm can provide reliable uncertainty maps. Results: We demonstrate the ability of the algorithm to recover high resolution high dynamic range radio maps from VLA data of the radio galaxy Cygnus A. We compare the quality of the recovered images with previous work relying on classically calibrated data. Furthermore we compare with a compressed sensing algorithm also incorporating direction-dependent gains. Conclusions: Including direction-dependent effects in the calibration model significantly improves the dynamic range of the reconstructed images compared to reconstructions from classically calibrated data. Compared to the compressed sensing reconstruction, the resulting sky images have a higher resolution and show fewer artifacts. For utilizing the full potential of radio interferometric data, it is essential to consider the direction dependence of the antenna gains.

astro-ph.IM↗

arXiv2023

Fast-Cadence High-Contrast Imaging with Information Field Theory

Although many exoplanets have been indirectly detected over the last years, direct imaging of them with ground-based telescopes remains challenging. In the presence of atmospheric fluctuations, it is ambitious to resolve the high brightness contrasts at the small angular separation between the star and its potential partners. Post-processing of telescope images has become an essential tool to improve the resolvable contrast ratios. This paper contributes a post-processing algorithm for fast-cadence imaging, which deconvolves sequences of telescope images. The algorithm infers a Bayesian estimate of the astronomical object as well as the atmospheric optical path length, including its spatial and temporal structures. For this, we utilize physics-inspired models for the object, the atmosphere, and the telescope. The algorithm is computationally expensive but allows to resolve high contrast ratios despite short observation times and no field rotation. We test the performance of the algorithm with point-like companions synthetically injected into a real data set acquired with the SHARK-VIS pathfinder instrument at the LBT telescope. Sources with brightness ratios down to $6\cdot10^{-4}$ to the star are detected at $185$ mas separation with a short observation time of $0.6\,\text{s}$.

astro-ph.IM↗

arXiv2022

Sparse Kernel Gaussian Processes through Iterative Charted Refinement (ICR)

Gaussian Processes (GPs) are highly expressive, probabilistic models. A major limitation is their computational complexity. Naively, exact GP inference requires $\mathcal{O}(N^3)$ computations with $N$ denoting the number of modeled points. Current approaches to overcome this limitation either rely on sparse, structured or stochastic representations of data or kernel respectively and usually involve nested optimizations to evaluate a GP. We present a new, generative method named Iterative Charted Refinement (ICR) to model GPs on nearly arbitrarily spaced points in $\mathcal{O}(N)$ time for decaying kernels without nested optimizations. ICR represents long- as well as short-range correlations by combining views of the modeled locations at varying resolutions with a user-provided coordinate chart. In our experiment with points whose spacings vary over two orders of magnitude, ICR's accuracy is comparable to state-of-the-art GP methods. ICR outperforms existing methods in terms of computational speed by one order of magnitude on the CPU and GPU and has already been successfully applied to model a GP with $122$ billion parameters.

cs.LG↗

arXiv2021

Probabilistic Autoencoder using Fisher Information

Neural Networks play a growing role in many science disciplines, including physics. Variational Autoencoders (VAEs) are neural networks that are able to represent the essential information of a high dimensional data set in a low dimensional latent space, which have a probabilistic interpretation. In particular the so-called encoder network, the first part of the VAE, which maps its input onto a position in latent space, additionally provides uncertainty information in terms of a variance around this position. In this work, an extension to the Autoencoder architecture is introduced, the FisherNet. In this architecture, the latent space uncertainty is not generated using an additional information channel in the encoder, but derived from the decoder, by means of the Fisher information metric. This architecture has advantages from a theoretical point of view as it provides a direct uncertainty quantification derived from the model, and also accounts for uncertainty cross-correlations. We can show experimentally that the FisherNet produces more accurate data reconstructions than a comparable VAE and its learning performance also apparently scales better with the number of latent space dimensions.

stat.ML↗

arXiv2021

The Galactic Faraday rotation sky 2020

This work gives an update to existing reconstructions of the Galactic Faraday rotation sky by processing almost all Faraday rotation data sets available at the end of the year 2020. Observations of extra-Galactic sources in recent years have, among other regions, further illuminated the previously under-constrained southern celestial sky, as well as parts of the inner disc of the Milky Way. This has culminated in an all-sky data set of 55,190 data points, which is a significant expansion on the 41,330 used in previous works, hence making an updated separation of the Galactic component a promising venture. The increased source density allows us to present our results in a resolution of about $1.3\cdot 10^{-2}\, \mathrm{deg}^2$ ($46.8\,\mathrm{arcmin}^2$), which is a twofold increase compared to previous works. As for previous Faraday rotation sky reconstructions, this work is based on information field theory, a Bayesian inference scheme for field-like quantities which handles noisy and incomplete data. In contrast to previous reconstructions, we find a significantly thinner and pronounced Galactic disc with small-scale structures exceeding values of several thousand $\mathrm{rad}\,\mathrm{m}^{-2}$. The improvements can mainly be attributed to the new catalog of Faraday data, but are also supported by advances in correlation structure modeling within numerical information field theory. We furthermore give a detailed discussion on statistical properties of the Faraday rotation sky and investigate correlations to other data sets.

astro-ph.GA↗

arXiv2021

Geometric variational inference

Efficiently accessing the information contained in non-linear and high dimensional probability distributions remains a core challenge in modern statistics. Traditionally, estimators that go beyond point estimates are either categorized as Variational Inference (VI) or Markov-Chain Monte-Carlo (MCMC) techniques. While MCMC methods that utilize the geometric properties of continuous probability distributions to increase their efficiency have been proposed, VI methods rarely use the geometry. This work aims to fill this gap and proposes geometric Variational Inference (geoVI), a method based on Riemannian geometry and the Fisher information metric. It is used to construct a coordinate transformation that relates the Riemannian manifold associated with the metric to Euclidean space. The distribution, expressed in the coordinate system induced by the transformation, takes a particularly simple form that allows for an accurate variational approximation by a normal distribution. Furthermore, the algorithmic structure allows for an efficient implementation of geoVI which is demonstrated on multiple examples, ranging from low-dimensional illustrative ones to non-linear, hierarchical Bayesian inverse problems in thousands of dimensions.

stat.ME↗

arXiv2021

Field dynamics inference for local and causal interactions

Inference of fields defined in space and time from observational data is a core discipline in many scientific areas. This work approaches the problem in a Bayesian framework. The proposed method is based on statistically homogeneous random fields defined in space and time and demonstrates how to reconstruct the field together with its prior correlation structure from data. The prior model of the correlation structure is described in a non-parametric fashion and solely builds on fundamental physical assumptions such as space-time homogeneity, locality, and causality. These assumptions are sufficient to successfully infer the field and its prior correlation structure from noisy and incomplete data of a single realization of the process as demonstrated via multiple numerical examples.

physics.data-an↗

arXiv2021

Reconstructing non-repeating radio pulses with Information Field Theory

Particle showers in dielectric media produce radio signals which are used for the detection of both ultra-high energy cosmic rays and neutrinos with energies above a few PeV. The amplitude, polarization, and spectrum of these short, broadband radio pulses allow us to draw conclusions about the primary particles that caused them, as well as the mechanics of shower development and radio emission. However, confidently reconstructing the radio signals can pose a challenge, as they are often obscured by background noise. Information Field Theory offers a robust approach to this challenge by using Bayesian inference to calculate the most likely radio signal, given the recorded data. In this paper, we describe the application of Information Field Theory to radio signals from particle showers in both air and ice and demonstrate how accurately pulse parameters can be obtained from noisy data.

astro-ph.HE↗

arXiv2021

Efficient wide-field radio interferometry response

Radio interferometers do not measure the sky brightness distribution directly but rather a modified Fourier transform of it. Imaging algorithms, thus, need a computational representation of the linear measurement operator and its adjoint, irrespective of the specific chosen imaging algorithm. In this paper, we present a C++ implementation of the radio interferometric measurement operator for wide-field measurements which is based on "improved $w$-stacking". It can provide high accuracy (down to $\approx 10^{-12}$), is based on a new gridding kernel which allows smaller kernel support for given accuracy, dynamically chooses kernel, kernel support and oversampling factor for maximum performance, uses piece-wise polynomial approximation for cheap evaluations of the gridding kernel, treats the visibilities in cache-friendly order, uses explicit vectorisation if available and comes with a parallelisation scheme which scales well also in the adjoint direction (which is a problem for many previous implementations). The implementation has a small memory footprint in the sense that temporary internal data structures are much smaller than the respective input and output data, allowing in-memory processing of data sets which needed to be read from disk or distributed across several compute nodes before.

astro-ph.IM↗

arXiv2021

Comparison of classical and Bayesian imaging in radio interferometry

CLEAN, the commonly employed imaging algorithm in radio interferometry, suffers from a number of shortcomings: in its basic version it does not have the concept of diffuse flux, and the common practice of convolving the CLEAN components with the CLEAN beam erases the potential for super-resolution; it does not output uncertainty information; it produces images with unphysical negative flux regions; and its results are highly dependent on the so-called weighting scheme as well as on any human choice of CLEAN masks to guiding the imaging. Here, we present the Bayesian imaging algorithm resolve which solves the above problems and naturally leads to super-resolution. We take a VLA observation of Cygnus~A at four different frequencies and image it with single-scale CLEAN, multi-scale CLEAN and resolve. Alongside the sky brightness distribution resolve estimates a baseline-dependent correction function for the noise budget, the Bayesian equivalent of weighting schemes. We report noise correction factors between 0.4 and 429. The enhancements achieved by resolve come at the cost of higher computational effort.

astro-ph.IM↗

arXiv2020

Towards Bayesian Data Compression

In order to handle large data sets omnipresent in modern science, efficient compression algorithms are necessary. Here, a Bayesian data compression (BDC) algorithm that adapts to the specific measurement situation is derived in the context of signal reconstruction. BDC compresses a data set under conservation of its posterior structure with minimal information loss given the prior knowledge on the signal, the quantity of interest. Its basic form is valid for Gaussian priors and likelihoods. For constant noise standard deviation, basic BDC becomes equivalent to a Bayesian analog of principal component analysis. Using Metric Gaussian Variational Inference, BDC generalizes to non-linear settings. In its current form, BDC requires the storage of effective instrument response functions for the compressed data and corresponding noise encoding the posterior covariance structure. Their memory demand counteract the compression gain. In order to improve this, sparsity of the compressed responses can be obtained by separating the data into patches and compressing them separately. The applicability of BDC is demonstrated by applying it to synthetic data and radio astronomical data. Still the algorithm needs further improvement as the computation time of the compression and subsequent inference exceeds the time of the inference with the original data.

physics.data-an↗

arXiv2020

Probabilistic simulation of partial differential equations

Computer simulations of differential equations require a time discretization, which inhibits to identify the exact solution with certainty. Probabilistic simulations take this into account via uncertainty quantification. The construction of a probabilistic simulation scheme can be regarded as Bayesian filtering by means of probabilistic numerics. Gaussian prior based filters, specifically Gauss-Markov priors, have successfully been applied to simulation of ordinary differential equations (ODEs) and give rise to filtering problems that can be solved efficiently. This work extends this approach to partial differential equations (PDEs) subject to periodic boundary conditions and utilizes continuous Gaussian processes in space and time to arrive at a Bayesian filtering problem structurally similar to the ODE setting. The usage of a process that is Markov in time and statistically homogeneous in space leads to a probabilistic spectral simulation method that allows for an efficient realization. Furthermore, the Bayesian perspective allows the incorporation of methods developed within the context of information field theory such as the estimation of the power spectrum associated with the prior distribution, to be jointly estimated along with the solution of the PDE.

math.NA↗

arXiv2020

Bayesian decomposition of the Galactic multi-frequency sky using probabilistic autoencoders

All-sky observations of the Milky Way show both Galactic and non-Galactic diffuse emission, for example from interstellar matter or the cosmic microwave background (CMB). The different emitters are partly superimposed in the measurements, partly they obscure each other, and sometimes they dominate within a certain spectral range. The decomposition of the underlying radiative components from spectral data is a signal reconstruction problem and often associated with detailed physical modeling and substantial computational effort. We aim to build an effective and self-instructing algorithm detecting the essential spectral information contained Galactic all-sky data covering spectral bands from $γ$-ray to radio waves. Utilizing principles from information theory, we develop a state-of-the-art variational autoencoder specialized on the adaption to Gaussian noise statistics. We first derive a generic generative process that leads from a low-dimensional set of emission features to the observed high-dimensional data. We formulate a posterior distribution of these features using Bayesian methods and approximate this posterior with variational inference. The algorithm efficiently encodes the information of 35 Galactic emission data sets in ten latent feature maps. These contain the essential information required to reconstruct the initial data with high fidelity and are ranked by the algorithm according to their significance for data regeneration. The three most significant feature maps encode astrophysical components: (1) The dense interstellar medium (ISM), (2) the hot and dilute regions of the ISM and (3) the CMB. The machine-assisted and data-driven dimensionality reduction of spectral data is able to uncover the physical features encoding the input data. Our algorithm is able to extract the dense and dilute Galactic regions, as well as the CMB, from the sky brightness values only.

astro-ph.IM↗

arXiv2020

hammurabi X: Simulating Galactic Synchrotron Emission with Random Magnetic Fields

We present version X of the hammurabi package, the HEALPix-based numeric simulator for Galactic polarized emission. Improving on its earlier design, we have fully renewed the framework with modern C++ standards and features. Multi-threading support has been built in to meet the growing computational workload in future research. For the first time, we present precision profiles of hammurabi line-of-sight integral kernel with multi-layer HEALPix shells. In addition to fundamental improvements, this report focuses on simulating polarized synchrotron emission with Gaussian random magnetic fields. Two fast methods are proposed for realizing divergence-free random magnetic fields either on the Galactic scale where a field alignment and strength modulation are imposed, or on a local scale where more physically motivated models like a parameterized magneto-hydrodynamic (MHD) turbulence can be applied. As an example application, we discuss the phenomenological implications of Gaussian random magnetic fields for high Galactic latitude synchrotron foregrounds. In this, we numerically find B/E polarization mode ratios lower than unity based on Gaussian realizations of either MHD turbulent spectra or in spatially aligned magnetic fields.

astro-ph.IM↗

arXiv2020

Metric Gaussian Variational Inference

Solving Bayesian inference problems approximately with variational approaches can provide fast and accurate results. Capturing correlation within the approximation requires an explicit parametrization. This intrinsically limits this approach to either moderately dimensional problems, or requiring the strongly simplifying mean-field approach. We propose Metric Gaussian Variational Inference (MGVI) as a method that goes beyond mean-field. Here correlations between all model parameters are taken into account, while still scaling linearly in computational time and memory. With this method we achieve higher accuracy and in many cases a significant speedup compared to traditional methods. MGVI is an iterative method that performs a series of Gaussian approximations to the posterior. We alternate between approximating the covariance with the inverse Fisher information metric evaluated at an intermediate mean estimate and optimizing the KL-divergence for the given covariance with respect to the mean. This procedure is iterated until the uncertainty estimate is self-consistent with the mean parameter. We achieve linear scaling by avoiding to store the covariance explicitly at any time. Instead we draw samples from the approximating distribution relying on an implicit representation and numerical schemes to approximately solve linear equations. Those samples are used to approximate the KL-divergence and its gradient. The usage of natural gradient descent allows for rapid convergence. Formulating the Bayesian model in standardized coordinates makes MGVI applicable to any inference problem with continuous parameters. We demonstrate the high accuracy of MGVI by comparing it to HMC and its fast convergence relative to other established methods in several examples. We investigate real-data applications, as well as synthetic examples of varying size and complexity and up to a million model parameters.

stat.ML↗

arXiv2020

A Bayesian Model for Bivariate Causal Inference

We address the problem of two-variable causal inference without intervention. This task is to infer an existing causal relation between two random variables, i.e. $X \rightarrow Y$ or $Y \rightarrow X$ , from purely observational data. As the option to modify a potential cause is not given in many situations only structural properties of the data can be used to solve this ill-posed problem. We briefly review a number of state-of-the-art methods for this, including very recent ones. A novel inference method is introduced, Bayesian Causal Inference (BCI), which assumes a generative Bayesian hierarchical model to pursue the strategy of Bayesian model selection. In the adopted model the distribution of the cause variable is given by a Poisson lognormal distribution, which allows to explicitly regard the discrete nature of datasets, correlations in the parameter spaces, as well as the variance of probability densities on logarithmic scales. We assume Fourier diagonal Field covariance operators. The model itself is restricted to use cases where a direct causal relation $X \rightarrow Y$ has to be decided against a relation $Y \rightarrow X$ , therefore we compare it other methods for this exact problem setting. The generative model assumed provides synthetic causal data for benchmarking our model in comparison to existing State-of-the-art models, namely LiNGAM , ANM-HSIC , ANM-MML , IGCI and CGNN . We explore how well the above methods perform in case of high noise settings, strongly discretized data and very sparse data. BCI performs generally reliable with synthetic data as well as with the real world TCEP benchmark set, with an accuracy comparable to state-of-the-art algorithms. We discuss directions for the future development of BCI .

stat.ML↗

arXiv2019

Probing Cosmic Ray Transport with Radio Synchrotron Harps in the Galactic Center

Recent observations with the MeerKAT radio telescope reveal a unique population of faint non-thermal filaments pervading the central molecular zone (CMZ). Some of those filaments are organized into groups of almost parallel filaments, seemingly sorted by their length, so that their morphology resembles a harp with radio emitting "strings". We argue that the synchrotron emitting GeV electrons of these radio harps have been consecutively injected by the same source (a massive star or pulsar) into spatially intermittent magnetic fiber bundles within a magnetic flux tube or via time-dependent injection events. After escaping from this source, the propagation of cosmic ray (CR) electrons inside a flux tube is governed by the theory of CR transport. We propose to use observations of radio harp filaments to gain insight into the specifics of CR propagation along magnetic fields of which there are two principle modes: CRs could either stream with self-excited magneto-hydrodynamical waves or diffuse along the magnetic field. To disentangle these possibilities, we conduct hydrodynamical simulations of either purely diffusing or streaming CR electrons and compare the resulting brightness distributions to the observed synchrotron profiles of the radio harps. We find compelling evidence that CR streaming is the dominant propagation mode for GeV CRs in one of the radio harps. Observations at higher angular resolution should detect more radio harps and may help to disentangle projection effects of the possibly three-dimensional flux-tube structure of the other radio harps.

astro-ph.HE↗

arXiv2019

The primordial magnetic field in our cosmic backyard

We reconstruct the 3D structure of magnetic fields, which were seeded by density perturbations during the radiation dominated epoch of the Universe and later on were evolved by structure formation. To achieve this goal, we rely on three dimensional initial density fields inferred from the 2M++ galaxy compilation via the Bayesian $\texttt{BORG}$ algorithm. Using those, we estimate the magnetogenesis by the so called Harrison mechanism. This effect produced magnetic fields exploiting the different photon drag on electrons and ions in vortical motions, which are exited due to second order perturbation effects in the Early Universe. Subsequently we study the evolution of these seed fields through the non-linear cosmic structure formation by virtue of a MHD simulation to obtain a 3D estimate for the structure of this primordial magnetic field component today. At recombination we obtain a reliable lower limit on the large scale magnetic field strength around $10^{-23} \mathrm{G}$, with a power spectrum peaking at about $ 2\, \mathrm{Mpc}^{-1}h$ in comoving scales. At present we expect this evolved primordial field to have strengthts above $\approx 10^{-27}\, \mathrm{G}$ and $\approx 10^{-29}\, \mathrm{G}$ in clusters of galaxies and voids, respectively. We also calculate the corresponding Faraday rotation measure map and show the magnetic field morphology and strength for specific objects of the Local Universe.

astro-ph.CO↗