Searcharxiv⌕ Search

arXiv subjects

Philipp Frank

Publications and source records attributed to Philipp Frank.

At least 19 recordsLinked to original sources

Forward modeling linear variable filter data: High resolution PAH and Paschen-$α$ maps from SPHEREx

Polycyclic aromatic hydrocarbon (PAH) emission is a powerful tool for the study of the interstellar medium (ISM) in both Galactic and extragalactic contexts. It can reveal ionization state, radiation field, metallicity, composition, and ISM structure as sculpted by processes like turbulence and stellar feedback. Here we use NASA's SPHEREx to produce clean, high-resolution PAH maps at spectral resolution R $\sim$ 35 (compared to WISE at R $\sim$ 2). Prior treatments of the SPHEREx linear variable filter data seek to convert them to more conventional band-averaged maps that can leave striping artifacts associated with spatially variable wavelength sampling, especially at emission lines. In contrast, we deliver a scalable forward-modeling method capable of generating high-quality all-sky SPHEREx emission-line maps at 8.5 arcsec FWHM angular resolution (on 6.4 arcsec pixels), while naturally producing maps without visible striping. We reformat the data into a pixel-level database, mask stellar contamination, and infill missing pixels with a conditional, data-driven, anisotropic Gaussian process. We apply our method to produce 3.3 $μ$m and 3.4 $μ$m PAH and Paschen-$α$ emission maps of a set of fields spanning a range of Galactic and extragalactic conditions including the Rosette nebula, the Corona Australis molecular cloud, M33, and M31. We demonstrate the quality and synergies of our SPHEREx 3.3 $μ$m PAH maps by comparisons with other available gas and dust tracer maps and achieve sensitivities up to $σ(I_{3.3}) = 2.7 \times 10^{-8}$ erg s$^{-1}$ cm$^{-2}$ sr$^{-1}$. We estimate that a two-year extension of the SPHEREx survey, with modest modifications to the scanning strategy, would match or exceed the PAH sensitivity of WISE over the whole sky.

astro-ph.GA↗

Bayesian Image Reconstruction with Spatially Variant PSFs in X-ray Astronomy

X-ray observatories introduce unwanted instrumental effects into photon count data that must be accurately accounted for. In particular, the spatially variant point spread function (PSF) and shot noise pose a non-trivial inverse problem. We apply Bayesian inference grounded in information field theory (IFT) combined with a fast spatially variant PSF representation to infer the posterior mean and uncertainty of the X-ray flux under a minimal set of physically motivated prior assumptions, thereby removing or reducing these instrumental effects. First, ignoring spatial PSF variability, the IFT-based deconvolution is benchmarked on a synthetic example against three variants of the Richardson-Lucy (RL) algorithm using SSIM, RMSE, and NLL as evaluation metrics. Second, the ability of an interpolated patch-based convolution scheme to accurately represent the complex, spatially variant Chandra PSF is assessed on simulated PSF data. Finally, deblurring and spatially variant PSF representation are combined and applied to real Chandra observations of the supernova remnant Cassiopeia A. The IFT-based deconvolution outperforms all tested RL variants on the synthetic benchmark. The patch-based convolution scheme accurately recovers the complex structure of the Chandra PSF. Applied to Cassiopeia A, the reconstructed image appears visually sharper than the exposure-corrected data, and the data residuals are consistent with pure noise, indicating an accurate model of both the instrument and the sky. A comparison with a highly resolved reference dataset confirms this impression. The presented framework enables the removal of spatially variant PSFs from noisy data and opens avenues for future work on cross-calibration between multiple observations or different X-ray instruments, building on the principled uncertainty quantification inherent to the Bayesian framework.

astro-ph.IM↗

Variational Inference Using a Differentiable Multigrid Linear Solver

Gradient-based Bayesian inference methods require efficient access to Jacobian and adjoint-Jacobian operators of high-dimensional forward models. While multigrid solvers provide near-optimal complexity for elliptic partial differential equations, they are rarely available in forms compatible with automatic differentiation (AD). We develop a differentiable multigrid solver for steady-state diffusion-absorption problems and derive its adjoint operations analytically through the full multigrid hierarchy. The resulting solver, DMGS, is implemented in C++ and interfaced with JAX to provide efficient Jacobian-vector and vector-Jacobian products for variational inference in the NIFTy framework. We validate the approach on a 3D inverse problem involving diffuse radiative transfer in tissue, reconstructing an effective radiative source from Monte Carlo-simulated data. The reconstruction reproduces the data at a reduced chi-squared of 1.1 and generalizes to 32 independent validation datasets. Benchmarks against a JAX-native multigrid implementation show comparable runtimes and consistently lower peak memory for the hand-derived adjoint, with modest reverse-mode overhead. These results establish differentiable multigrid solvers as practical building blocks for variational inference in PDE-constrained problems.

math.NA↗

Milky Way Atlas: A radial-velocity-resolved, three-dimensional map of H I within 1.25 kpc

We present a velocity-resolved three-dimensional map of local atomic hydrogen (HI) within 1.25 kpc of the Sun, tackling the challenge of converting emission from position-position-velocity space into true 3D structure. Our method combines the HI4PI full-sky survey with the Edenhofer et al. (2024) 3D dust map in the framework of Information Field Theory, enabling a joint reconstruction of the local HI density, radial velocity field, and effective line width while also separating emission arising inside the mapped local volume from more distant Galactic HI. The inference is driven by morphological matching between dust and HI structures together with kinematic coherence in 3D space. Synthetic data tests show that the method recovers the local density and velocity structure, even in the presence of substantial contamination from distant emission. The resulting map reveals a smoother, more diffuse local HI distribution than the dust, a declining HI-to-dust ratio toward high dust column densities consistent with the atomic-to-molecular transition, and a velocity field that captures both large-scale Galactic rotation and local non-circular velocities. Independent comparisons with maser and young stellar cluster velocities agree with the recovered kinematics. This HI map provides a new three-dimensional, kinematically resolved view of the nearby atomic interstellar medium and a foundation for localising other velocity-resolved Galactic emission in physical space.

astro-ph.GA↗

Application of Bayesian Statistical Tools to SKA Telescopes Polarization Surveys to Study Magnetization of the Large-scale Structure of the Universe

Understanding cosmological magnetic fields requires a detailed knowledge of magnetism in the different environments of the large-scale structure of the Universe. Magnetic fields are well known to inhabit galaxy clusters, and recently their presence has been detected between galaxy clusters, along filaments extending up to 10-15 Mpc. Beyond that, there is limited information on the existence of magnetic fields in sheets and voids of the cosmic web. We propose a Bayesian statistical approach to study magnetic fields on large scales through observations of the Faraday rotation effect in large samples of polarized point-like background radio sources. We present the expectations to detect magnetization in environments of the large-scale structure with the SKA-Mid polarization survey planned by the SKAO Magnetism Science Working Group and with SKA-Low with AA4 telescopes, and discuss the required level of accuracy on the redshifts of the host galaxies for such a study. We find that about 50,000 mid-frequency Faraday rotation measurements complemented by high-precision redshifts are needed to constrain magnetization of dense environments as galaxy clusters. Investigation of magnetization in weakly-magnetized low-density enviroments, as filaments, will remain challenging, but low frequencies radio observations and spectroscopic redhifts for at least 17,000 will allow us to put first constraints.

astro-ph.IM↗

GraphGP: Scalable Gaussian Processes with Vecchia's Approximation

Gaussian processes are a powerful tool for modeling continuous fields, but their naive $\mathcal{O}(N^3)$ computational cost and $\mathcal{O}(N^2)$ memory requirement often limit their practical use. Vecchia's approximation is a sparse precision matrix approximation for stationary, decaying kernels that conditions each point only on its $k$ nearest neighbors. We present GraphGP, a GPU algorithm for Vecchia's approximation that scales to nearly a billion parameters with linear time and memory requirements, handling arbitrary point distributions over a large dynamic range. Our key contributions are (1) a bit-reversed k-d tree ordering that allows efficient neighbor searches while also maximizing batch parallelism, and (2) a differentiable CUDA implementation, which is substantially faster and more memory efficient than our pure JAX baseline. GraphGP provides the building blocks for inference, including forward generation, inverse application, log-determinant, and kernel parameter derivatives.

stat.CO↗

The 3D Structure and Kinematics of the Local Disk-Halo Interface: Intermediate-velocity Clouds are the Minority of High-altitude Clouds in the Solar Neighborhood

Studies of the Milky Way's disk-halo interface have historically identified inflowing and outflowing gas incompatible with disk rotation on the basis of radial velocity, leading to the well-known categories of intermediate-velocity clouds (IVCs) and high-velocity clouds (HVCs). In this work, we leverage recent progress in 3D dust mapping of the Solar Neighborhood to perform the first 3D spatial search for anomalous-velocity clouds at the local disk-halo interface. We identify 1,695 dust clouds within 1.25 kpc of the Sun (with altitudes ranging between z=-646 pc to z=+928 pc) by applying a topological structure finding method to a parsec-resolution 3D dust map. We then evaluate the morphological similarity between these clouds and HI 21 cm emission to measure cloud kinematics, and construct a sample of 519 clouds with high-confidence distances, 3D morphologies, and radial velocities. Among these are several IVCs (embedded within the well-known Intermediate Velocity Arch complex) now identified in 3D for the first time, enabling direct measurement of their distances, sizes, densities, masses, pressures, and dust-to-gas ratios. We observe a pronounced asymmetry in the vertical distribution of all clouds in the Solar Neighborhood, with $(2.9 \pm 0.2) \times$ more clouds in the Northern Galactic hemisphere than the Southern above altitudes at which IVCs are present ($|z| \geq 480$ pc). IVCs make up only 18% of the total number of clouds located at these high altitudes, with the remainder having low velocities -- highlighting the importance of accounting for low-radial-velocity structures when evaluating the local disk-halo interface and modeling feedback-driven Galactic fountain flows.

astro-ph.GA↗

A Three-Dimensional Tomographic Reconstruction of the Galactic Cosmic-Ray Proton Density

Cosmic rays (CRs) are a ubiquitous non-thermal component of the interstellar medium (ISM). A data-driven three-dimensional (3D) map of their distribution is essential for understanding CR transport and constraining the spatial distribution of their sources. In this work, we reconstructed the 3D spatial distribution of the Galactic cosmic-ray proton (CRp) density. We model the diffuse gamma-ray emission arising from inelastic hadronic interactions between CRps and interstellar gas. Using a map of dust-correlated diffuse gamma-ray emission based on ten years of Fermi-LAT observations together with a three-dimensional gas density model, we infer the spatial CRp distribution through a morphological matching approach. The logarithmic CRp density field is described by a Gaussian process defined on a spherical-times-radial grid, while both the field and its correlation structure are inferred simultaneously using Iterative Charted Refinement. The posterior distribution of the reconstructed 3D CRp density field is approximated using geometric variational inference. The reconstructed CRp density exhibits a smooth but spatially structured distribution with a limited dynamical range across the Galactic disk. We find a moderate enhancement of the CRp density toward the inner Galaxy. The inferred normalization at the Solar position is consistent with local CR measurements by the AMS-02 instrument.

astro-ph.HE↗

DEFROST: Detecting Excess in Faraday Rotation thrOugh Sophisticated analysis Techniques

Understanding origin and evolution of cosmological magnetic fields requires knowledge of magnetic fields in different extragalactic environments. In this context, a powerful tool is the statistical analysis of the Faraday effect on the linear polarization of a sample of radio sources. This effect carries information about the magnetic fields in our Galaxy, extragalactic environments between the sources and the observer, and within the emitting radio source itself. An accurate disentangling of all these components is crucial to characterize magnetic fields in the LSS of the Universe. The significant amount of data delivered by new radio instruments enables the investigation of increasingly weak magnetic fields. However, a trustworthy characterization is only possible with advanced analysis techniques. In this work, we present a new algorithm capable of simultaneously disentangling the Faraday effect due to our Galaxy from extragalactic contributions, by properly taking into account the observing noise. The algorithm takes as an input a catalog of RM complemented by auxiliary information as, e.g., the redshift. We tested the algorithm with synthetic data to assess its performance and identify the range of Galactic magnetic field power spectrum slopes that allows us to properly disentangle Galactic and extragalactic terms. Furthermore, we tested the algorithm with synthetic catalogs, based on m- and cm-data currently available, corresponding to different observing setups, noise, and cuts in the absolute value of the Galactic latitude of the radio sources. Considering noise values and density of polarized sources consistent with existing catalogs, we demonstrated that the most robust results are obtained with sources with absolute Galactic latitude > 45deg, with inference of the extragalactic parameters at most within 5sigma, both for dispersion in Faraday rotation of ~1 and 10rad/m2.

astro-ph.IM↗

The radial component of the local Galactic magnetic field in 3D

We present a distance-resolved reconstruction of the local line-of-sight Galactic magnetic field, $B_{||}$, by combining a 3D electron density ($n_{e}$) map derived from dust map-informed simulations and a full-sky map of Faraday rotation measure (RM). The forward model evaluates RM on the same 3D grid as the $n_{e}$ map and compares to the Galactic Faraday rotation sky. We infer $B_{||}$ with a Gaussian-process prior whose power spectrum is inferred from the data using geometric variational inference. The result is a local (within 1.25 kpc where $|b|>5^{\circ}$) map of $B_{||}$ with uncertainties. The reconstructed RM sky reproduces prominent features of Faraday rotation sky, with a root mean square average strength of $B_{||}$ of $1.63\pm 0.16$ $μ$G. In face-on views, the magnetic field exhibits coherent patches with alternating sign and hints of kpc-scale modulations, but with significant structure seen on scales of order 100 pc. The $B_{||}$ field is seen to exhibit a 3D power spectrum with an average slope of $-2.73 \pm 0.19$. We validate our $B_{||}$ reconstruction with Galactic pulsars. Predicted RMs (computed by integrating $n_{e}B_{||}$ to each pulsar's distance) correlates with observed RMs, and predicted dispersion measures (DMs) from the $n_{e}$ map also correlate with measured DMs, albeit with significant scatter.

astro-ph.GA↗

A Bayesian Method for Air-Shower Reconstruction using Information Field Theory

The radio detection of extensive air showers provides a powerful method for studying the origin of high-energy cosmic rays. The Low-Frequency Array (LOFAR) offers unprecedentedly detailed measurements of the radio emission footprint. However, fully exploiting this information requires advanced reconstruction techniques. In this paper, we introduce a novel framework for air shower reconstruction based on Bayesian inference and Information Field Theory (IFT). Our method is built on a fully differentiable forward model of the radio signal, which incorporates a physical emission parameterization and a precise wavefront model. Additionally, we augment this physical model with Gaussian processes to account for systematic uncertainties in both the signal fluence and arrival timing. By leveraging gradient information, our approach enables efficient (three orders of magnitude acceleration w.r.t.\ the legacy method) and robust inference of the underlying physical shower parameters, such as primary energy and the depth of shower maximum, $X_\text{max}$. This work provides not only point estimates but also a rigorous quantification of uncertainties. We achieve a resolution in $X_\text{max}$ of $25\,\mathrm{g/cm^2}$ and a radiation energy resolution of $12\%$ on simulations for LOFAR.

astro-ph.HE↗

Correlation as a Resource in Unitary Quantum Measurements

Quantum measurement is a physical process. What physical resources and constraints does quantum mechanics require for measurement to produce the classical world we observe? Treating measurement as a fully unitary quantum process, our goal is to show that objective, redundant, and correctly aligned outcomes are possible iff the environment begins in a specially structured, correlated subspace. We start with a minimal set of assumptions: unitarity, orthogonality of conditional environment branches, and finite-dimensional Hilbert spaces. Using these, we demonstrate that generic environmental states cannot support redundant and mutually consistent records of the signal, the measured quantum system. The admissible initial states form a subspace on which the measurement maps obey the Knill-Laflamme error-correction conditions, revealing that the emergence of classical objectivity relies on the environment behaving like a quantum error-correcting code. The post-measurement subspace naturally factorizes into a ``pointer'' to hold measurement outcomes and ``memory'' to retain pre-measurement quantum information about the environment's state, thereby respecting the no-deletion theorem. This further allows the identification of correlation as a finite resource consumed during measurement. Through an explicit qudit model with local interactions, we demonstrate how correlated environments yield redundant observer networks. Simulations show that record fidelity and redundancy depend on the initial correlations in the environment. This perspective links quantum Darwinism to error correction and raises the possibility that natural processes may prepare and evolutionarily favour environments capable of supporting reliable measurement.

quant-ph↗

aim-resolve: Automatic Identification and Modeling for Bayesian Radio Interferometric Imaging

Modern radio interferometers deliver large volumes of data containing high-sensitivity sky maps over wide fields-of-view. These large area observations can contain various and superposed structures such as point sources, extended objects, and large-scale diffuse emission. To fully realize the potential of these observations, it is crucial to build appropriate sky emission models which separate and reconstruct the underlying astrophysical components. We introduce aim-resolve, an automatic and iterative method that combines the Bayesian imaging algorithm resolve with deep learning and clustering algorithms in order to jointly solve the reconstruction and source extraction problem. The method identifies and models different astrophysical components in radio observations while providing uncertainty quantification of the results. By using different model descriptions for point sources, extended objects, and diffuse background emission, the method efficiently separates the individual components and improves the overall reconstruction. We demonstrate the effectiveness of this method on synthetic image data containing multiple different sources. We further show the application of aim-resolve to an L-band (856 - 1712 MHz) MeerKAT observation of the radio galaxy ESO 137-006 and other radio galaxies in that environment. We observe a reasonable object identification for both applications, yielding a clean separation of the individual components and precise reconstructions of point sources and extended objects along with detailed uncertainty quantification. In particular, the method enables the creation of catalogs containing source positions and brightnesses and the corresponding uncertainties. The full decoupling of sky emission model and instrument response makes the method applicable to a wide variety of instruments or wavelength bands.

astro-ph.IM↗

ReplicationBench: Can AI Agents Replicate Astrophysics Research Papers?

Frontier AI agents show increasing promise as scientific research assistants, and may eventually be useful for extended, open-ended research workflows. However, in order to use agents for novel research, we must first assess the underlying faithfulness and correctness of their work. To evaluate agents as research assistants, we introduce ReplicationBench, an evaluation framework that tests whether agents can replicate entire research papers drawn from the astrophysics literature. Astrophysics, where research relies heavily on archival data and computational study while requiring little real-world experimentation, is a particularly useful testbed for AI agents in scientific research. We split each paper into tasks which require agents to replicate the paper's core contributions, including the experimental setup, derivations, data analysis, and codebase. Each task is co-developed with the original paper authors and targets a key scientific result, enabling objective evaluation of both faithfulness (adherence to original methods) and correctness (technical accuracy of results). ReplicationBench is extremely challenging for current frontier language models: even the best-performing language models score under 20%. We analyze ReplicationBench trajectories in collaboration with domain experts and find a rich, diverse set of failure modes for agents in scientific research. ReplicationBench establishes the first benchmark of paper-scale, expert-validated astrophysics research tasks, reveals insights about agent performance generalizable to other domains of data-driven science, and provides a scalable framework for measuring AI agents' reliability in scientific research.

cs.CL↗

Spatially Coherent 3D Distributions of HI and CO in the Milky Way

The spatial distribution of the gaseous components of the Milky Way is of great importance for a number of different fields, for example, Galactic structure, star formation, and cosmic rays. However, obtaining distance information to gaseous clouds in the interstellar medium from Doppler-shifted line emission is notoriously difficult given our vantage point in the Galaxy. It requires spatial knowledge of gas velocities and generally suffers from distance ambiguities. Previous works often assumed the optically thin limit, had a fixed velocity field, and lacked resolution overall. We aim to overcome these issues and improve previous reconstructions of the gaseous constituents of the interstellar medium of the Galaxy. We used 3D Gaussian processes to model correlations in the interstellar medium, including correlations between different lines of sight, and enforce a spatially coherent structure in the prior. For modelling the transport of radiation from the emitting gas to us as observers, we took absorption effects into account. A special numerical grid ensures that there is high resolution nearby. We inferred the spatial distributions of atomic hydrogen (HI), carbon monoxide (CO), their emission line widths, and the Galactic velocity field in a joint Bayesian inference. We further constrained these fields with complementary data from Galactic masers and young stellar object clusters. Our main result consists of a set of samples that implicitly contain statistical uncertainties. The resulting maps are spatially coherent and reproduce the data with high fidelity. We confirm previous findings regarding the warping and flaring of the Galactic disc. A comparison with 3D dust maps reveals a good agreement on scales larger than approximately 400 pc. While our results are not free of artefacts, they present a big step forward in obtaining high-quality 3D maps of the interstellar medium.

astro-ph.GA↗

The influence of the 3D Galactic gas structure on cosmic-ray transport and $γ$-ray emission

Cosmic rays (CRs) play a major role in the dynamics of the interstellar medium (ISM). Their interactions and transport ionize, heat, and push the ISM thereby coupling different regions of it. The spatial distribution of CRs depends on the distribution of their sources as well as the ISM constituents they interact with, such as gas, starlight, and magnetic fields. Particularly, gas influences CR fluxes and $γ$-ray emission. We illustrate the influence of realistic and largely structured 3D gas distributions on CR transport and $γ$-ray emission, by studying their correlation using the PICARD code and multiple samples of recent 3D reconstructions of the HI and H2 Galactic gas constituents. We adjust the diffusion coefficient $D_{xx}$ and Alfvén speed $v_{\text{A}}$ to reproduce local measurements of B/C abundances and find that these parameters depend non-linearly on the local distribution of gas. When simulating CR transport, the distributions of CR fluxes exhibit energy-dependent structures that vary for all CR species due to their corresponding loss processes. Regions of enhanced secondary (primary) species are spatially correlated (anti-correlated) with the gas density. Furthermore, we show that the morphology of gas clouds alone impacts CR flux predictions. For $γ$-ray emission, we observe a high sensitivity of the $γ$-ray emissivities to gas structures, as these determine the spatial distributions of hadronic interactions and bremsstrahlung. This way, we have for the first time calculated how well-defined uncertainties in a structured gas model propagate to CR transport and $γ$-ray emission.

astro-ph.HE↗

Bayesian Multi-wavelength Imaging of the LMC SN1987A with SRG/eROSITA

The eROSITA Early Data Release (EDR) and eROSITA All-Sky Survey (eRASS1) data have already revealed a remarkable number of undiscovered X-ray sources. Using Bayesian inference and generative modeling techniques for X-ray imaging, we aim to increase the sensitivity and scientific value of these observations by denoising, deconvolving, and decomposing the X-ray sky. Leveraging information field theory, we can exploit the spatial and spectral correlation structures of the different physical components of the sky with non-parametric priors to enhance the image reconstruction. By incorporating instrumental effects into the forward model, we develop a comprehensive Bayesian imaging algorithm for eROSITA pointing observations. Finally, we apply the developed algorithm to EDR data of the Large Magellanic Cloud (LMC) SN1987A, fusing data sets from observations made by five different telescope modules. The final result is a denoised, deconvolved, and decomposed view of the LMC, which enables the analysis of its fine-scale structures, the identification of point sources in this region, and enhanced calibration for future work.

astro-ph.IM↗

Information Field Theory with JAX infers Air Shower Electric Currents from Antenna Signal Traces

Direct imaging of cosmic-ray-induced particle showers during daylight is a long-standing challenge in astroparticle physics. A promising avenue for capturing images of these showers is through the radio emissions generated by their electrically charged particles. Their corresponding current vectors evolve over time as the particle shower propagates through the Earth's atmosphere leading to a characteristic time-dependent electric field in an antenna array. In this work, we harness modern Bayesian inference techniques within the Python toolkit for numerical information field theory NIFTy, coupled with the high-performance numerical computing capabilities of the Python library JAX. This innovative combination enables us to reconstruct the particle shower and its temporal development from data collected by a ground-based antenna array. Our approach opens an initial pathway for detailed imaging of cosmic-ray showers, potentially advancing our understanding of high-energy astrophysical processes.

astro-ph.IM↗