[Paper Review] M87* in space, time, and frequency

Philipp Arras, Philipp Frank+5 more|arXiv (Cornell University)|Feb 12, 2020

Radio Astronomy Observations and Technology16 references4 citations

TL;DR

This paper presents a Bayesian imaging algorithm that reconstructs time- and frequency-resolved, 2+1+1 dimensional images of M87* from sparse, low-SNR Very Long Baseline Interferometry (VLBI) data, accounting for temporal evolution and uncertainty. The method infers spatial and temporal correlations directly from data, revealing variable emission structures outside the photon ring on day-timescales.

ABSTRACT

Observing the dynamics of compact astrophysical objects provides insights into their inner workings, thereby probing physics under extreme conditions. The immediate vicinity of an active supermassive black hole with its event horizon, photon ring, accretion disk, and relativistic jets is a perfect pace to study general relativity, magneto-hydrodynamics, and high energy plasma physics. The recent observations of the black hole shadow of M87* with Very Long Baseline Interferometry (VLBI) by the Event Horizon Telescope (EHT) open the possibility to investigate its dynamical processes on time scales of days. In this regime, radio astronomical imaging algorithms are brought to their limits. Compared to regular radio interferometers, VLBI networks typically have fewer antennas and low signal to noise ratios (SNRs). If the source is variable during the observational period, one cannot co-add data on the sky brightness distribution from different time frames to increase the SNR. Here, we present an imaging algorithm that copes with the data scarcity and the source's temporal evolution, while simultaneously providing uncertainty quantification on all results. Our algorithm views the imaging task as a Bayesian inference problem of a time-varying brightness, exploits the correlation structure between time frames, and reconstructs an entire, $\mathbf{2+1+1}$ dimensional time-variable and spectrally resolved image at once. The degree of correlation in the spatial and the temporal domains is inferred from the data and no form of correlation is excluded a priori. We apply this method to the EHT observation of M87* and validate our approach on synthetic data. The time- and frequency-resolved reconstruction of M87* confirms variable structures on the emission ring on a time scale of days. The reconstruction indicates extended and time-variable emission structures outside the ring itself.

Motivation & Objective

• Address the challenge of imaging highly variable, compact astrophysical sources like M87* using sparse, low-SNR VLBI data with limited temporal and spectral resolution.
• Overcome the limitations of traditional imaging algorithms that cannot co-add data across time due to source variability and low signal-to-noise ratios.
• Simultaneously reconstruct a 2+1+1 dimensional image (spatial, temporal, spectral) while quantifying uncertainty in all reconstructed components.
• Infer the correlation structure between time frames and spatial pixels directly from data, without assuming a priori temporal or spatial correlation models.
• Enable robust inference of dynamical processes in the immediate vicinity of M87*, including emission outside the photon ring, using real EHT observations.

Proposed method

• Frame the imaging problem as a Bayesian inference task for a time-varying, spectrally resolved brightness distribution.
• Model the source brightness as a spatio-temporal-spectral field with a Gaussian process prior that captures unknown correlations in space, time, and frequency.
• Use a hierarchical prior to infer the degree of correlation in both spatial and temporal domains directly from the data, without assuming fixed correlation lengths.
• Apply a variational inference approach to efficiently compute the posterior distribution over the full 2+1+1 dimensional image, enabling uncertainty quantification.
• Incorporate the full baseline visibility data from EHT observations, including amplitude and phase information, into the likelihood model.
• Validate the method on synthetic data with known dynamics before applying it to real EHT observations of M87*.

Experimental results

Research questions

• RQ1 Can a Bayesian imaging framework reconstruct time- and frequency-resolved images of M87* from sparse, low-SNR VLBI data while accounting for source variability?
• RQ2 What is the degree of spatial and temporal correlation in the emission structure of M87* as inferred directly from the data?
• RQ3 Are there extended, time-variable emission structures outside the photon ring in M87* on day-timescales?
• RQ4 How does the uncertainty in the reconstructed image vary across space, time, and frequency?
• RQ5 Can the method reliably distinguish between intrinsic source variability and imaging artifacts in low-SNR VLBI data?

Key findings

• The algorithm successfully reconstructs a 2+1+1 dimensional image of M87* from real EHT data, resolving emission structures across space, time, and frequency.
• The reconstruction reveals variable structures on the emission ring with timescales of days, indicating dynamic processes in the accretion flow.
• Extended and time-variable emission features are detected outside the photon ring, suggesting complex emission morphology beyond the ring structure.
• The method provides full uncertainty quantification across all dimensions, enabling confidence assessment of reconstructed features.
• The inferred correlation structure in time and space is data-driven and does not assume fixed correlation lengths, improving robustness to source variability.
• Validation on synthetic data confirms the method's ability to recover known dynamical features under realistic observational conditions.