The Event Horizon Telescope recently observed the first shadow of a black hole. Images like this can potentially be used to test or constrain theories of gravity and deepen the understanding in plasma physics at event horizon scales, which requires accurate parameter estimations. In this work, we present Deep Horizon, two convolutional deep neural networks that recover the physical parameters from images of black hole shadows. We investigate the effects of a limited telescope resolution and observations at higher frequencies. We trained two convolutional deep neural networks on a large image library of simulated mock data. The first network is a Bayesian deep neural regression network and is used to recover the viewing angle $i$, and position angle, mass accretion rate $\dot{M}$, electron heating prescription $R_{\rm high}$ and the black hole mass $M_{\rm BH}$. The second network is a classification network that recovers the black hole spin $a$. We find that with the current resolution of the Event Horizon Telescope, it is only possible to accurately recover a limited number of parameters of a static image, namely the mass and mass accretion rate. Since potential future space-based observing missions will operate at frequencies above 230 GHz, we also investigated the applicability of our network at a frequency of 690 GHz. The expected resolution of space-based missions is higher than the current resolution of the Event Horizon Telescope, and we show that Deep Horizon can accurately recover the parameters of simulated observations with a comparable resolution to such missions.

We investigate the use of deep learning in the context of X-ray polarization detection from astrophysical sources as will be observed by the Imaging X-ray Polarimetry Explorer (IXPE), a future NASA selected space-based mission expected to be operative in 2021. In particular, we propose two models that can be used to estimate the impact point as well as the polarization direction of the incoming radiation. The results obtained show that data-driven approaches depict a promising alternative to the existing analytical approaches. We also discuss problems and challenges to be addressed in the near future.

Space asteroseismology reveals that stellar structure and evolution models of intermediate- and high-mass stars are in need of improvement in terms of angular momentum and chemical element transport. We aim to probe the interior structure of a hot massive star in the core-hydrogen burning phase of its evolution. We analyse CoRoT space photometry, Gaia DR2 space astrometry, and high-resolution high signal-to-noise HERMES and HARPS time-series spectroscopy of the slowly rotating B2V star HD 170580. From the time-series spectroscopy we derive $vsini=4\pm 2\,$km\,s$^{-1}$, where the uncertainty results from the complex pulsational line-profile variability. We detect 42 frequencies with amplitude above 5$\sigma$. Among these we identify 5 rotationally split triplets and 1 quintuplet. Asteroseismic modelling based on CoRoT, Gaia DR2 and spectroscopic data leads to a star of 8M$_\odot$ near core-hydrogen exhaustion and an extended overshoot zone. The detected low-order pressure-mode frequencies cannot be fit within the uncertainties of the CoRoT data by models without atomic diffusion. Irrespective of this limitation, the low-order gravity modes reveal HD 170580 to be a slow rotator with an average rotation period between 73 and 98 d and a hint of small differential rotation. Future Gaia DR3 data taking into account the multiplicity of the star, along with long-term TESS photometry would allow to put better observational constraints on the asteroseismic models of this blue evolved massive star. Improved modelling with atomic diffusion, including radiative levitation, is needed to achieve compliance with the low helium surface abundance of the star. This poses immense computational challenges but is required to derive the interior rotation and mixing profiles of this star.

We develop a novel method based on machine learning principles to achieve optimal initiation of CPU-intensive computations for forward asteroseismic modeling in a multi-D parameter space. A deep neural network is trained on a precomputed asteroseismology grid containing about 62 million coherent oscillation-mode frequencies derived from stellar evolution models. These models are representative of the core-hydrogen burning stage of intermediate-mass and high-mass stars. The evolution models constitute a 6D parameter space and their predicted low-degree pressure- and gravity-mode oscillations are scanned, using a genetic algorithm. A software pipeline is created to find the best fitting stellar parameters for a given set of observed oscillation frequencies. The proposed method finds the optimal regions in the 6D parameters space in less than a minute, hence providing the optimal starting point for further and more detailed forward asteroseismic modeling in a high-dimensional context. We test and apply the method to seven pulsating stars that were previously modeled asteroseismically by classical grid-based forward modeling based on a $\chi^2$ statistic and obtain good agreement with past results. Our deep learning methodology opens up the application of asteroseismic modeling in +6D parameter space for thousands of stars pulsating in coherent modes with long lifetimes observed by the $Kepler$ space telescope and to be discovered with the TESS and PLATO space missions, while applications so far were done star-by-star for only a handful of cases. Our method is open source and can be used by anyone freely.

We present a new method to interpret the $\gamma$-ray data of our inner Galaxy as measured by the Fermi Large Area Telescope (Fermi LAT). We train and test convolutional neural networks with simulated Fermi-LAT images based on models tuned to real data. We use this method to investigate the origin of an excess emission of GeV $\gamma$-rays seen in previous studies. Interpretations of this excess include $\gamma$ rays created by the annihilation of dark matter particles and $\gamma$ rays originating from a collection of unresolved point sources, such as millisecond pulsars. Our new method allows precise measurements of the contribution and properties of an unresolved population of $\gamma$-ray point sources in the interstellar diffuse emission model.

The Fermi Collaboration has recently updated their analysis of gamma rays from the center of the Galaxy. They reconfirm the presence of an unexplained emission feature which is most prominent in the region of $1-10$ GeV, known as the Galactic Center GeV excess (GCE). Although the GCE is now firmly detected, an interpretation of this emission as a signal of self-annihilating dark matter (DM) particles is not unambiguously possible due to systematic effects in the gamma-ray modeling estimated in the Galactic Plane. In this paper we build a covariance matrix, collecting different systematic uncertainties investigated in the Fermi Collaboration's paper that affect the GCE spectrum. We show that models where part of the GCE is due to annihilating DM can still be consistent with the new data. We also re-evaluate the parameter space regions of the minimal supersymmetric Standard Model (MSSM) that can contribute dominantly to the GCE via neutralino DM annihilation. All recent constraints from DM direct detection experiments such as PICO, LUX, PandaX and Xenon1T, limits on the annihilation cross section from dwarf spheroidal galaxies and Large Hadron Collider limits are considered in this analysis. Due to a slight shift in the energy spectrum of the GC excess with respect to the previous Fermi analysis, and the recent limits from direct detection experiments, we find a slightly shifted parameter region of the MSSM compared to our previous analysis that is consistent with the GCE. Neutralinos with a mass between $85-220$ GeV can describe the excess via annihilation into a pair of $W$-bosons or top quarks. Remarkably, there are low fine-tuning models among the regions that we have found. The complete set of solutions will be probed by upcoming direct detection experiments and with dedicated searches in the upcoming data of the Large Hadron Collider.

Observations with the Fermi Large Area Telescope (LAT) indicate an excess in gamma rays originating from the center of our Galaxy. A possible explanation for this excess is the annihilation of Dark Matter particles. We have investigated the annihilation of neutralinos as Dark Matter candidates within the phenomenological Minimal Supersymmetric Standard Model (pMSSM). An iterative particle filter approach was used to search for solutions within the pMSSM. We found solutions that are consistent with astroparticle physics and collider experiments, and provide a fit to the energy spectrum of the excess. The neutralino is a Bino/Higgsino or Bino/Wino/Higgsino mixture with a mass in the range $84-92$~GeV or $87-97$~GeV annihilating into W bosons. A third solutions is found for a neutralino of mass $174-187$~GeV annihilating into top quarks. The best solutions yield a Dark Matter relic density $0.06 < \Omega h^2 <0.13$. These pMSSM solutions make clear forecasts for LHC, direct and indirect DM detection experiments. If the MSSM explanation of the excess seen by Fermi-LAT is correct, a DM signal might be discovered soon.