2nd Workshop on Compact Objects, Gravitational Waves and Deep Learning

20 Jun 2022, 08:45 → 23 Jun 2022, 17:00 Europe/Madrid

ADEIT, University of Valencia

Antonio Onofre (LIP), Carlos Herdeiro (Mat. Department, University of Aveiro), Jose Antonio Font (Universitat Valencia)

The Workshop on Compact Objects, Gravitational Waves and Deep Learning, intends to bring together experimental and theoretical Astrophysicists, Relativists and Cosmologists in a close collaboration with High Energy Particle Physicists to explore the Universe from its most fundamental particle constituents, to the events that happen at a Cosmological scale. The workshop is particularly focused in providing young Master and PhD researchers state of the art knowledge on Astrophysics, Cosmology and Particle Physics, with particularly useful Hands-On sessions, using state of the art Machine Learning methods, to explore signals of new physics that can span the boundary between Particle Physics performed at Colliders, Relativistic Astrophysics, Strong Gravity and Cosmology.

ADEIT.png

Monday 20 June

08:45 → 09:00 Opening and Wellcome

Speaker: Jose Antonio Font (Departamento de Astronomía y Astrofísica, Universitat de València, Spain)

09:00 → 10:00 Gravitational Waves Bayesian Analysis and Parameter Estimation

Speaker: Juan Calderon Bustillo (University of Santiago de Compostela, Spain)

Valencia_WS.pdf

10:00 → 11:00 Neutron Stars and Equations of State (EOS)

Speaker: Micaela Oertel (LUTH, Observatoire de Paris, France )

lecture1.pdf

11:00 → 11:30 Coffee Break

11:30 → 12:30 Event Horizon Telescope Results

Speaker: Iván Martí-Vidal (University of Valencia, Spain)

12:30 → 13:30 Compact Objects in Modified Gravity

Speaker: Diego Rubiera-Garcia (Universidad Complutense de Madrid, Spain)

BlackHoles230622D.pdf

13:30 → 15:00 Lunch

15:00 → 17:00 Hands-On Session on EOS

hands-on session on Compose (https://compose.obspm.fr/)

Speaker: Micaela Oertel (Observatoire de Paris)

17:00 → 17:30 Coffee Break

17:30 → 18:30 Hands-On Session on EOS

hands-on session on Compose (https://compose.obspm.fr/)

Speaker: Micaela Oertel (Observatoire de Paris)

Tuesday 21 June

09:00 → 10:00 Gravitational Waves Bayesian Analysis and Parameter Estimation

Speaker: Juan Calderon Bustillo (University of Santiago de Compostela, Spain)

10:00 → 11:00 Modeling Binary Neutron Star Mergers

Speaker: Roberto De Pietri (Parma University and INFN, Italy)

11:00 → 11:30 Coffee Break

11:30 → 12:30 Compact Objects in Modified Gravity

Speaker: Diego Rubiera-Garcia (Universidad Complutense de Madrid, Spain)

12:30 → 13:30 Gravitational Waveforms from Generative Adversarial Networks (GANs)

Speaker: Felipe Freitas (University of Aveiro, Portugal)

FFFreitas / 2nd-Workshop-on-Compact-Objects-Gravitational-Waves-and-Deep-Learning

FFFreitas / Compact-Objects-Gravitational-Waves-and-Deep-Learning- Public

13:30 → 15:00 Lunch

15:00 → 17:00 Hands-On: Gravitational Waveforms from Generative Adversarial Networks (GANs)

Speaker: Felipe Freitas (University of Aveiro, Portugal)

17:00 → 17:30 Coffee Break

17:30 → 18:30 Hands-On: Gravitational Waveforms from Generative Adversarial Networks (GANs)

Speaker: Felipe Freitas (University of Aveiro, Portugal)

Wednesday 22 June

09:00 → 10:00 Gravitational Waves Bayesian Analysis and Parameter Estimation

Speaker: Juan Calderon Bustillo (University of Santiago de Compostela, Spain)

10:00 → 11:00 Core-Collapse Supernovae and Light Curves

Speaker: MARTIN FRANZ OBERGAULINGER (University of Valencia, Spain)

SN-Lectures.pdf

11:00 → 11:30 Coffee Break

11:30 → 12:30 Gravitational Waves Data Analysis

Speaker: Alejandro Torres-Forné (University of Valencia)

12:30 → 13:30 Gravitational Waves Data Analysis

Speaker: Alejandro Torres-Forné (University of Valencia)

13:30 → 15:00 Lunch

15:00 → 15:20 Equatorial timelike circular orbits around generic ultracompact objects

For a stationary, axisymmetric, asymptotically flat, ultra-compact [i.e. containing light-rings (LRs)] object, with a Z2 north-south symmetry fixing an equatorial plane, we establish that the structure of timelike circular orbits (TCOs) in the vicinity of the equatorial LRs, for either rotation direction, depends exclusively on the radial stability of the LRs. Thus, an unstable LR delimits a region of unstable TCOs (no TCOs) radially above (below) it; a stable LR delimits a region of stable TCOs (no TCOs) radially below (above) it. Corollaries are discussed for both horizonless ultra-compact objects and black holes. We illustrate these results with a variety of exotic stars examples and non-Kerr black holes, for which we also compute the efficiency associated with converting gravitational energy into radiation by a material particle falling under an adiabatic sequence of TCOs. For most objects studied, it is possible to obtain efficiencies larger than the maximal efficiency of Kerr black holes, i.e. larger than 42%.

Speaker: Jorge Delgado

EquatorialTCOs_Talk.pdf

15:20 → 15:40 Black Holes spontaneous scalarization

Spontaneous scalarisation of electrically charged, asymptotically flat Reissner–Nordstrom black holes (BHs) has been recently demonstrated to occur in Einstein-Maxwell–Scalar (EMS) models. This phenomenon is allowed by a non-minimal coupling between the scalar and the Maxwell fields and does not require non-minimal couplings of the scalar field to curvature invariants. EMS BH scalarisation presents a technical simplification over the BH scalarisation that has been conjectured to occur in extended scalar-tensor Gauss-Bonnet (eSTGB) models. It is then natural to ask: (1) how universal are the conclusions extracted from the EMS model? And (2) how much do these conclusions depend on the choice of the non-minimal coupling function?

Speaker: Alexandre Pombo

AlexandreP_SpontScal.pdf

15:40 → 16:00 Quantitative comparison of deep-learning network architectures for gravitational wave parameter inference

Quick and accurate parameter inference in gravitational wave (GW) data is highly desirable for online searches of GW sources. For this purpose, the usage of machine learning and deep learning algorithms is an attractive possibility. However, while general time-series classification has been extensively studied in literature, there being a number of architectures designed specifically with this task in mind, the usage of these techniques to obtain some generating parameters is comparatively rare. In this talk, I will discuss some of the leading architectures for time series classification and present a performance comparison of these networks when adapted to a GW parameter estimation task.

Speaker: Osvaldo Freitas (University of Minho)

GW_workshop_vlc.pdf

16:00 → 16:20 Deep-Learning Inference of Rotational Core-Collapse Supernovae with Numerically-Generated Gravitational-Wave Signals

The gravitational collapse of the core of massive stars and the subsequent explosion of such stars as supernovae (CCSN in the following) is one of the most interesting sources of gravitational waves expected to be detected in the coming years. CCSN hold great scientific interest as the analysis of their complex waveforms can potentially provide valuable information about the underlying physical processes operating during the gravitational collapse of the iron cores of massive stars. Deep Learning techniques will be used to analyze rotational CCSN signals by classifying and correctly separating them from transient sources of noise of the detectors (i.e. glitches). The objective of this analysis will be to estimate some physical parameters of the systems that produce the signals. To detect the presence of astrophysical signals from rotational CCSN in noisy detector data, a network will be trained using a dataset of Time Series from CCSN waveforms injected into the real noise of the Advanced LIGO and Advanced Virgo detectors (corresponding to the O3 observing run). Another network will estimate the physical parameters of the system, specifically, the amplitude of the gravitational signal and the maximum (peak) frequency.

Speaker: Solange Nunes

Deep-Learning Inference of CCSN.pdf

16:20 → 16:40 Core-collapse supernovae from red super giant stars

Supernova (SN) explosions are one of the most energetic events in the observable universe.
Given that, they are the best natural laboratories to investigate extreme physical phenomena, that otherwise would not be reproducible on Earth.
During these powerful explosions chemical elements are also produced, that go to enrich the amount of heavy elements in the interstellar medium.
Three-dimensional long-time simulations of core-collapse supernovae (CCSNe) are crucial to better understand the connection between the progenitor star and the supernova remnants.
These studies have been performed using mainly two approaches: (i) a detailed 3D analysis of individual events, e.g. SN 1987A (M\"uller et al. 1991; Orlando et al. 2015, 2020), or (ii) 1D surveys of stars with different masses and initial conditions (Ugliano et al. 2012; Sukhbold et al. 2016; Ertl et al. 2020).
Here, we intend to extend the current 3D models in the fashion of the latter 1D simulations, considering SNe originated by different red super giant (RSG) progenitors with zero-age main-sequence (ZAMS) masses between 12.5𝑀⊙ and 27𝑀⊙.

This study shows a variety of explosion morphologies: all the models (except one) show the development of asymmetries in the ejecta from very early stages.
The growth of these asymmetries can be related to the motion of the central neutron star (NS): it has been observed that the strongest deviations from spherical symmetry are associated with higher kick velocity of the NS.

Speaker: Beatrice Giudici (Universitat de València)

giudici_slides.pdf

16:40 → 17:00 A new model for the turbulent stress tensors from simulations of the magnetorotational instability

The study of some astrophysical scenarios such as binary neutron star mergers or supernovae can only be fully understood with the aid of numerical-relativity simulations. Insufficient numerical resolution of current grid-based simulations prevents the correct description of the development of turbulence at small scales, which is a key mechanism in instabilities like the magnetorotational instability (MRI) or the Kelvin-Helmholtz instability (KHI), responsible, for example, for the amplification of weak magnetic fields in hypermassive neutron stars, the transient compact objects that are produced after the collision of two neutron stars in compact binaries. A possible remedy is to use subgrid models to reproduce the effects of the turbulence at small scales in terms of the resolved scales, in order to capture the physics of the problem with affordable computational resources. Here we discuss the performance of a new subgrid models using box simulations of the MRI.

Speaker: Miquel Miravet-Tenés (University of Valencia)

model-stress-tensors.pdf

17:00 → 17:30 Coffee Break

17:30 → 17:50 A numerical-relativity gravitational-wave catalogue of spinning Proca-star collisions

We have performed a systematic study of the dynamics and the emission of gravitational radiation in head-on collisions of dynamically robust spinning vector boson stars, aka Proca stars. We find that the wave-like nature of bosonic stars has large impact on the gravitational-wave emission. The energy emitted in gravitational waves critically depends on the difference between the oscillation frequencies of the primary and secondary stars Δ𝜔/𝜇=(𝜔1−𝜔2)/𝜇 in a non-monotonic way. In the unequal-mass case we observe a periodic modulation of the radiated energy as a function of 𝜔2/𝜇 of the secondary star with fixed 𝜔1/𝜇 that we relate to constructive and destructive interference due to the interaction of the Proca field with itself.

Speaker: Nicolas Sanchis-Gual

2nd_2022.pdf

17:50 → 18:10 Impact of ultralight bosonic dark matter on the dynamical bar-mode instability of rotating neutron stars

We investigate the effects ultralight bosonic field dark matter may have on the dynamics of unstable differentially-rotating neutron stars prone to the bar-mode instability. To this aim we perform numerical simulations in general relativity of rotating neutron stars accreting an initial spherically symmetric bosonic field cloud, solving the Einstein-(complex, massive) Klein Gordon-Euler and the Einstein-(complex) Proca-Euler systems. We find that the presence of the bosonic field can critically modify the development of the bar-mode instability of neutron stars, depending on the total mass of the bosonic field and on the boson particle mass. In some cases, the accreting bosonic field can even quench the dominant l = m = 2 mode of the bar deformation by dynamically forming a mixed (fermion-boson) star that retains part of the angular momentum of the original neutron star. However, the mixed star undergoes the development of a mixed bar that leads to significant gravitational-wave emission, substantially different to that of the isolated neutron star. Our results indicate that dark-matter accretion in neutron stars could change the frequency of the expected emission of the bar-mode instability, which would have an important impact on ongoing searches for continuous gravitational waves.

Speaker: Fabrizio Di Giovanni (University of Valencia)

Talk_Workshop.tar.xz

18:10 → 18:30 Fermion-axion stars: static solutions and numerical evolutions

The non-linear stability analysis of mixed stars made by fermionic and bosonic matter is a branch of investigation which allows us to prove the existence of neutron stars with accreted bosonic dark matter. Such hypothetical fermion-boson stars are gravitationally bound solutions of the Einstein-Klein-Gordon-Euler theory where we modeled the bosonic matter as a complex scalar field with a specific potential 𝑉(𝜙). The aim of our work is to study the non-linear stability using a particular choice of the scalar potential: the axion-like particle potential. Changing the potential of the bosonic matter, it is possible to have a completely different behavior of the solutions leading to the possibility to have new stable regions in the parameters space. The fermion-boson models are defined by the central rest-mass density for the fermionic matter 𝜌𝑐 and the central value of the scalar field 𝜙𝑐; in addition, the axion-like particle potential depends on two parameters: the mass of the scalar field 𝑚𝑎 and the decay constant 𝑓𝑎. After building such spherically-symmetric models called fermion-axion stars, we perform numerical relativity simulations to evolve and analyze them in the non-linear regime, confirming that the linearly stable configurations remain stable. We observe three different fates for the unstable models: there are models which collapse to black holes, others that migrate to different stable configurations and finally in a small region of the parameter space very close to the first unstable branch of pure axion stars there are models where the scalar field disperses, bringing with itself also the small fraction of baryonic matter forming those configurations.

Speaker: Davide Guerra (University of Valencia)

FAS_slides.pdf

Thursday 23 June

09:00 → 10:00 Hands-On: Gravitational Waves Bayesian Analysis and Parameter Estimation

Speaker: Juan Calderon Bustillo (University of Santiago de Compostela, Spain)

10:00 → 11:00 Electromagnetic Phenomenology of Neutron Stars (pulsars and magntars)

Speaker: Michael Gabler (University of Valencia, Sapin)

EM_NS_Gabler.pdf

11:00 → 11:30 Coffee Break

11:30 → 12:30 Asteroseismology of Proto-Neutron Stars

Speaker: Pablo Cerdá-Durán (Departamento de Astronomía y Astrofísica, Universitat de València, Spain)

Cerda-Duran.pdf

12:30 → 13:30 Ultra-Light Particles

Speaker: António Morais (University of Aveiro)

Ultralight-1.pdf

13:30 → 15:00 Summary Talk and Lunch

Speaker: Jose Antonio Font (Departamento de Astronomía y Astrofísica, Universitat de València, Spain)