Dr Vanessa Graber

Welcome to my website. I am a theoretical astrophysicist and machine-learning practitioner, who studies neutron stars, some of the most exciting objects in our Universe.

night sky showing the milky way and a shooting star

Back

Research

Neutron stars are stellar remnants that form in the supernova explosions of massive stars. Each of these objects contains a mass comparable to that of our Sun within a radius of about ten kilometres and exhibits extreme gravity, high densities, fast rotation and large magnetic fields. Such conditions cannot be recreated on Earth, making neutron stars amazing cosmic laboratories to study matter under extreme conditions.

The extreme nature of neutron stars requires us to combine knowledge from many scientific areas to understand their fascinating properties. As a result, my work crosses traditional discipline boundaries. In particular, I focus on the interface of condensed-matter physics and astrophysics as well as the connections between astronomy and artificial intelligence. More details can be found here:

Back

Back

Condensed-matter physics & astrophysics

One of the most exciting aspects of neutron stars is that quantum mechanics strongly influences their interiors. To better understand this behaviour, I study macroscopic quantum condensates in neutron-star interiors.

Like the Earth, neutron stars are composed of distinct layers. They have a solid crust and a fluid interior that contain neutrons, protons, electrons and possibly exotic particles. In terms of their high densities, neutron stars are very cold, giving the protons and neutrons the special ability to flow without friction: The charged protons form a so-called superconductor, whereas the neutrons are referred to as a superfluid.

These two are exotic versions of quantum states observed in condensed-matter experiments on Earth. From their laboratory counterparts, we know that superconductors and superfluids create vortices that can be envisaged as tiny, rapidly rotating tornadoes. These small structures interact with their surroundings, affecting the large-scale dynamics of the star.

waterspout off the coast of Florida, 1969 — Vortices can be envisaged as tiny, rapidly rotating tornadoes. Image credit: NOAA Photo Library

I research different approaches to include these small-scale effects into theoretical models of neutron stars.

Using techniques that are well-known from standard magnetohydrodynamics, I have for example studied the evolution of the magnetic field in the interior of superconducting neutron stars. I have also presented novel ways that low-temperature laboratory experiments could be used to make progress in understanding neutron star astrophysics. I am currently working with condensed-matter experts to explore these analogies further.

Furthermore, I have been analysing how coupling processes in the interior affect the star's response after a so-called glitch. These sudden spin-ups interrupt the regular spin-down of pulsars and are thought to be a macroscopic manifestation of superfluidity. By connecting the physics on different length scales, I developed predictive models of the glitch rise, showing that assumptions about the microphysics of vortices crucially affect the star's rotational behaviour. Comparing my predictions to the first pulse-to-pulse glitch observations, reported by Palfreyman et al. (2018), I derived constraints on the strength of the frictional mechanisms in the star's interior. An improved analysis of the data, revealing novel details about the internal components of the star, was published in Nature Astronomy.

Using techniques well-known from the study of laboratory superconductors, so-called Ginzburg-Landau models, I have also been exploring the microscale characteristics of the superconducting protons in the neutron star core. Their properties are poorly understood but could significantly impact the stellar magnetism and are thus crucial to understanding the macroscopic magnetic field properties of compact objects. By adapting the Ginzburg-Landau description to the neutron-star interior and connecting it with realistic superfluid parameters and equations of state, my collaborators and I have constructed superconducting phase diagrams and found that the outer core of neutron stars exhibits so-called type-1.5 superconductivity, rather than type-II superconductivity as generally assumed. We are currently exploring the impact this new state has on astrophysical observables.

More information can be found on my publication page.

Back

Back

Astronomy & artificial intelligence

Astronomy has seen extensive investments in the past decade that have led to the development of new telescope facilities that take data across the entire electromagnetic spectrum, detect gravitational waves as well as neutrinos. Neutron stars are the primary targets for such multi-messenger observations. However, the large data volumes make it increasingly difficult to analyse these data by hand. For several upcoming observatories, this will become impossible all together. The astronomy community is, therefore, exploring new scientific approaches.

This is where artificial intelligence (AI) techniques such as machine learning come in. Machine learning is an AI implementation that allows systems to automatically improve and learn from previous experiences without being explicitly told how to do so. They are, thus, ideal to analyse large and complex astronomical datasets.

In particular, I have been using machine learning in the field of neutron-star population synthesis.

Although about a billion neutron stars are expected to exist in our own galaxy, observational constraints limit us to only detecting a small fraction of them; we only know around 3,500 of these compact objects to date. To overcome this gap, we use population synthesis to model the full population theoretically. Based on our current knowledge of input physics, these approaches focus on simulating synthetic neutron-star populations. Once the simulated samples are created, I compare these to real observations using machine-learning techniques to identify discrepancies and subsequently adjust our theoretical models. This global approach allows us to constrain the input physics and learn more about neutron-star properties.

artist impression of a neutron star — Artist impression of a neutron star and its magnetic field. Image credit: ESO/L.Calçada

In a proof of concept study, the MAGNESIA population synthesis team explored the possibility of inferring the properties of the Galactic neutron-star population through deep learning. In particular, we focused on their dynamical characteristics and showed that convolutional neural networks are able to accurately estimate the parameters, which control the positions of synthetic pulsars. Our analysis also highlighted the need for increasing the sample of known neutron stars and accurately classifying them, which is one of the main science drivers for the upcoming Square Kilometer Array.

Recently, I led a study that applied so-called simulation-based inference (SBI) in the context of neutron stars for the first time. We used this new machine-learning technique to infer the parameters that control the magnetic and rotational properties of radio-emitting neutron stars. SBI is crucial for population synthesis as our simulation framework is too complex to use standard Bayesian inference tools, such as Markov chain Monte Carlo. We specifically used a method called neural posterior estimation and successfully trained deep neural networks to infer posterior distributions. This work will form the basis for future multi-wavelength analyses of Galactic neutron-stars that are currently ongoing.

More information can be found on my publication page.

Back

Back

Publications

A full list of papers can be found on the online databases ADSbeta logo ADS, arXiv logo arXiv or ORCID iD icon ORCID. A few selected publications are given below.

J. R. Fuentes and V. Graber, Superfluid Spin-up: 3D Simulations of Postglitch Dynamics in Neutron Stars Cores, Astrophys. J., 974, 300 (2024)
V. Graber, M. Ronchi, C. Pardo-Araujo, and N. Rea, Isolated pulsar population synthesis with simulation-based inference, Astrophys. J., 968, 16 (2024)
S. Ascenzi, V. Graber, and N. Rea, Neutron-star Measurements in the Multi-messenger Era, Astroparticle Phys., 158, 102935 (2024)
M. Ronchi, N. Rea, V. Graber, and N. Hurley-Walker, Long-period pulsars as evidence of supernova fallback accretion, Astrophys. J., 934, 184 (2022)
T. S. Wood and V. Graber, Superconducting phases in neutron star cores, Univ. 8, 228 (2022)
M. Ronchi, V. Graber, A. Garcia-Garcia, J. A. Pons, and N. Rea, Analyzing the Galactic pulsar population with machine learning, Astrophys. J., 916, 100 (2021)
G. Ashton, P. D. Lasky, V. Graber, and J. Palfreyman, Rotational evolution of the Vela pulsar during the 2016 glitch, Nature Astron., 3, 1143 (2019)
V. Graber, A. Cumming, and N. Andersson, Glitch rises as a test for rapid superfluid coupling in neutron stars, Astrophys. J., 865, 23 (2018)
V. Graber, N. Andersson, and M. Hogg, Neutron stars in the laboratory, Intern. J. Mod. Phys. D, 26, 1730015 (2017)
V. Graber, N. Andersson, K. Glampedakis, and S. K. Lander, Magnetic field evolution in superconducting neutron stars, Mon. Not. Roy. Astron. Soc., 453, 671 (2015)

Back

Back

Talks

I have presented my research at a large number of international conferences and given many invited colloquia and seminars over the years. Details are shown below.

Back

Back