Postdoctoral Fellow
Research group | Rosseland Centre for Solar Physics
Main supervisor | Tiago Pereira
Co-supervisor | -
Affiliation | Institute of Theoretical Astrophysics
Contact | valeriia.liakh@astro.uio.no
Short bio
I received my PhD in Astrophysics from the Instituto de Astrofísica de Canarias (IAC) in 2021. My thesis, titled “Large-Amplitude Oscillations in Quiescent and Active Solar Prominences”, focused on modeling various aspects of prominence dynamics in the pre-eruptive stage, when the structure evolves into a coronal mass ejection (CME), potentially leading to geomagnetic storms if Earth-directed. After my PhD, I worked at the Centre for Mathematical Plasma-Astrophysics at KU Leuven, where I continued modeling prominences using advanced numerical methods.
I am currently a Postdoctoral Fellow at the Rosseland Centre for Solar Physics, where I use different methods to study prominence "tornadoes"—highly dynamic structures whose role as potential CME precursors remains an open question in solar physics. My approach includes numerical modeling and machine learning techniques to better understand their nature and evolution.
Research interests and hobbies
I am interested in solar physics, particularly the dynamics and evolution of solar prominences. My research focuses on modeling prominence eruptions and their role in space weather, with a special emphasis on prominence "tornadoes" and their potential as CME precursors. I use a combination of numerical simulations and machine learning techniques to investigate these complex plasma structures.
Outside of work, I enjoy playing padel and ping pong, ice skating, dancing Latin styles, and traveling.
DSTrain project
Unravelling the Enigma of Solar Prominences Rotation with Magnetohydrodynamic Simulations and Radiative Transfer Using Deep Learning
Prominences, intricate phenomena in the solar corona, are the birthplaces for coronal mass ejections (CMEs), energetic events that trigger geomagnetic storms affecting Earth's magnetosphere and space infrastructure. With a synergy of modern high-resolution observations with advanced numerical modeling, radiative transfer techniques, and data science, we aim to deepen our understanding of prominences and their dynamic behaviors.
Computational modeling and modern data science approaches for synthesizing spectra have become important for interpreting modern observations. A notable gap exists in the 3D numerical modeling of rotational flows within solar prominences, especially including crucial factors like non-adiabatic effects and turbulent heating. Moreover, the 3D non-Local Thermodynamic Equilibrium (NLTE) spectra are essential to connect with observations, but the computational expense of obtaining these for prominence plasma has hindered progress in this area.
In this project, I propose an innovative solution using SunnyNet, an advanced neural network-based tool developed at the Rosseland Center for Solar Physics. SunnyNet offers accelerated calculations of NLTE spectra, reducing computation times from tens of millions of CPU hours to just a few days using GPUs. This breakthrough approach will be applied to 3D simulations of solar "tornadoes" conducted with the MHD code Bifrost, also developed at the Rosseland Center for Solar Physics.
Publications
DSTrain publications
Previous publications