?
Postdoctoral Fellow
Research group |?Rorsseland Centre for Solar Physics (RoCS)
Main supervisor |?Tiago Pereira
Co-supervisor |?-
Affiliation |?Institute of Theoretical Astrophysics
Contact |?harshm@uio.no
Short bio
I obtained my PhD in Astronomical Instrumentation from the Indian Institute of Astrophysics, Bengaluru, where I investigated the chromospheric magnetic field through multi-line spectropolarimetric observations and non-LTE modeling of solar spectral lines. My work combined observations, radiative-transfer modeling, and numerical analysis to study the diagnostic potential of chromospheric lines such as Hα and Ca II.
During my doctoral research, I developed expertise in spectropolarimetric diagnostics, forward modeling, and inversion techniques, and contributed to studies on chromospheric shock waves, polarization signatures, and observational constraints on magnetic fields.
I am currently a postdoctoral researcher at the University of Oslo under the DSTrain MSCA fellowship, working at the Rosseland Centre for Solar Physics (RoCS). My research focuses on combining radiative transfer modeling with machine learning methods to improve the interpretation of solar spectropolarimetric observations and to enable next-generation inversion techniques.
Research interests and hobbies
My research aims to improve our ability to infer physical conditions in stellar atmospheres from observations, particularly magnetic fields in the solar chromosphere. I am especially interested in the intersection of radiative transfer, plasma physics, and data science, and in developing computational tools that bridge simulations and observations. My work involves non-LTE modeling, inversion techniques, and machine-learning-based diagnostics for spectropolarimetric data.
I enjoy developing scientific software and exploring new computational approaches for astrophysical problems.
Outside research, I like to spend time with family, nature (treks), and enjoy music and films.
DSTrain project
PRABHA: Facilitating 3D non-LTE inversions through artificial neural networks
Magnetic fields govern many dynamic processes in the solar atmosphere, including flares, winds, and coronal mass ejections. However, they cannot be measured directly and must instead be inferred from spectropolarimetric observations through complex radiative-transfer modeling. With next-generation telescopes such as DKIST and the Swedish Solar Telescope producing high-resolution data, there is an urgent need for faster and more realistic diagnostic tools.
The PRABHA project aims to develop a novel inversion framework capable of performing full 3D non-LTE inversions of chromospheric spectral lines such as Hα and Ca II 8542 ?. The approach combines realistic 3D radiative-MHD simulations with neural networks that act as fast forward solvers for non-LTE populations and radiative transfer. This will allow efficient synthesis of Stokes profiles and response functions while incorporating horizontal radiative transfer effects.
The trained models and inversion framework will be validated using state-of-the-art simulations and spectropolarimetric observations from existing solar telescopes, with future applications to DKIST and SST datasets. By integrating solar physics, computational modeling, and machine learning, PRABHA will help bridge the gap between modern observations and physically realistic modeling, enabling more accurate measurements of chromospheric magnetic fields and advancing our understanding of solar activity and space-weather processes.
Publications
DSTrain publications
Previous publications
?
?
?
?
?
?