Projects

under construction

Velocity Reconstruction with Autodifferentiable Forward Modelling and Optimization

We present an optimal estimator for the peculiar radial velocity field by jointly maximizing the posterior of the CMB temperature anisotropy given galaxy survey induced by the kinetic SZ effect.

Learn More
Velocity Reconstruction with Autodifferentiable Forward Modelling and Optimization

Two-field formalism for a neural network-enhanced non-Gaussianity search with halos

We introduce a new, neural-net-enhanced method for measuring the local non-Gaussianity \(f_{NL}\) from the dark matter halos with the estimator that reconstructs the local value of \(\sigma_8\) and the large-scale linear dark matter field

Learn More
Two-field formalism for a neural network-enhanced non-Gaussianity search with halos

Reconstruction of the non-linear cosmological density fields from galaxies

We develop a new method of reconstruction of the cosmological fields, such as dark matter and baryon density, from the observed luminous galaxies

Learn More
Reconstruction of the non-linear cosmological density fields from galaxies

KSZ Velocity Reconstruction with ACT and DESI‑LS using a Tomographic QML Power Spectrum Estimator

We perform a reconstruction of the radial velocity with the kSZ effect. We use the ACT-DR6 CMB data and the DESI-LS galaxy sample. We find a strong cross correlation detection signifficance (\(\sim 11 \sigma\))

Learn More
KSZ Velocity Reconstruction with ACT and DESI‑LS using a Tomographic QML Power Spectrum Estimator

TP Bench – Theoretical Physics Benchmark for AI

A curated dataset and evaluation suite designed to measure the reasoning capabilities of AI models in theoretical physics

Learn More
TP Bench – Theoretical Physics Benchmark for AI