Origin and evolution of the universe
The thesis reflects broad interests across cosmological theory, perturbations, and early-universe dynamics, combining analytical modeling with data-facing implications.
Research
From cosmology and gravity to machine learning and statistical inference.
Darsh’s research spans theoretical physics, cosmological data analysis, and ML-driven scientific methods. The unifying thread is first-principles reasoning paired with practical computational techniques.
Research arc
Early work focused on gravity, black holes, and cosmological perturbations; recent work includes deep learning methods for CMB reconstruction and statistical inference.
Output
9 peer-reviewed publications across JCAP, Physical Review D, Open Journal of Astrophysics, and related journals.
9Peer-reviewed publications
DPhilOxford thesis on the origin and evolution of the universe
ML + PhysicsApplied deep learning to core cosmology data problems
Thesis
Doctoral work
Publications
Selected papers and results
Inpainting CMB maps using Partial Convolutional Neural Networks
Introduces PCNN-based inpainting for masked CMB maps, reaching very high reconstruction fidelity and showing the practical value of deep learning in precision cosmology.
Initial conditions of the universe: Decaying tensor modes
Explores observational consequences of decaying tensor modes without locking to a specific early-universe model, and studies resulting CMB signatures.
Screened fifth forces in parity breaking correlation functions
Computes modified-gravity signatures in parity-breaking galaxy correlations, linking screening physics to potentially observable large-scale structure effects.
Initial conditions of the universe: A sign of the sine mode
Analyzes decaying scalar perturbation modes and quantifies their potential CMB impact through both analytical development and numerical treatment.
Parameter dependent covariance and cosmological inference
Tests when covariance-matrix parameter dependence materially impacts parameter estimation, showing practical regimes where simplifying assumptions remain reliable.
A self-consistency check for unitary propagation of Hawking quanta
Develops a consistency framework around assumptions in the black-hole information problem, focusing on unitarity constraints in Hawking-radiation propagation.
Memory effect from supernova neutrino shells
Studies how relativistic neutrino shells from supernovae perturb geometry and generate measurable signatures in pulsar scintillation and interferometric setups.
Pulsar acceleration shifts from nearby supernova explosions
Shows characteristic timing-derivative signatures in pulsar signals induced by nearby supernova events, outlining potential detectability with stable millisecond pulsars.
Fixed-point alloy cells for in situ thermocouple calibration
Presents stable zinc-bismuth and aluminum-indium monotectic alloy fixed-point cells with robust phase-transition plateaus for calibration workflows.