Research atlas

Cosmic history,
read at many scales.

Two connected tracks: cosmology across cosmic history, and interdisciplinary inference where the same numerical ideas meet other complex systems.

Primary programme

Cosmology

01 / 05

Inflation · primordial features · non-Gaussianity

The early Universe

I work on inflationary models and the primordial perturbations they generate. A recurring question is whether localized departures from a smooth primordial spectrum—suppression, oscillations or other features—can illuminate the dynamics of inflation.

This includes exact numerical solutions of scalar and tensor perturbations, two-field models, primordial black-hole formation, stochastic gravitational waves and the bispectrum. Whipped Inflation, Wiggly Whipped Inflation and the BINGO code grew out of this programme.

Related publications →

Four classes of primordial power-spectrum features generated by Wiggly Whipped Inflation — A feature atlas from Wiggly Whipped Inflation: suppression, localized structure, extended oscillations and mixed signatures. Paper & figure source ↗

FIG. 01AJCAP 2020

Predicted stochastic gravitational-wave spectra across pulsar timing, space-based and terrestrial detector windows

Across detector windowsPredicted stochastic gravitational-wave spectra from a two-field inflation model, shown against sensitivity windows from pulsar timing arrays to terrestrial interferometers. Open paper ↗

FIG. 01BJCAP 2013

BINGO numerical result for the non-Gaussianity parameter over triangular wave-vector configurations

Bispectrum over triangle spaceThe numerical BINGO result for the non-Gaussianity parameter across allowed triangular wave-vector configurations in quadratic inflation. Open paper ↗

02 / 05

CMB · inverse problems · parameter estimation

Cosmological inference

Model comparison and reconstruction are complementary. I develop statistical methods that reconstruct primordial physics directly from cosmological observations, then use those hints to build and test models.

The work spans modified Richardson–Lucy deconvolution, optimized free-form binning, Gaussian processes and Bayesian parameter estimation. Recent analyses examine the Planck–ACT discordance and how different CMB datasets map onto the primordial spectrum.

Related publications →

Primordial power spectrum reconstructions from Planck and ACT — Free-form reconstructions show where Planck PR3 and ACT DR4 prefer different primordial structure. Paper & figure source ↗

FIG. 02AJCAP 2014

Color-coded Planck primordial power-spectrum reconstructions and their corresponding CMB temperature angular power spectra

Reconstructed PPS and angular spectraColour traces the improvement in Planck likelihood across the reconstruction: each primordial spectrum is paired with its predicted CMB temperature spectrum and residuals. Open paper ↗

03 / 05

Cosmic dawn · ionization history · neutral hydrogen

Reionization

Reionization connects the first luminous sources with the CMB and the intergalactic medium. I combine CMB optical depth, UV luminosity density and neutral-hydrogen observations to reconstruct this history with as few imposed shapes as possible.

Poly-Reion provides flexible, physical histories; other work reconstructs the source and sink terms of the ionization equation and applies Gaussian processes to the evolving ionized fraction.

Related publications →

Reconstructed ionized fraction of hydrogen across redshift from CMB and astrophysical observations — Free-form reionization histories jointly constrained by CMB polarization, UV luminosity density and neutral-hydrogen data. Paper & figure source ↗

FIG. 03AJCAP 2024

Galaxy ultraviolet luminosity-function data at redshift four with Schechter and Gaussian-process fits

UV luminosity-function fit at z = 4The Gaussian-process fit follows the observed bright-galaxy excess that the Schechter form misses; the blue and orange curves show posterior samples of the two fits. Open paper ↗

04 / 05

Galaxy clustering · EFTofLSS · primordial features

Large-scale structure

The late-time matter distribution carries a transformed record of primordial physics. I study how inflationary features propagate through nonlinear evolution into galaxy clustering and halo abundance.

The programme includes Lyman-α forest statistics and isotropy tests across survey footprints, halo formation, fluid and N-body simulations, and effective-field-theory modelling of large-scale structure, with connections to Euclid and future surveys.

Related publications →

Linear and logarithmic primordial features evolved with linear theory, effective field theory, infrared resummation and N-body calculations — Primordial-feature signals evolved to redshift zero: one-loop EFTofLSS with infrared resummation closely tracks the N-body fit. Paper & figure source ↗

Methods across fields

Interdisciplinary

05 / 05

Epidemiology · Bayesian inference · public-health data

Interdisciplinary inference

I also adapt the numerical solvers, model comparison and Bayesian inference machinery of cosmology to problems outside the field. ELiXSIR is a fast Fortran extended, zone-linked nine-compartment epidemic solver with age-structured contacts, policy transitions and migration between regions.

In the published INDSCI-SIM study, the framework was applied not only to Delhi, but across Karnataka districts and state, Bengaluru, Chennai, Mumbai, Pune and India as a whole. Joint fits to infections and deaths reconstructed reporting bias, infection fatality ratios and the evolving reproduction number.

Open the ELiXSIR dossier →

ELiXSIR analysis of Karnataka COVID-19 cases, deaths, total infections, reporting bias, fatality ratio and reproduction number — Data, posterior bands and inferred epidemic quantities for Karnataka—one component of the multi-scale India study by Hazra et al. (2022, CC BY). Paper & figure source ↗

Continue to computational work →