Research
The cosmic web holds far more information than its simplest statistics reveal. I build the tools to read that extra information, from galaxy surveys and from radio maps of neutral hydrogen.
The idea
For decades, two-point statistics such as the power spectrum have been the cornerstone of large-scale structure studies. The power spectrum measures how strongly matter clusters on each length scale. For a Gaussian field, that is the whole story.
The Universe we observe is not Gaussian. Gravity draws matter into filaments, sheets and clusters, and this non-linear growth leaves imprints that the power spectrum cannot see. Higher-order statistics capture that missing information, so we can learn more from the same galaxy surveys and 21-cm observations.
The bispectrum is the next statistic after the power spectrum, and the first that responds to non-Gaussian features. It matters at both ends of cosmic history. At late times, it traces the non-linearity built up by gravitational instability. In the very early Universe, it constrains primordial non-Gaussianity, which opens a window onto the physics of inflation.
Contributions
Most of this work began during my PhD at IIT Kharagpur. It runs from the first estimators to the first measurements on real radio data, listed from earliest to most recent.
We developed a fast estimator that quantifies how the 3D bispectrum depends on the shape of triangles.
Using simulated signals, we forecast the monopole and quadrupole moments of the 21-cm bispectrum from the Epoch of Reionization. The results show that the bispectrum can be a powerful diagnostic of how reionization began and how it ended.
We measured how the bispectrum monopole of SDSS galaxies depends on the size and shape of triangles. Next, we plan to extend this analysis to higher-order multipoles.
Our Brisk estimator computes all non-zero angular multipoles of the anisotropic redshift-space bispectrum. It applies directly to ongoing and upcoming galaxy surveys such as DESI, Euclid and LSST.
We built a visibility-based estimator for the 21-cm angular bispectrum. It works directly on gridded data from a single frequency channel of radio interferometers such as the uGMRT, MWA and SKA.
We extended this to many frequencies at once, with a fast, FFT-accelerated estimator for the multi-frequency angular bispectrum (MABS) and the 3D 21-cm bispectrum. On simulated MWA observations, it recovers the input signal to within about 20%, mostly consistent with statistical noise.
I laid the theoretical foundations for using the bispectrum multipoles of galaxies and other cosmological tracers to test theories of gravity beyond general relativity.
We applied these estimators to MWA observations at z = 8.2. The first paper studies the foregrounds and sets preliminary upper limits on the 21-cm bispectrum from reionization. The second refines the analysis by filtering out the smooth component of the data.
Current work
Late-time Universe
Galaxy surveys give us an information-rich, three-dimensional map of the Universe. I build tools that extract as much of that information as possible, including on small, non-linear scales, to understand how the Universe has evolved at late times.
Epoch of Reionization
The 21-cm line of neutral hydrogen lets us study the Epoch of Reionization. We map the sky at radio frequencies with interferometers and extract the information carried by 21-cm fluctuations. I am currently using 21-cm forest data to constrain models of reionization.
The 21-cm window
Neutral hydrogen emits a faint spectral line at a rest frequency of 1420 MHz, a wavelength of 21 cm. As the Universe expands, the line is redshifted, so each frequency we observe corresponds to a different cosmic epoch. Tuning a radio telescope to lower frequencies means looking further back in time.
νobs = 1420 MHz1 + z
For example, a signal emitted at redshift z = 9 reaches us at 142 MHz.
Visual tour
