Reconstruction on cosmological and astronomical data in a model independent manner using machine learning techniques such as Gaussian Process Regression (GPR) and Principal Component Analysis (PCA). A recent work looked at the effect of choice of kernel function in the reconstruction of H(z) observations using GPR. Using the modifeid version of GaPP code, we showed that polynomial kernels are a better choice compared to the widely used RBF and Matern kernels.
Planning to extend such analyses to a wide variety of observational datasets.
Testing GR by studying the observational consequences of potential modifications. Applications in cosmology and observations related to compact objects.
We have studied the background and perturbed cosmological evolution of the late Universe in non-standard cosmological models such as f(R) theories, interacting dark sector models and more general Horndeski theories and beyond. We studied the effect of these modifications compared to the standard cosmological model and showed that studying the perturbed evolution is key to detect the deviations from GR using cosmological observations.
We also explored the recently suggested Ricci-inverse gravity and its potential as a viable candidate as late-Universe cosmological model.
We have also studied the early Universe evolution and inflationary dynamics in the modified gravity theories (f(R,phi)).
I have also co-authored a review article containing these studies and more on modified theories gravity titled Modified theories of Gravity: Why, How and What?
In this study, we devised a test of Einstein's Equivalence Principle (EEP) by studying the geodesic of the photon around a blackhole where electromagnetic field is non-minimally coupled (NMC) to gravity. We explicitly showed that the NMC of the electromagnetic field introduces observable modifications to the black hole image, and these modifications are potentially detectable by future observations.