I'm a cosmologist working at the interface of cosmological theory and data analysis. The main goal of my research is to find new ways to test and constrain fundamental physics with cosmological data. The initial perturbations of the universe are believed to have been created by a quantum process, and the statistical properties of the resulting density field contain information about the ultra high energy physics of the primordial universe.
Recently I have worked on using the cross correlation of the kinetic Sunyaev Zeldovich effect (kSZ) in the CMB and galaxy surveys to reconstruct the large-scale velocity field of the universe. With my collaborators I have proposed an optimal estimator for this reconstruction and shown that it has high signal-to-noise in upcoming experiments. This velocity reconstruction is a new source of information for cosmologists that has already found multiple interesting applications. In particular I have led the development of a method to constrain primordial non-Gaussianity of the local type using the kSZ velocity field and a tracer of galaxies, which will lead to improved constraints with respect to previous methods on the important fnl parameter.
I also investigate how machine learning methods can be used to solve technical problems in cosmological parameter estimation. It appears likely that machine learning will be a key technology to deal with the non-Gaussian, non-linear physics and large computational challenges of future experiments. However, current work on machine learning in cosmology is often basic, for example using convolutional neural networks developed for image analysis as a "black box" to extract parameters from idealized simulations. I believe that machine learning, combined with forward simulations, will come to dominate cosmological data analysis, but in the form of specialized well-understood tools used at different steps of the analysis. In this way machine learning will be combined with classical statistical methods. In my recent publication 1905.05846, I developed a method to perform Wiener filtering of Gaussian fields with an innovative neural network architecture. My neural network, after training, is able to Wiener filter CMB maps a factor of a 1000 times faster than the standard conjugate gradient method, with minimal loss of optimality. Wiener filtering is the computational bottleneck of optimal analyses of near-Gaussian random fields in cosmology, including power spectrum or non-Gaussianity estimation. In addition to its practical value, the paper demonstrates how physical insight can be included in machine learning analysis.
Another area of my research is finding or constraining massive particles during inflation. In principle the density perturbations created by inflation are sensitive to much higher masses than could ever be probed with a terrestrial collider and thus "cosmological collider physics" could give unique insights into ultra high energy particle physics. In many cases such a measurement would require futuristic experimental data and I have worked on forecasting the sensitivity of different experimental setups. For some theoretical models, even current CMB data can be used to obtain interesting and new constraints on massive states. In previous and ongoing work I am searching for such signatures in Planck CMB data. Cosmological collider physics is still a young field and I believe there will be interesting developments in the years and decades to come.
I am member of the Planck experiment, the Simons Observatory, and CHIME and have contributed data analysis and forecasts to these collaborations. I am also contributing to the Modern Inflationary Cosmology collaboration within the Simons Foundation Origins of the Universe Initiative.