Research interests

My research is in cosmology, both theoretical and observational. My work spans analytical and numerical modeling, confrontation between theoretical predictions and observations, survey planning and data reduction and analysis. As well as my work on DESI that I've started when I joined LBNL, I have worked and am working on the following projects or topics.

HETDEX

The Hobby-Eberly Telescope Dark Energy Experiment is a blind spectroscopic survey that aims to measure the matter power spectrum P(k) and the expansion rate of the Universe H(z) at redshifts 1.7<z<3.5 to 2% and 1% accuracy respectively using Lyman-alpha emitting galaxies as tracers. The survey instrument, VIRUS, consists of 78 integral field unit fed double spectrographs. First light is expected to be in the second half of 2016.

My involvement in this project has been as part of the software team lead by Niv Drory. I have been responsible for developing the scheduling software for the survey. This has involved simulating weather patterns typical at the McDonald Observatory (where the HET is located), taking into account the particularities of tracking targets with the HET (which has a fixed elevation) to simulate the execution of the survey to minimise the duration of the survey while maximising the throughput for observations and avoiding undesirable patterns in the survey window function. We have used my code to find the optimal survey footprint and target tiling and are using it to find the best observation strategy to ensure the survey achieves its science goals. This software is now being integrated into the HET queuing system to instruct the telescope operators which field to observe.

I am also responsible for developing an alternative spectral extraction code within Cure, the HETDEX data reduction pipeline. Extraction techniques currently being used in optical (and near-UV) fiber spectroscopy assume that the fiber point spread function is separable along axes parallel to the rows and columns of CCD pixels, an assumption that very rarely holds in reality. An optimal spectral extraction algorithm, a.k.a. "spectro-perfectionism" (Bolton & Schlegel 2010) does not rely on this assumption and was shown to work well for the case where spectra are uncorrelated (on the sky), at the price of needing very large computing resources. My work on this part of the project has two aspects. Firstly, adapt the algorithm to integral field fiber spectroscopy, where adjacent fibers on the detector are also adjacent on the sky, which introduces extra correlations. Secondly, improve the algorithm to make it computationally tractable, and hence useful in the case of an actual survey like HETDEX.

To make HETDEX a possibility, the HET’s field of view had to be increased, requiring a new corrector and a new tracker. Over the past two and a half years, I have been involved in the development of the new telescope control system (TCS) for the upgraded HET. Specifically, I am responsible for the following aspects: adapting and integrating the object trajectory generation library into the new TCS; developing and implementing the mount model for various components of the telescope such as the tracker’s flexure and positioning of guide cameras; developing the coordinate transform library to go back and forth between the physical coordinates and orientation of the telescope tracker, instruments and probes and where they are pointing on sky; I am also participating in the deployment of this software; parts of this process are already in an advanced stage, in particular those relating to the tracker; what remains will be deployed after the hardware upgrade is completed.

Cosmic strings

The most important part of my theoretical work concerns the cosmological and astrophysical consequences of the formation of cosmic strings in the early Universe. There has been a renewal of interest in cosmic strings in the past decade, due to developments in theoretical physics. Indeed, it has been shown that all cosmologically viable supersymmetric grand unified theories lead to the formation of topologically stable cosmic strings at the end of the inflation. Also, braneworld cosmologies, in which our 3+1 dimension Universe is a brane in a higher dimensional bulk, generically predict that "cosmic superstrings" will be copiously produced at the end of inflation.

My major contribution in this field has been the development of an efficient method to solve the Einstein-Boltzmann equations on a 3D grid with the direct input of high resolution simulations of cosmic string networks. This code has been massively parallelised using OpenMP directives and has achieved very high scalability. Our first results were all-sky maps at a resolution sufficient to study large angular scales. The power spectra we extracted from these maps enabled us to infer the variation of the string linear energy density as a function of the cosmological constant. However, since observations have excluded that strings are the primary source of cosmological perturbations, the main interest lies in the study of small angular scales, where a subdominant string contribution could still be detected due to their non-Gaussian signature. The first maps that we have computed with an angular resolution varying from 0.05 to 0.2 degrees have enabled us to verify this hypothesis which has been confirmed in our more recent work which features maps of higher resolution (0.7' to 4.2') and simulations that spanned a much longer dynamic range (a total of 180 in conformal time).

This work has recently been extended on several fronts. We have published work using networks’ unequal time correlators to directly compute the CMB angular power spectrum. I also expect to submit a paper on the polarisation of the CMB seeded by networks of cosmic strings in the near future. More recently, I have reworked and integrated the 3D grid method and the ray-tracing codes outlined above to take advantage of the significantly increased shared memory available for our simulations and thus eliminate the need for large intermediary output files whose size had become the limiting factor of achievable resolution. I plan to use this improved pipeline to compute high resolution all-sky CMB maps, thus updating our previous results, but also to compute maps of other observables, such as weak lensing maps and 21cm intensity maps. My collaborators and I have also used the pipeline I developed to compute maps of CMB fluctuations seeded by networks of domain walls.