Research

Our research program focuses on understanding the nature of dark matter and dark energy. We pursue this via new observations of the universe using existing facilities, though development of new facilities and instrumentation, and a laboratory search for dark photon dark matter.   Our group uses optical and near-IR ground-based telescopes (the Keck telescopes, NOAO 4-meter telescopes, and the Subaru telescope) equipped with state-of-the-art instruments along with space-based facilities such as HST (optical), Spitzer (IR), GALEX (UV), and Chandra and XMM (X-ray) to follow up discoveries made in the Deep Lens Survey. Some of these programs are described below. Much of our work is done in collaboration with others in the cosmology group here, as well as the LSST Project and the LSST Science Collaborations.

A key focus of our research activity is optimizing LSST performance.   For this, we use the DLS data (and our DLS research results) to develop new data analysis algorithms required for LSST. An important area of our research is improved photometric redshift calibration. We also have laboratory experiments to measure systematic errors in the LSST CCDs, and develop an understanding of the 3-D electron transport effects, leading to a correction that can be applied in the LSST data software pipeline.

 

The Big Picture: The Power of Surveys

The application of CCDs to astronomy has generated an exponential increase in science capability.   Ultimately, LSST will provide a relational database of tens of trillions of photometric measurements of 30 billion objects.  The plot below charts this amazing history, quantified in terms of the number of galaxies (at a given S/N ratio) surveyed per unit time, at each epoch over the past 30 years.   The trend is driven by microelectronics technology: the size of focal planes in pixels, and the exponential capability of our computers to process and analyze the exponential growth in data.

CCD Data Trends2.jpg

 

LSST CCDs put to the Test

Experiments usually are limited by systematic errors, sometimes due to sample selection bias, and sometimes due to systematics in the detector.  Our LSST CCDs are novel segmented thick 3-D devices, and the charge transport in these thick silicon CCDs is complex.  In order to map these effects, we have built a LSST optical beam simulator, which allows us to operate the CCDs in a realistic f/1.2 beam illumination.

UCDavis_CCD_Lab.jpg

The system shown above is used to fully characterize the various systematic effects in our CCDs, from astrometric systematics, to the related PSF systematics, to the related photometric systematics -- all in the context of the resulting residual weak lens shear systematic error.  A link to a recent paper is arXiv:1703.05823. This R&D is key to the success of LSST as a dark energy physics mission.  Its operations are a collaboration between our group, the LSST Project, and the Dark Energy Science Collaboration. Initially funded through an NSF AST ATI grant, the facility now is supported by private philanthropy and the DOE Office of High Energy Physics grant DE-SC0009999.

 

Dark Matter Fluctuations

The amplitude of dark matter clustering is a key cosmological parameter. Recently Jee et al. found a normalization for the dark matter power spectrum which is slightly higher than previous estimates.  One week later, the Plank team published a similar estimate.  Our analysis used a new shear estimation algorithm applied to a million galaxies in the DLS in a 2-D analysis of shear-shear correlation vs angle.  

DLS-sigma8.png

We have repeated this 2-D analysis with full cosmic shear tomography, making use of improved photometric redshifts and the new sFIT shear estimator. This yielded higher precision cosmology measurements: Jee et al. 2016 ApJ 824, 77.

 

Growth of Large Scale Dark Matter Structure

Galaxies are embedded in dark matter over-densities, which should grow with cosmic time. Using weak gravitational lensing of a million galaxies of known redshifts in the Deep Lens Survey, Choi et al. were able to detect the growth of large scale mass directly for the first time. In the plot below, mass over-density associated with lens galaxies in three lens redshift bins (0.3, 0.5, 0.8) is shown as a function of projected radius from the galaxy.  The observed growth in galaxy-mass correlation is consistent with LCDM cosmology, including galaxy bias.

g-m.png

The next step is to fit a halo model to the data, informed by the galaxy clustering observed in each redshift bin. This will enable separation of bias from growth, and a more direct test of cosmology.  Michael Schneider et al. are currently working on this, informed by covariance analysis of n-body simulations.  With the LSST data, we expect 50x smaller errors!   LSST will also probe the physics of dark matter in multiple ways.

 

Towards LSST Precision Photometric Redshifts

Sam Schmidt has been working on techniques to improve the calibration of photometric redshifts for LSST.  As training data will likely be incomplete, particularly for the faintest galaxies, methods that can calibrate the redshift distribution with non uniform training data are of particular importance.  The most promising technique takes advantage of the fact that galaxies are embedded in a cosmic web of large scale structure.  In essence, if we have a set of objects for which we know the redshifts, even if such a sample is bright and not representative of the entire population, we can measure the cross-correlation signal between these known objects and an unknown sample to determine the redshift distribution of the unknown sample.  In recent work we implemented such a technique, and showed that it can be extended into the quasi-linear regime of clustering to increase signal-to-noise, while still maintaining an accurate recovery of the redshift distribution.  The figure shows a Gaussian redshift distribution drawn from mock galaxy light cones (blue) in only 300 square degrees in both a constant bias scenario (left), and galaxy bias evolving with redshift (right).  

pz.jpg
Shown: the raw estimated redshift distribution from our technique in red, as well as an additional correction in blue-grey, for 3 different clustering scales.  The blue line is the true underlying redshift distribution. In all cases, we can accurately reconstruct the underlying distribution.  Small scale cross-correlation measurements can also be used for sensitive tests for catastrophic outlier detection, which will be very important for LSST.

 

Laboratory search for dark photon dark matter

Many types of astronomical observations show decisively that most of the mass in the Universe is of an unknown form, unlike ordinary matter. This "dark matter" fills the universe and clumps over cosmic time under its own gravitational self attraction. Our current understanding of physics cannot explain dark matter; its existence is evidence for new physics! Its physical nature is a central unanswered question in science. Sensitive searches for weakly interacting massive particles in the GeV range have found nothing. Other possibilities for dark matter, such as the ultra-low mass nano eV to milli eV regime remain unexplored. A natural candidate for vector dark matter is the hidden photon, which can couple to electromagnetism. We are working on an extremely sensitive laboratory experiment: "Dark E-field Radio"  which leverages cryogenic microwave detectors and FPGA technology in a GHz wide real-time spectral analysis. The result will be a 10,000-fold improvement over current astrophysical limits in dark matter detection searches in this vast unexplored ultra-low mass regime.

 10_31_19_LimitPlot.png