David Gerdes

Arthur F. Thurnau Professor of Physics

University of Michigan


Driving DECamMy research addresses basic questions about the large-scale structure and evolution of the universe. How did the universe evolve from its very homogeneous state shortly after the Big Bang to the rich and varied structures we see today? How is this evolution affected by the properties and abundance of the universe's main ingredients: dark energy, dark matter, and baryons? I'm part of several large astronomical surveys that are attempting to answer these questions. The Dark Energy Survey is an ongoing optical imaging survey of 5000 square degrees of the sky using the Blanco 4-meter telescope at CTIO in Chile. Over the course of 525 nights of observing between 2013 and 2018, we'll observe about 300 million galaxies in five different regions of the visible and near-infrared spectrum (the grizY bands), and discover approximately 4000 Type Ia supernovae. With this data set, we'll be able to measure the expansion history of the universe using four different, complementary techniques:

  • Weak gravitational lensing The small distortions of galaxy shapes caused by the bending of light by massive objects (such as galaxy clusters) that the light encounters on its way here can tell us about the distribution of dark matter at various distances and how it has evolved over time.
  • Baryon acoustic oscillations This refers to the tendency of galaxies to "clump" together on a characteristic distance scale. This measurement probes the geometry of the universe on different distance scales.
  • Galaxy clusters The number and size of galaxy clusters as a function of redshift is a sensitive probe of dark energy.
  • Type Ia supernovae This was the technique that was used to discover dark energy in the late 1990s. Type Ia supernovae are "standard candles" that allow us to reconstruct the Hubble diagram of distace vs. redshift over more than half the age of the universe.
Most of these measurements require knowing the redshift of the object being measured. Ideally, this would be measured via spectroscopy, but this is impractical for a sample of 300 million objects. I'm interested in using machine-learning techniques to estimate galaxy redshifts empirically using only the five observed DES magnitudes, and perhaps a few other factors like size, shape, and proximity to other galaxies. As a step in this direction, my graduate student Adam Sypniewski and I have developed the ArborZ photometric redshift algorithm, which uses boosted decision trees and a spectroscopic training sample to provide a full photo-z probability distribution function, P(z). We've shown that P(z) provides a more faithful reconstruction of the true redshift distribution than an estimate based on single best-estimate photo-zs alone. The ArborZ code is freely available and we encourage its use by the broader astronomical community.

Looking to the longer term, I'm part of the DESI project ("Dark Energy Spectroscopic Instrument"), a proposed 5000-fiber R=5000 spectrograph intended for installation on the 4-meter Mayall telescope at Kitt Peak National Observatory. DESI will measure baryon acoustic oscillation features in galaxies and in hydrogen gas over a wide survey area out to redshifts of 3.5. A large spectroscopic survey like this is the logical next step following a wide-field imaging survey like DES.

I'm an "adult-onset astronomer" who came to this field after working for a number of years in experimental high-energy physics. From 1992-2008 I was part of the CDF Experiment at Fermilab, where I contributed to the discovery and study of the top quark. Then the Tevatron was shut down.

My research is supported by a grant from the US Department of Energy, Office of Science.