Robert Gibson, Research Associate
Contact Information
Email: rgibson@astroOffice : C304
Cell : 206.794.8197
Fax : 206.685.0403
Dept. of Astronomy
University of Washington
Box 351580
Seattle, WA 98195-1580
Shipping Address:
Physics/Astronomy Bldg. C319
3910 15th Ave NE
Seattle, WA 98195-1580
About Me
After undergraduate study in math and physics at the University of Michigan, I worked for several years as a software engineer at QUALCOMM, Inc., then returned to study astrophysics in graduate school at MIT. There, I studied high-resolution X-ray spectroscopy of AGN with Claude Canizares and the Chandra HETGS group. After grad school, I worked with Niel Brandt at Penn State. I now work at the University of Washington with Andy Connolly, Zeljko Ivezic, and the whole gang. When I'm not working, I'm enjoying life with my wife, Ann, and young children, Ross and Eleanor.
Here are some of the things I'm interested in and have worked on recently.
Multiwavelength Surveys
AGN Absorber Structure and Ionization
Large-Scale Survey Science and Methods
Preparing the Future...
Surveys &mdash with a variety of sky coverage, depth, and wavebands &mdash are continually adding new value to archived data from previous observing missions. The LSST will be particularly valuable for the study of large, multiwavelength data sets. While the cost of obtaining simultaneous multiwaveband observations is usually prohibitive, the LSST cadence will naturally produce near-simultaneous observations with any other contemporaneous mission.
As one example of multiwavelength survey science, I and collaborators analyzed hundreds of SDSS quasars serendipitously observed with Chandra and XMM-Newton and found that, while quasar emission may vary strongly, very few optically-selected quasars have weak or disrupted X-ray emission. Apparently, almost every quasar that has the ability to produce strong optical/UV emission also produces strong X-ray emission.
About 15% of quasars show very broad absorption lines in their UV spectra. These "BAL" absorption features are a powerful diagnostic to directly measure winds (apparently) accelerated from the accretion disk. These winds can reach very high velocities (30,000 km/s or more) and may carry enough energy to significantly impact their host galaxies and surroundings. It's possible that these broad absorption lines are related to narrower absorption features such as mini-BALs; perhaps the narrower features broaden when accelerated by radiation pressure.
As you might expect from the dynamic, small-scale environment in the heart of an AGN, UV-absorbing outflows are variable over observable time scales. We also have indications from high-resolution X-ray spectroscopy indicating that we may be able to observe structural changes in X-ray absorbers on time scales as small as a few hours. As we observe more sources with increasingly sensitive spectroscopy in a variety of wavebands, we expect to learn about the structure and evolution of the material in the heart of AGN, as well as the physics that launches and accelerates AGN outflows.
Large-scale, multiwavelength science requires a lot of planning and effort. In my past and ongoing research, I have developed methods to efficiently fit 80,000 SDSS spectra, reduce and analyze thousands of X-ray observations, and examine variations in spectra taken at different times with (possibly) different observatories. For the SDSS DR5 BAL Catalog, I wrote code to fit quasar spectra automatically and then provide a simple interface to quickly inspect and adjust fits for these 80,000 spectra. The model fits were then distributed to collaborators and re-adjusted based on their comments. The final catalog was then "compiled" from the final set of fits. Even with this high degree of automation and delegation, this approach would not be feasible for a data set that was more than about 100,000 spectra.
Experience pointedly demonstrates that current surveys are already taxing traditional methods of both analysis and also interpretation. For example, visual inspection by experts will not work for samples even double current sizes. Projects must be carefully planned and delegated before analysis begins, as unexpected limitations or errors found part-way through can be disastrous for large-scale projects.
One way to prepare for the future is to simulate it. LSST image simulations help scientists, educators, and developers practice working with the flood of data that will be generated in the LSST survey. The way we work with astrophysical data will soon need to evolve. Network bandwidth, storage space, and processing power are annoying, but easily solvable, details. More pressing is the need to develop new methods of data analysis (especially error analysis and quality assurance!) for huge data sets where traditional methods like visual inspection will not be possible.
