Research Interests
I have been studying the central regions of galaxies in order to
understanding their evolutionary history. The fist step involves
measuring the mass of any central black hole. Together with the Nuker Team ,
we have found that nearly all galaxies contain a central supermassive
black hole. Furthermore, the mass of the black hole strongly
correlates with various galaxy properties. I have been working on a Black Hole Webpage that
describes the data and results. In order to measure the mass of the
black hole, we use a sophisticated orbit-based model to represent the
galaxy. These models provide one of the most general solutions for how
stars can orbit in a galaxy. This modeling code allows us to not
only measure the central black hole accurately, but also to determine
how the stars orbit throughout the galaxy. Both of these relate to how
the galaxy formed and evolved.
The plot on the right represents the paths of stars as they orbit in a
galaxy that contains a black hole. Each color is a different star.
|
|  |
HETDEX stands for the Hobby-Eberly Telescope Dark Energy Experiment.
Dark energy is one of the most important quantities of our Universe,
but we have essentially no idea as to what it is. Along with Gary Hill
and Phillip MacQueen, we have designed and are building an instrument
capable of accurately measuring potential evolution of dark energy.
The experiment is described
here . We will use the HET to conduct a large redshift survey,
detecting nearly 1 million galaxies over a huge volume. These results
will determine the expansion history of the universe using baryonic
oscillations, thereby determing the evolution of dark energy.
I have been measuring the Fundamental Plane for field ellipticals out
to z around 1.0 using data gathered for the DEEP project, on the Keck
telescope. The data show strong luminosity evolution: around two
magnitudes at z=1.0. This amount of galaxy brighening suggests a
rather recent formation epoch--around z=1.5--in contrast to results
for cluster galaxies, which suggest formation epochs before
z=3.0. These observations will provide a strong constraint on
evolutionary models for galaxy formation.
Dynamics of globular clusters continue to be one of my main areas of
interest. Tad Pryor, Ted Williams, and I have pioneered the use of an
imaging Fabry-Perot (FP) for acquiring large velocity samples in
globular cluster (both Galactic and Magellenic). This work is a
continuation of my thesis work, where I developed the observing
strategies and analysis techniques for using the FP. Current
techniques, such as fibers and multi-slit spectrographs, suffer from
crowding problems in dense regions since light from a single bright
star will dominate the neighboring starlight. The two-dimensional
information of the FP allows one to properly apportion the light. This
technique is the most efficient for obtaining velocity information in
crowded regions, as we have increased the number of stars with
measured velocities in the central regions by factors of 10--100 with
only 3--4 hours of observing time per cluster. I have now begun
collaborations with other astronomers who are interested in using this
technique.
With the velocity measurements from the FP, I have been developing new
techniques which can determine the mass density profile
non-parametrically. The technique takes estimates of the velocity
dispersion and the surface brightness profile and, after inversion
through the Abel integrals, uses the Jeans equation, assuming
isotropy, to provide a non-parametric estimate of the mass density and
mass-to-light (M/L) ratio as a function of radius. Applying this
technique to many clusters, we have found increases in the M/L in the
central and the outer regions. The central increase is explained
through mass segregation and, for the first time, we are able to
directly estimate the heavy remnant population. The increase in the
outer parts can be explained with having a population of low mass
stars at those radii. The advantage of using the non-parametric
techniques is that we are now able to put strong constraints on
models, either N-body or Fokker-Plank, and we can directly estimate
the present-day mass function. The mass functions for all of the
clusters studied show a significant number of objects with a mass
around 0.7~M$_\odot$. If we assume these are primarily white dwarfs,
we can use their numbers to place constraints on the initial mass
function.
From the two-dimensional FP data, I am able to measure a velocity map,
pixel-by-pixel, using the integrated light of the cluster, providing
an accurate measure of any rotation. For most of clusters, there is a
clear indication of rotation in the inner 0.5 parsecs. We also have a
measure of the rotation at 3 parsecs, and the amplitude of the
rotation is generally the same. Although not dynamically significant,
the flat rotation curve has not been seen in standard models and
N-body codes, since solid body rotation is expected. For M15, we have
measured a significant increase in the rotation amplitude at small
radii. This increase is not expected from evolutionary models and may
be the result of a central mass concentration. A $1000\Msun$ black
hole is consistent with both the rotation and velocity dispersion
profiles.
Adaptive optics will quickly become one of the most important astronomical
tools. I have recently been involved in two AO projects: kinematic observations
for core-collapse clusters and M/L gradient in late-type galaxies.
Globular cluster systems around early-type galaxies provide an
excellent opportunity to study the evolutionary and formation history
of the host galaxy. Working with Markus Kissler-Patig we have used the
radial velocities of the clusters around M87 to measure its spin
parameter. The results demonstrate significant rotation in the outer
halo of M87, consistent with simulations that include a merger of two
massive objects; it is difficult to create such a high spin parameter
from accretion by random infall of nearby smaller galaxies.
I have also worked with T.C. Beers in the area of clusters of
galaxies. At the time, this was an area that lacked rigorous
statistical analysis and I worked mainly on developing the statistical
tools to properly analyze the cluster data. We wrote a statistical
package, ROSTAT, which applied a variety of estimators for location,
scale, confidence bands, and goodness-of-fit to a specific
distribution, among many other tests, to a given dataset. This package
was distributed among the astronomical community and has received
considerable use and praise from the users. Some of the subroutines
which are more heavily used are the biweight estimator for location
and scale and the bootstrap technique for determining confidence
bands.
I analyzed velocity measurements from clusters of galaxies to
determine both the significance of cD galaxy velocity offsets and the
existence of bound populations around cD galaxies. Using robust
statistical techniques we showed that a much smaller percentage of cD
galaxies exhibited significant velocity offsets or bound populations
than had previously been reported. This difference was due to the more
robust statistical analysis and we encouraged a better understanding
in the astronomical community of current statistical techniques.
To study positional data from galaxy clusters, I implemented an
adaptive kernel technique (Silverman, B.W. 1986, Density Estimation
for Statistics and Data Analysis, Chapman and Hall, London) for
density estimation, which allowed us to better determine the
significance of substructure. In addition, we used clustering
algorithms, incorporating both velocity and positional data, to
provide an estimate of sub-clumps within a cluster. When applied to
Abell 400, these techniques showed that the cluster is best modeled as
two groups which are presently merging. This result has an important
effect on estimates of the velocity dispersion of the cluster since
the dispersion was significantly higher when using all the velocities
from the whole system as compared to calculating the dispersions
separately for the two groups. The inferred M/L was a factor of four
lower if the subclumps were considered separately. We have looked at
other galaxy clusters and, after applying the adaptive kernel to
determine the membership of particular sub-groups, we have calculated
robust velocity dispersions and locations in order to better
understand the present-day kinematical states of the clusters. We
found that if substructure is present, but ignored, the derived
velocity dispersion may be overestimated by a factor of two.