I study galaxies, stars, and how stars change galaxies.

I'm a Ph.D. student in the UC Berkeley Department of Astronomy. I use supercomputer simulations and data from big telescopes to better understand how the universe works.

My speciality is galaxy formation, with a focus on low-mass galaxies. I'm also interested in a range of problems in stellar physics and, more generally, in developing new methods to extract information from data. My advisors are Eliot Quataert and Dan Weisz .

I was an undergrad at Yale University, where I worked with Marla Geha. I graduated with a B.S. in astrophysics in 2016.

Research Interests

I am interested in a broad range of topics at the intersection of cosmology, galaxy formation and evolution, and stellar physics. I try to work at the interface between of theory and observations, emphasizing thorough tests of theoretical predictions with observational data.

Brief summaries of some of my projects are available below.

Dynamical effects of stellar feedback on low-mass galaxies

Using galaxy simulations from the Feedback in Realistic Environments (FIRE) project, I have shown that feedback-driven gas outflows can couple energy from stellar feedback to the orbits of stars and dark matter. This causes stars to migrate coherently outward, producing metallicity gradients and apparent outside-in quenching signatures similar to those observed in nearby dwarf galaxies.

Because stellar feedback disrupts star forming gas clouds, the star formation rate of low-mass galaxies varies in concert with the dynamics and spatial distribution of stars and dark matter. I have shown that this leads to observationally testable predicted correlations between galaxies' star formation rates and their sizes, velocity dispersions, and dark matter density profiles.

Feedback-driven outflows are dynamically important when (a) star formation is bursty, (b) the potential is shallow, and (c) the baryon fraction is relatively high. I have shown that these conditions are satisfied not only in dwarf galaxies, but also in the high-redshift progenitors of Milky Way-mass galaxies. As a result, many of the dynamical processes discussed above also apply. In particular, I predict the oldest stars in the Milky Way to have migrated outwards since they formed, so that ancient stars are found primarily in the inner stellar halo at $z=0$.

For details, see the paper Where are the most Ancient Stars in the Milky Way?.

Some popular science coverage related to these papers can be found in this article. A follow-up investigation can be found in this paper.

Stellar feedback as a driver of gas kinematics and morphology

Stars inject energy into their host galaxies' gas through supernovae, radiation pressure, and stellar winds. I have shown that, particularly at low masses, the resulting gas outflows dramatically change galaxies' observable properties, heating their gas and preventing the formation of stable disks. This failure of disk formation turns out to be closely related to the stochastic star formation histories our model predicts for low-mass galaxies.

Spatially resolved gas kinematic maps are expensive and observationally in short supply. Using simulated galaxies, I have shown that one can also extract kinematic information from spatially unresolved observations, which have already been obtained for large, morphologically unbiased samples of galaxies in the local Universe. I have then used mock-observations of unresolved HI to compare simulations and data.

Globular cluster formation

I have developed a semi-analytic model to predict the globular cluster populations of galaxies based on their assembly histories. One of the lessons of this model is that many properties of globular cluster populations are primarily set during hierarchical assembly and are not sensitive to details of the globular cluster formation process.

Stellar multiplicity

On much smaller scales, I have studied how the presence of unrecognized binarity, which affects nearly half of stars targeted by spectroscopic surveys of the Milky Way, can bias stellar parameters and abundances derived from fitting spectra. I have also developed a binary spectral model that corrects these baises.

With a binary spectral model in hand, I analyzed spectra of main-sequence stars from the APOGEE survey in order to find targets whose spectra show evidence of unresolved multiplicity. Our model made it possible to identify 2-3 times more binaries and triples than can be identified by traditional radial velocity methods, which are only sensitive to short-period systems.

More recently, I've studied the occurence rate and separation distribution of very wide binaries, with separations of tens to tens of thousands of AU. Gaia astrometry makes it possible to identify gravitationally bound binaries and eliminate chance alignments with ruthless efficiency, allow us to create a very pure catalog. Using the catalog, we showed that there are fewer-than-expected very wide binaries containing a white dwarf. We proposed that these binaries have been disrupted by velocity kicks during white dwarf formation due to anisotropic mass loss.

Cross-matching the binary catalog with Galactic spectroscopic surveys allows us to measure how the binary fraction varies with metallicity. We find that it becomes strongly anti-correlated with metallicity at separations less than 200 AU, likely corresponding to the separation below which binaries form by disk fragmentation.

The initial mass function of low-mass galaxies

Despite 60+ years of research, it remains unclear whether the stellar initial mass function (IMF) is constant or varies with environment. I have carried out statistical tests to determine what observational data is needed to robustly identify IMF variations and have designed an efficient observing strategy to search for IMF variations in nearby dwarf galaxies with JWST.

The initial-final mass relation

The recent Gaia data release provided accurate parallaxes for a homogenous sample of tens of thousands of field white dwarfs, revealing never-before-seen substructure in the color magnitude diagram (CMD). I developed a model to predict the white dwarf CMD for a given initial-final mass relation (IFMR) and used it to constrain the IFMR for stars with a wide range of initial masses.

For details, see the paper An Empirical Measurement of the Initial-Final Mass Relation with Gaia White Dwarfs. Here's some popular coverage of the paper (in German).

Publications and Curriculum Vitae

My publications: ADS or arXiv.

My CV (current as of June 25, 2019) is available here.

Public Code and Data

Wide binaries with spectroscopic metallicities
The catalog of wide binaries within 200 pc that were observed by a spectroscopic survey and analyzed in the paper The wide binary fraction of solar-type stars: emergence of metallicity dependence at a < 200 AU is available here .

Gaia wide binary catalog
The catalog of high-confidence wide binaries within 200 pc produced in the paper Imprints of white dwarf recoil in the separation distribution of Gaia wide binaries is available here .

Separation distribution inference
The code used to infer the intrinsic separation distribution of wide binaries (correcting for incompleteness at small separations) in the paper Imprints of white dwarf recoil in the separation distribution of Gaia wide binaries is available here .

Initial-final mass relation
A tabulated version of the IFMR derived in the paper An empirical measurement of the initial-final mass relation with Gaia white dwarfs is available here .

Spectral models
A public version of the code BinSpec, which models and fits spectra of single- and multiple-star systems, is available on available GitHub .

Spectroscopic binary catalogs
Several catalogs accompanying the paper Discovery and Characterization of 3000+ Main-Sequence Binaries from APOGEE Spectra are available here .

FIRE galaxy simulation angular momentum
True and mock-observed baryon angular momenta for simulated galaxies from the paper Gas kinematics, morphology and angular momentum in the FIRE simulations are available here .


LAMOST spectra along the main sequence

The animation below shows how the normalized optical spectrum of main-sequence stars changes with mass along the main sequence, assuming LAMOST-like resolution ($R\sim 2000$) and wavelength coverage. The spectral model uses The Payne, developed by Yuan-Sen Ting, to interpolate between synthetic model spectra.

LAMOST binary spectrum

This animation shows how the LAMOST-like spectrum of an unresolved main-sequence binary system with a solar-type primary changes with mass ratio. The bottom-right panel shows that the spectral signature of binarity is maximized at $q\sim 0.8$. For details, see El-Badry et al. 2018.

Contact Me

email: kelbadry{at}berkeley{dot}edu
skype: kareemelbadry1
office: 407 Campbell Hall, Berkeley, CA 94720