X. Fan

The University of Arizona, Tucson, Arizona, United States

Are you X. Fan?

Claim your profile

Publications (147)270.73 Total impact

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: High-redshift quasars are currently the only probes of the growth of supermassive black holes and potential tracers of structure evolution at early cosmic time. Here we present our candidate selection criteria from the Panoramic Survey Telescope & Rapid Response System 1 and follow-up strategy to discover quasars in the redshift range 5.7<z<6.2. With this strategy we discovered eight new 5.7<z<6.0 quasars, increasing the number of known quasars at z>5.7 by more than 10%. We additionally recovered 18 previously known quasars. The eight quasars presented here span a large range of luminosities (-27.3 < M_{1450} < -25.4; 19.6 < z_ps1 < 21.2) and are remarkably heterogeneous in their spectral features: half of them show bright emission lines whereas the other half show a weak or no Ly$\alpha$ emission line (25% with rest-frame equivalent width of the Ly$\alpha$ + Nv line lower than 15{\AA}). We find a larger fraction of weak-line emission quasars than in lower redshift studies. This may imply that the weak-line quasar population at the highest redshifts could be more abundant than previously thought. However, larger samples of quasars are needed to increase the statistical significance of this finding.
    The Astronomical Journal 05/2014; 148(1). · 4.97 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present spectral energy distributions (SEDs) of 69 QSOs at z>5, covering a rest frame wavelength range of 0.1mu to ~80mu, and centered on new Spitzer and Herschel observations. The detection rate of the QSOs with Spitzer is very high (97% at lambda_rest ~< 4mu), but drops towards the Herschel bands with 30% detected in PACS (rest frame mid-infrared) and 15% additionally in the SPIRE (rest frame far-infrared; FIR). We perform multi-component SED fits for Herschel-detected objects and confirm that to match the observed SEDs, a clumpy torus model needs to be complemented by a hot (~1300K) component and, in cases with prominent FIR emission, also by a cold (~50K) component. In the FIR detected cases the luminosity of the cold component is on the order of 10^13 L_sun which is likely heated by star formation. From the SED fits we also determine that the AGN dust-to-accretion disk luminosity ratio declines with UV/optical luminosity. Emission from hot (~1300K) dust is common in our sample, showing that nuclear dust is ubiquitous in luminous QSOs out to redshift 6. However, about 15% of the objects appear under-luminous in the near infrared compared to their optical emission and seem to be deficient in (but not devoid of) hot dust. Within our full sample, the QSOs detected with Herschel are found at the high luminosity end in L_UV/opt and L_NIR and show low equivalent widths (EWs) in H_alpha and in Ly_alpha. In the distribution of H_alpha EWs, as determined from the Spitzer photometry, the high-redshift QSOs show little difference to low redshift AGN.
    02/2014; 785(2).
  • [Show abstract] [Hide abstract]
    ABSTRACT: The Sloan Digital Sky Survey discovered a remarkable set of quasars at z = 2.2 ‑ 5.9 having extremely weak rest-frame UV emission lines in their spectra (hereafter, WLQs). One hypothesis explaining the extreme weakness of their emission lines is that WLQs are high accretion rate sources, and one manifestation of a high accretion rate is a steep hard-X-ray spectral slope. We therefore performed spectral analysis of all the high-quality X-ray observations of WLQs available in the archive. This analysis includes new XMM-Newton observations of SDSS J0928+1848 and SDSS J1231+0138, at z = 3.8 and z = 3.2 respectively, alongside the spectral analysis of two archival XMM-Newton sources, SDSS J1141+0219 and SDSS J1012+5313, at z = 3.6 and z = 3.0 respectively. With a total exposure time of ˜78 ks and an average of ˜1200 counts per source, we derived the hard-X-ray photon index for each object, as well as the mean photon index derived from joint-fitting these spectra, while accounting for the possibility of intrinsic absorption for which we found no evidence. Our results show that the mean photon index of the four WLQs is below the average observed in luminous radio-quiet quasars, consistent with the fact that all of the WLQs in the XMM-Newton archive are radio intermediate. This is mainly a consequence of selecting the X-ray brightest WLQs for XMM-Newton observations. In order to be able to test the hypothesis that WLQs have high accretion rates, future XMM-Newton observations should target radio-quiet WLQs, and suitable candidates likely lie at lower redshifts, i.e. at z < 2.2.
    01/2014;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We have undertaken a survey of the most massive and overdense galaxy structures in the early universe at z=2-4, traced by regions of the highest optical depth from IGM Lyα absorption over large scale of 10-30 Mpc. These structures represent the most biased regions in the universe at the peak of galaxy formation epoch, bridging the most luminous quasars observed at z∼ 6-7 and the most massive clusters in the local universe. Cosmological simulation and early observations show a strong correlation between the optical depth of 1-D intergalactic Lyα absorption and the 3-D mass overdensities. Based on the simulation and MMT high S/N observations, we have developed a novel technique to identify high-redshift overdense region by searching for high optical depth regions in quasar spectra and requiring that they are not associated with discrete high column density absorbers such as sub-damped Lyα (sub-DLA) or DLA systems. By examining absorption spectra of ∼ 140,000 sight lines provided by SDSS-III quasar survey, we have searched the IGM overdensities over a volume of 10 Gpc^3. This survey volume is more than two orders of magnitude larger than any existing Lyman Break Galaxy survey, enabling the discovery of the rarest and most massive systems. We have selected one unique field with three large Lyα absorption and seven quasars within 20 h^-1 Mpc. Our preliminary and shallow KPNO-4m narrowband + broadband imaging have revealed a significant overdensity of Lyα emitters at bright-end associating with this system. In future, we propose to carry out the LBT/LBC multi-color imaging to probe this field in much deeper using BX galaxies down to L^*, and further take the LBT/MODS observations to spectroscopically follow-up the ∼100 BX galaxies in this most overdense field and map the 3-D structure associated with the largest intergalactic HI overdensity.
    01/2014;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present Herschel far-infrared (FIR) photometry of eleven quasars at redshift z>5 that have previously been detected at 1.2mm. We perform full spectral energy distribution (SED) fits over the wavelength range lambda_rest ~0.1-400mu for those objects with good Herschel detections. These fits reveal the need for an additional far-infrared (FIR) component besides the emission from a dusty AGN-powered torus. This additional FIR component has temperatures of T_FIR ~ 40-60K with luminosities of L_(8-1000mu) ~ 10^13 L_sun (accounting for 25-60% of the bolometric FIR luminosity). If the FIR dust emission is due to star formation it would suggest star formation rates in excess of 1000 solar masses per year. We show that at long wavelengths (lambda_rest > 50mu) the contribution of the AGN-powered torus emission is negligible. This explains how previous FIR studies of high-redshift quasars that relied on single component fits to (ground-based) observations at lambda_obs > 350mu reached T_FIR and L_FIR values similar to our complete SED fits. Stacking the Herschel data of four individually undetected sources reveals a significant average signal in the PACS bands but not in SPIRE. The average SED of sources with individual Herschel detections shows a striking surplus in near- and mid-infrared emission when compared to common AGN templates. The comparison between two average SEDs (sources with and without individual Herschel detections) matched in the UV/optical indicates that for these objects the strength of the MIR emission may correlate with the strength of the FIR emission.
    The Astrophysical Journal 05/2013; 772(2). · 6.73 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Gravitational lensing magnification bias, boosted by the observed steep luminosity function of high redshift quasars, strongly suggests that lenses should be common amongst the highest redshift quasars. Currently over 60 quasars at z>5.7 have been discovered in wide-area ground-based imaging surveys; however, only a handful have been imaged with sufficient resolution to identify subarcsecond scale lenses. I will present results from an ongoing HST SNAP survey of all known 6 quasars, including some promising candidates for gravitational lenses. The observed sample is large enough to place strong constraints on the quasar luminosity function at 6, particularly on the contribution of faint quasars to the reionizing photon budget. Constraining the lensing rate among this unique sample is also needed to correctly derive physical parameters related to black hole growth from the observational data and to interpret quasar / host galaxy correlations at high redshift.
    01/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We present a survey of bright 3 LBGs in the the NOAO Bootes fields and SDSS deep stripe field. Our new LBT U-band survey, combined with the existing NOAO Bootes field survey, allows us to build up a sample of 15,000 photometrically-selected LBGs at 3 in the 9 square degree survey area. With the large survey area, we study UV luminosity function, stellar mass function and clustering in bright 3 LBGs. We compare our results with the results measured in other high-z LBG samples to test the cold flow accretion model and put strong constraint on the cosmic star formation efficiency and star formation history in these galaxies. Furthermore, the new Bootes wide field survey and our newly coadded imaging data in SDSS deep stripe field provide a survey area two orders of magnitude larger than any of previous deep field survey, which allow us to reveal a population of the most luminous LBGs at 3. With L> 7L* and star formation rate (SFR) ~500 M/yr, these spectroscopically-confirmed LBGs are some of the rarest and most intensive star forming systems in the early Universe. The fellow-up deep spectroscopic observations and approved Spitzer and HST observations will reveal the physical properties, place these newly discovered galaxies in the context of galaxy growth through merger and cold flow accretion at the peak era of cosmic star formation,and provide a unique laboratory for galaxy formation theory.
    01/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Huge amounts of far-infrared (FIR) emitting warm dust and molecular gas have been discovered in the host galaxies of ~30% of the quasars at 6, indicating the presence of massive star formation coeval with rapid supermassive black hole (SMBH) accretion in these earliest quasar host galaxies close to end of cosmic reionization. We present our recent ALMA observations of the [C II] 158 micron fine structure line emission from the host galaxies of four millimeter bright quasars at 6. The detections of [C II] line emission provide strong evidence of active star formation and SMBH-galaxy coevolution in the most distant universe. The line velocity maps of three of them at ~0.6" spatial resolution show indications of velocity gradient along the major axis direction, which suggests the presence of a nuclear star-forming disk over the central a few kpc region of the quasar hosts. These results reveal important constraints on the distributions of nuclear star formation and dynamical properties of the cool atomic gas in the young quasar host galaxies. They are the keys to understand the growth of the first SMBHs and their host galaxies close to the cosmic reionization epoch.
    01/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: The Large Synoptic Survey Telescope (LSST; http://lsst.org) will revolutionize our understanding of active galactic nuclei (AGN) and their environments. The decade-long survey will discover at least 10 million AGN across 18,000 square degrees on the sky, with between about 50 to 200 visits per source for each of the ugrizy filters. A combination of the LSST sub-arcsecond astrometry, six-band photometry, and unprecedented cadence will enable the most efficient AGN selection, with additional characterization through the use of sophisticated star-galaxy separation techniques. The time-domain nature of the survey will provide invaluable information on the physics of the AGN central engine, as well as on transient fueling events, and will allow real-time alerts that will trigger follow-up observations. Several LSST "deep drilling" fields will help discover the faintest AGN at high redshift, enhancing the value of current and planned multiwavelength pencil-beam surveys while providing hours-to-years temporal information on thousands of AGN. The wide ranges of both luminosity and redshift spanned by LSST, including the discovery of over 1000 quasars at z>6.5, will dramatically improve the quantification of the optical AGN luminosity function. Measurements of AGN clustering at high redshift will be used to determine the relationship between AGN and dark matter. The discovery of about 8000 gravitationally lensed quasars, including 1000 systems with measurable time delays, will place significantly tighter constraints on key cosmological parameters.
    01/2013;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We present preliminary results of quasar point source subtractions at z=2 and z=6 with WFC3 IR data. We have developed a new Python-based Markov Chain Monte Carlo wrapper for the TinyTim PSF modeling software to allow Bayesian estimation of PSF aberration parameters. We apply this to 18 quasars at z=2 and 2 at z=6, including an improved fit of SDSS J1148+5251. We also present fits of several stars to test the reliability of the algorithm.
    01/2013;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We present a measurement of the quasar luminosity function in the range 0.68<z<4 down to extinction corrected magnitude g_dered=22.5, using a simple and well understood target selection technique based on the time-variability of quasars. The completeness of our sample was derived directly from a control sample of quasars, without requiring complex simulations of quasar light-curves or colors. A total of 1877 quasar spectra were obtained from dedicated programs on the Sloan telescope (as part of the SDSS-III/BOSS survey) and on the Multiple Mirror Telescope. They allowed us to derive the quasar luminosity function. It agrees well with previously published results in the redshift range 0.68<z<2.6. Our deeper data allow us to extend the measurement to z=4. We measured quasar densities to g_dered<22.5, obtaining 30 QSO per deg^2 at z<1, 99 QSO per deg^2 for 1<z<2.15, and 47 QSO per deg^2 at z>2.15. Using pure luminosity evolution models, we fitted our LF measurements, and predicted quasar number counts as a function of redshift and observed magnitude. These predictions are useful inputs for future cosmology surveys such as those relying on the observation of quasars to measure baryon acoustic oscillations.
    Astronomy and Astrophysics 09/2012; · 5.08 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In December 2010, NASA created a Science Definition Team (SDT) for WFIRST, the Wide Field Infra-Red Survey Telescope, recommended by the Astro 2010 Decadal Survey as the highest priority for a large space mission. The SDT was chartered to work with the WFIRST Project Office at GSFC and the Program Office at JPL to produce a Design Reference Mission (DRM) for WFIRST. Part of the original charge was to produce an interim design reference mission by mid-2011. That document was delivered to NASA and widely circulated within the astronomical community. In late 2011 the Astrophysics Division augmented its original charge, asking for two design reference missions. The first of these, DRM1, was to be a finalized version of the interim DRM, reducing overall mission costs where possible. The second of these, DRM2, was to identify and eliminate capabilities that overlapped with those of NASA's James Webb Space Telescope (henceforth JWST), ESA's Euclid mission, and the NSF's ground-based Large Synoptic Survey Telescope (henceforth LSST), and again to reduce overall mission cost, while staying faithful to NWNH. This report presents both DRM1 and DRM2.
    08/2012;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We measure the quasar two-point correlation function over the redshift range 2.2<z<2.8 using data from the Baryon Oscillation Spectroscopic Survey. We use a homogeneous subset of the data consisting of 27,129 quasars with spectroscopic redshifts---by far the largest such sample used for clustering measurements at these redshifts to date. The sample covers 3,600 square degrees, corresponding to a comoving volume of 9.7(Gpc/h)^3 assuming a fiducial LambdaCDM cosmology, and it has a median absolute i-band magnitude of -26, k-corrected to z=2. After accounting for redshift errors we find that the redshift space correlation function is fit well by a power-law of slope -2 and amplitude s_0=(9.7\pm 0.5)Mpc/h over the range 3<s<25Mpc/h. The projected correlation function, which integrates out the effects of peculiar velocities and redshift errors, is fit well by a power-law of slope -1 and r_0=(8.4\pm 0.6)Mpc/h over the range 4<R<16Mpc/h. There is no evidence for strong luminosity or redshift dependence to the clustering amplitude, in part because of the limited dynamic range in our sample. Our results are consistent with, but more precise than, previous measurements at similar redshifts. Our measurement of the quasar clustering amplitude implies a bias factor of b~3.5 for our quasar sample. We compare the data to models to constrain the manner in which quasars occupy dark matter halos at z~2.4 and infer that such quasars inhabit halos with a characteristic mass of ~10^{12}Msun/h with a duty cycle for the quasar activity of 1 per cent.
    03/2012;
  • [Show abstract] [Hide abstract]
    ABSTRACT: High redshift quasars mark the growth of supermassive black holes in the early universe, and determining the evolution of the quasar population at high redshift provides strong constraints on early structure formation and the buildup of massive galaxies. The SDSS-III BOSS is now providing the best measurements of the quasar luminosity function (QLF) at 2.2 < z < 3.5, but is not designed to probe higher redshifts. Using the many epochs of SDSS imaging available in the Stripe 82 region we have identified a large number of z∼5 quasar candidates to a depth of i(AB)=22, and are confirming these objects with BOSS spectra obtained through ancillary science programs, as well as observations at the MMT and Magellan telescopes. These observations probe much deeper into the QLF than previous surveys at these redshifts and provide a key connection between lower redshift results and the highest redshift quasars at z=6-7.
    01/2012;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We present deep U band imaging data covering an area of 9 Square deg in the NOAO Deep Wide-Field Survey (NDWFS) Bootes fields. The data were takenby the Large Binocular Camera (LBC) on the 2x8.4-m Large Binocular Telescope (LBT). Combined with deep Bw and R-band images from NWDFS, we selected 15,000 Lyman Break Galaxy (LBG) candidates at z down to R=25.5 in the entire Bootes Field, the largest z 3 LBG sample in existence. Using this sample, the rest-frame UV and near-IR luminosity functions (LFs) of the LBGs is measured. The large survey area also helps us to eliminate the cosmic variance and put strong constraint on the bright end of the LFs. The spatial distributions of two bright LBG samples (23.5<R<24.0 and 24.0<R<24.5) are studied. The comoving correlation lengths increase with the R band brightness, implying that the clustering strength increases with rest-frame UV luminosities.
    01/2012;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We calculate and present the quasar luminosity function (QLF) using data from 30,000 new, z>2 quasars from the SDSS-III: Baryon Oscillation Spectroscopic Survey (BOSS). The BOSS data, along with the original SDSS-I/II, and the 2SLAQ QSO survey, will help cover luminosity-redshift (L-z) plane, such that there is a 4 magnitude dynamic range from redshifts z=0-4. In particular, the g=22.0 BOSS quasars fill in the z=2.2-3.5 redshift range, key for investigations for the faint-end of the QLF at the height of the ``quasar epoch''. We use our new measurements to map the black hole accretion history of the Universe and constrain how quasars contribute to the ``AGN feedback'' cycle in galaxy formation.
    01/2012;
  • [Show abstract] [Hide abstract]
    ABSTRACT: The Sloan Digital Sky Survey has revolutionized our understanding of quasar clustering at high redshift. But, above redshift 2.5, SDSS-I and II mainly took spectra of quasars at the bright end of the luminosity function. The clustering of fainter quasars above redshift 2.5 can break degeneracies between quasar fueling and host galaxy mass, newly constraining the length of the quasar phase near the peak of quasar activity. The SDSS-III/BOSS survey is compiling a sample of around 150,000 quasars above redshift 2.2, around half of which are above redshift 2.5. By design, a core subsample of BOSS quasars have a well-controlled angular selection function that facilitates clustering measurements. I will present measurements of quasar clustering from the first 2 years of BOSS, focusing on redshifts around 2.2 < z < 2.7. As statistics develop over the full 5-year survey, BOSS will measure the clustering of low-luminosity quasars at 2.5 < z < 3.5 with unprecedented precision.
    01/2012;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Future cosmological spectroscopic surveys of quasars will demand quasi-exhaustive target selection in both the "low-z” (redshifts of about 1.0 to 2.5) and the "mid-z” (redshifts of about 2.5 to 4.0) regimes. In combination, low-z and mid-z quasars can constrain the geometry of the Universe from the Baryon Acoustic Oscillation (BAO) feature in both quasar clustering and in the Lyman-alpha forest. But, target selection of quasars to the depths afforded by next-generation imaging surveys remains difficult, in particular because color-based selection of mid-z quasars_for which the optical colors of stars and quasars are similar_is highly inefficient. To help refine the selection of faint quasars for future surveys like BigBOSS, we have developed a method based on optical variability to target quasars up to g 23 and z 4, with a completeness better than 85%. Our method overcomes the drawbacks of color selection, which systematically misses quasars near 2.5 < z < 3.5. We have conducted a pilot survey within SDSS III/BOSS covering 16 square degrees, complemented by a 4 square degree survey on the MMT to study the fainter targets. We aim to determine the number density of quasars that BigBOSS will be able to use to constrain dark energy. ADM acknowledges support from the Alexander von Humboldt Foundation.
    01/2012;
  • [Show abstract] [Hide abstract]
    ABSTRACT: We present constraints on the properties of QSO host galaxies at z=6, using rest-frame ultraviolet imaging of the QSO J1148+5251. The data were collected by the Hubble Space Telescope Wide Field Camera 3 infrared channel in the F125W and F160W filters, and include contemporaneous observations of nearby stars to measure time-dependent components of the instrument point spread function (PSF). Using this observed PSF as a template, we construct model PSFs which are used to subtract the QSO point source from the images. We obtain a clean subtraction, but no detection of the host galaxy, to a 2σ limiting surface brightness of J > 24.45 AB mag arcsec-2 (H > 24.90 AB mag arcsec-2). Assuming a Sérsic profile with re = 0.4" and n = 4, this corresponds to a host galaxy with an integrated total magnitude of J > 23.05 AB mag (H > 23.50 AB mag). Further, we test the reliability of this method to recover the host galaxy flux by simulating host galaxies with varying parameters and conducting the same analysis. We discuss cosmological implications of the associated luminosity, size, and mass limits. Support for HST GO program #12332 is provided by NASA through a grant from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555.
    01/2012;
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Quasars (QSOs) at the highest known redshift (z {approx} 6) are unique probes of the early growth of supermassive black holes (BHs). Until now, only the most luminous QSOs have been studied, often one object at a time. Here we present the most extensive consistent analysis to date of 4 < z < 6.5 QSOs with observed near-infrared spectra, combining three new z {approx} 6 objects from our ongoing Very Large Telescope-Infrared Spectrometer And Array Camera program with nineteen sources from the literature. The new sources extend the existing Sloan Digital Sky Survey (SDSS) sample toward the faint end of the QSO luminosity function. Using a maximum likelihood fitting routine optimized for our spectral decomposition, we estimate the BH mass (M{sub BH}), the Eddington ratio (defined as L{sub bol}/L{sub Edd}), and the Fe II/Mg II line ratio, a proxy for the chemical abundance, to characterize both the central object and the broad-line region gas. The QSOs in our sample host BHs with masses of {approx}10{sup 9} M{sub sun} that are accreting close to the Eddington luminosity, consistent with earlier results. We find that the distribution of observed Eddington ratios is significantly different than that of a luminosity-matched comparison sample of SDSS QSOs at lower redshift (0.35 < z < 2.25): the average (log (L{sub bol}/L{sub Edd})) = -0.37 (L{sub bol}/L{sub Edd} {approx} 0.43) with a scatter of 0.20 dex for the z > 4 sample and the (log (L{sub bol}/L{sub Edd})) = -0.80 (L{sub bol}/L{sub Edd} {approx} 0.16) with a scatter of 0.24 dex for the 0.35 < z < 2.25 sample. This implies that, at a given luminosity, the M{sub BH} at high-z is typically lower than the average M{sub BH} of the lower-redshift population, i.e., the z > 4 sources are accreting significantly faster than the lower-redshift ones. We show that the derived Fe II/Mg II ratios depend sensitively on the analysis performed: our self-consistent, homogeneous analysis significantly reduces the Fe II/Mg II scatter found in previous studies. The measured Fe II/Mg II line ratios show no sign of evolution with cosmic time in the redshift range 4 < z < 6.5. If the Fe II/Mg II line ratio is used as a secondary proxy of the Fe/Mg abundance ratio, this implies that the QSOs in our sample have undergone a major episode of Fe enrichment in the few 100 Myr preceding the cosmic age at which they are observed.
    The Astrophysical Journal 10/2011; 739(2). · 6.73 Impact Factor

Publication Stats

2k Citations
270.73 Total Impact Points

Institutions

  • 2003–2014
    • The University of Arizona
      • Department of Astronomy
      Tucson, Arizona, United States
  • 2011
    • United States Naval Observatory
      Washington, Maine, United States
  • 2005
    • University of Washington Seattle
      • Department of Astronomy
      Seattle, Washington, United States
    • Pennsylvania State University
      • Department of Astronomy and Astrophysics
      University Park, Maryland, United States
  • 1997–2000
    • Princeton University
      • Department of Astrophysical Sciences
      Princeton, New Jersey, United States