George Lake

University of Zurich, Zürich, Zurich, Switzerland

Are you George Lake?

Claim your profile

Publications (108)446.51 Total impact

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Metal poor globular clusters (MPGCs) are a unique probe of the early universe, in particular the reionization era. Systems of globular clusters in galaxy clusters are particularly interesting as it is in the progenitors of galaxy clusters that the earliest reionizing sources first formed. Although the exact physical origin of globular clusters is still debated, it is generally admitted that globular clusters form in early, rare dark matter peaks (Moore et al. 2006; Boley et al. 2009). We provide a fully numerical analysis of the Virgo cluster globular cluster system by identifying the present day globular cluster system with exactly such early, rare dark matter peaks. A popular hypothesis is that that the observed truncation of blue metal poor globular cluster formation is due to reionization (Spitler et al. 2012; Boley et al. 2009; Brodie & Strader 2006); adopting this view, constraining the formation epoch of MPGCs provides a complementary constraint on the epoch of reionization. By analyzing both the line of sight velocity dispersion and the surface density distribution of the present day distribution we are able to constrain the redshift and mass of the dark matter peaks. We find and quantify a dependence on the chosen line of sight of these quantities, whose strength varies with redshift, and coupled with star formation efficiency arguments find a best fitting formation mass and redshift of $\simeq 5 \times 10^8 \rm{M}_\odot$ and $z\simeq 9$. We predict $\simeq 300$ intracluster MPGCs in the Virgo cluster. Our results confirm the techniques pioneered by Moore et al. (2006) when applied to the the Virgo cluster and extend and refine the analytic results of Spitler et al. (2012) numerically.
    Monthly Notices of the Royal Astronomical Society 03/2014; 442(3). DOI:10.1093/mnras/stu1057 · 5.23 Impact Factor
  • Source
    H. Lux · J. I. Read · G. Lake · K. V. Johnston
    [Show abstract] [Hide abstract]
    ABSTRACT: Tidal streams are a powerful probe of the Milky Way (MW) potential shape. In this paper, we introduce a simple test particle method to fit stream data, using a Markov Chain Monte Carlo technique to marginalise over uncertainties in the progenitor's orbit and the Milky Way halo shape parameters. Applying it to mock data of thin streams in the MW halo, we show that, even for very cold streams, stream-orbit offsets - not modelled in our simple method - introduce systematic biases in the recovered shape parameters. For the streams that we consider, and our particular choice of potential parameterisation, these errors are of order ~20% on the halo flattening parameters. However, larger systematic errors can arise for more general streams and potentials; such offsets need to be correctly modelled in order to obtain an unbiased recovery of the underlying potential. Assessing which of the known Milky Way streams are most constraining, we find NGC 5466 and Pal 5 are the most promising candidates. These form an interesting pair as their orbital planes are both approximately perpendicular to each other and to the disc, giving optimal constraints on the MW halo shape. We show that - while with current data their constraints on potential parameters are poor - good radial velocity data along the Pal 5 stream will provide constraints on qz - the flattening perpendicular to the disc. Furthermore, as discussed in a companion paper, NGC 5466 can provide rather strong constraints on the MW halo shape parameters, if the tentative evidence for a departure from the smooth orbit towards its western edge is confirmed.
    Monthly Notices of the Royal Astronomical Society 08/2013; 436(3). DOI:10.1093/mnras/stt1744 · 5.23 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Abstract Every Galactic environment is characterized by a stellar density and a velocity dispersion. With this information from literature, we simulated flyby encounters for several Galactic regions, numerically calculating stellar trajectories as well as orbits for particles in disks; our aim was to understand the effect of typical stellar flybys on planetary (debris) disks in the Milky Way Galaxy. For the solar neighborhood, we examined nearby stars with known distance, proper motions, and radial velocities. We found occurrence of a disturbing impact to the solar planetary disk within the next 8 Myr to be highly unlikely; perturbations to the Oort cloud seem unlikely as well. Current knowledge of the full phase space of stars in the solar neighborhood, however, is rather poor; thus we cannot rule out the existence of a star that is more likely to approach than those for which we have complete kinematic information. We studied the effect of stellar encounters on planetary orbits within the habitable zones of stars in more crowded stellar environments, such as stellar clusters. We found that in open clusters habitable zones are not readily disrupted; this is true if they evaporate in less than 10(8) yr. For older clusters the results may not be the same. We specifically studied the case of Messier 67, one of the oldest open clusters known, and show the effect of this environment on debris disks. We also considered the conditions in globular clusters, the Galactic nucleus, and the Galactic bulge-bar. We calculated the probability of whether Oort clouds exist in these Galactic environments. Key Words: Stellar interactions-Galactic habitable zone-Oort cloud. Astrobiology 13, 491-509.
    Astrobiology 05/2013; DOI:10.1089/ast.2012.0842 · 2.51 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: We trace the evolution of central galaxies in three ∼ 1013 M ⊙ galaxy groups simulated at high resolution in cosmological hydrodynamical simulations. In all three cases, the evolution in the group potential leads, at z = 0, to central galaxies that are massive, gas-poor early-type systems supported by stellar velocity dispersion resembling either elliptical or S0 galaxies, even without including energy feedback from active galactic nuclei (AGNs). Their z ∼ 2−2.5 main progenitors are massive (M ∗ ∼ 3−10×10 10 M⊙), star forming (20−60M⊙/yr) galaxies which host substantial reservoirs of cold gas ( ∼ 5 × 109 M⊙) in extended gas disks. Our simulations thus show that star forming galaxies observed at z ∼ 2 are likely the main progenitors of central galaxies in galaxy groups at z = 0. At z ∼ 2 the stellar component of all galaxies is compact, with a half-mass radius < 1 kpc. The central stellar density stays approximatively constant from such early epochs down to z = 0. Instead, the galaxies grow inside-out, by acquiring a stellar envelope outside the innermost ∼ 2 kpc. Consequently the density within the effective radius decreases by up to two orders of magnitude. Both major and minor mergers contribute to most (70 +20
  • T. Bruch · J. Read · L. Baudis · G. Lake
    [Show abstract] [Hide abstract]
    ABSTRACT: In hierarchical structure formation models of disk galaxies, a dark matter disk forms as massive satellites are preferentially dragged into the disk-plane where they dissolve. Here, we quantify the importance of this dark disk for direct and indirect dark matter detection. The low velocity of the dark disk with respect to the Earth enhances detection rates in direct detection experiments at low recoil energy. For WIMP masses MW IMP � 50 GeV/c 2, the detection rate increases by up to a factor of 3 in the 5 − 20 keV recoil energy range. Comparing this with rates at higher energy is sensitive to MW IMP, providing stronger mass constraints particularly for MW IMP � 100 GeV/c 2. The annual modulation signal is significantly boosted by the dark disk and the modulation phase is shifted by ∼ 3 weeks relative to the dark halo. The variation of the observed phase with recoil energy determines MW IMP, once the dark disk properties are fixed by future astronomical surveys. The low velocity of the particles in the dark disk with respect to the solar system significantly enhances the capture rate of WIMPs in the Sun, leading to an increased flux of neutrinos from the Sun which could be detected in current and future neutrino telescopes. The dark disk contribution to the muon flux from neutrino back conversion at the Earth is increased by a factor of ∼ 5 compared to the SHM, for ρd/ρh = 0.5.
  • Source
    Silvia Garbari · Chao Liu · Justin I. Read · George Lake
    [Show abstract] [Hide abstract]
    ABSTRACT: We apply a new method to determine the local disc matter and dark halo matter density to kinematic and position data for \sim2000 K dwarf stars taken from the literature. Our method assumes only that the disc is locally in dynamical equilibrium, and that the 'tilt' term in the Jeans equations is small up to \sim1 kpc above the plane. We present a new calculation of the photometric distances to the K dwarf stars, and use a Monte Carlo Markov Chain to marginalise over uncertainties in both the baryonic mass distribution, and the velocity and distance errors for each individual star. We perform a series of tests to demonstrate that our results are insensitive to plausible systematic errors in our distance calibration, and we show that our method recovers the correct answer from a dynamically evolved N-body simulation of the Milky Way. We find a local dark matter density of {\rho}dm = 0.025+0.014-0.013 M\odotpc^{-3} (0.95+0.53-0.49 GeV cm^{-3}) at 90% confidence assuming no correction for the non-flatness of the local rotation curve, and {\rho}dm = 0.022+0.015-0.013 M\odotpc^-3 (0.85+0.57-0.50 GeV cm^{-3}) if the correction is included. Our 90% lower bound on {\rho}dm is larger than the canonical value typically assumed in the literature, and is at mild tension with extrapolations from the rotation curve that assume a spherical halo. Our result can be explained by a larger normalisation for the local Milky Way rotation curve, an oblate dark matter halo, a local disc of dark matter, or some combination of these.
    Monthly Notices of the Royal Astronomical Society 05/2012; 425(2). DOI:10.1111/j.1365-2966.2012.21608.x · 5.23 Impact Factor
  • Source
    H. Lux · J. I. Read · G. Lake · K. V. Johnston
    [Show abstract] [Hide abstract]
    ABSTRACT: Stellar streams provide unique probes of galactic potentials, with the longer streams normally providing the cleaner measurements. In this paper, we show an example of a short tidal stream that is particularly sensitive to the shape of the Milky Way's dark matter halo: the globular cluster tidal stream NGC 5466. This stream has an interesting deviation from a smooth orbit at its western edge. We show that such a deviation favours an underlying oblate or triaxial halo (irrespective of plausible variations in the Milky Way disc properties and the specific halo parametrisation chosen); spherical or prolate halo shapes can be excluded at a high confidence level. Therefore, more extensive data sets along the NGC 5466 tidal stream promise strong constraints on the Milky Way halo shape.
    Monthly Notices of the Royal Astronomical Society 04/2012; 424(1). DOI:10.1111/j.1745-3933.2012.01276.x · 5.23 Impact Factor
  • Source
    Silvia Garbari · Justin I. Read · George Lake
    [Show abstract] [Hide abstract]
    ABSTRACT: We study the systematic problems in determining the local dark matter density $\rho_{dm}(R_\odot)$ from kinematics of stars in the Solar Neighbourhood, using a simulated Milky Way-like galaxy. We introduce a new unbiased method for recovering $\rho_{dm}(R_\odot)$ based on the moments of the Jeans equations, combined with a Monte Carlo Markov Chain technique and apply it to real data. ov
    The European Physical Journal Conferences 11/2011; 19(2100-014X). DOI:10.1051/epjconf/20121901008
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We have computed trajectories, distances and times of closest approaches to the Sun by stars in the Solar neighbourhood with known position, radial velocity and proper motions. For this purpose we have used a full potential model of the Galaxy that reproduces the local z-force, the Oort constants, the local escape velocity, and the rotation curve of the Galaxy. From our sample we constructed initial conditions, within observational uncertainties, with a Monte Carlo scheme for the twelve most suspicious candidates because of their small tangential motion. We find that the star Gliese 710 will have the closest approach to the Sun, with a distance of approximately 0.34 pc at 1.36 Myr in the future. We show that the effect of a flyby with the characteristics of Gliese 710 on a 100 AU test particle disk representing the Solar system is negligible. However, since there is a lack of 6D data for a large percentage of stars in the Solar neighbourhood, closer approaches may exist. We calculate parameters of passing stars that would cause noticeable effects on the Solar disk. Regarding the birth cloud of the Sun, we performed experiments to reproduce roughly the observed orbital parameters such as eccentricities and inclinations of the Kuiper Belt. It is known now that in Galactic environments, such as stellar formation regions, the stellar densities of new born stars, are high enough to produce close encounters within 200 AU. Moreover, in these Galactic environments, the velocity dispersion is relatively low, typically approximately 1-3 km s-1.We find that with a velocity dispersion of approximately 1 km s-1 and an approach distance of about 150 AU, typical of these regions, we obtain approximately the eccentricities and inclinations seen in the current Solar system. Simple analytical calculations of stellar encounters effects on the Oort cloud are presented.
    Monthly Notices of the Royal Astronomical Society 08/2011; 418. DOI:10.1111/j.1365-2966.2011.19579.x · 5.23 Impact Factor
  • Source
    Silvia Garbari · Justin I. Read · George Lake
    [Show abstract] [Hide abstract]
    ABSTRACT: We revisit systematics in determining the local dark matter density (rho_dm) from the vertical motion of stars in the Solar Neighbourhood. Using a simulation of a Milky Way-like galaxy, we determine the data-quality required to detect the dark matter density at its expected local value. We introduce a new method for recovering rho_dm that uses moments of the Jeans equations, combined with a Monte Carlo Markov Chain technique to marginalise over the unknown parameters. Given sufficiently good data, we show that our method can recover the correct local dark matter density even in the face of disc inhomogeneities, non-isothermal tracers and a non-separable distribution function. We illustrate the power of our technique by applying it to Hipparcos data [Holmberg & Flynn 2000,2004]. We first make the assumption that the A and F star tracer populations are isothermal. This recovers rho_dm=0.003^{+0.009}_{-0.007}Msun/pc^3 (with 90 per cent confidence), consistent with previous determinations. However, the vertical dispersion profile of these tracers is poorly known. If we assume instead a non-isothermal profile similar to the blue disc stars from SDSS DR-7 [Abazajian et al. 2009] measured by Bond et al. (2009), we obtain a fit with a very similar chi^2 value, but with rho_dm=0.033^{+0.008}_{-0.009}Msun/pc^3 (with 90 per cent confidence). This highlights that it is vital to measure the vertical dispersion profile of the tracers to recover an unbiased estimate of the local dark matter density.
    Monthly Notices of the Royal Astronomical Society 05/2011; 416(3). DOI:10.1111/j.1365-2966.2011.19206.x · 5.23 Impact Factor
  • Source
    H. Lux · J. I. Read · G. Lake
    [Show abstract] [Hide abstract]
    ABSTRACT: We calculate orbits for the Milky Way dwarf galaxies with proper motions, and compare these to subhalo orbits in a high-resolution cosmological simulation. We use the simulation data to assess how well orbits may be recovered in the face of measurement errors, a time-varying triaxial gravitational potential and satellite-satellite interactions. For present measurement uncertainties, we recover the apocentre ra and pericentre rp to ~40 per cent. With improved data from the Gaia satellite we should be able to recover ra and rp to ~14 per cent, respectively. However, recovering the 3D positions and orbital phase of satellites over several orbits is more challenging. This owes primarily to the non-sphericity of the potential and satellite interactions during group infall. Dynamical friction, satellite mass-loss and the mass evolution of the main halo play a more minor role in the uncertainties. We apply our technique to nine Milky Way dwarfs with observed proper motions. We show that their mean apocentre is lower than the mean of the most massive subhaloes in our cosmological simulation, but consistent with the most massive subhaloes that form before z = 10. This lends further support to the idea that the Milky Way's dwarfs formed before reionization.
    01/2010; 406(4). DOI:10.1063/1.3458551
  • Source
    Silvia Garbari · George Lake · Justin I. Read
    [Show abstract] [Hide abstract]
    ABSTRACT: We examine systematic problems in determining the local matter density from the vertical motion of stars, i.e. the 'Oort limit'. Using collisionless simulations and a Monte Carlo Markov Chain technique, we determine the data quality required to detect local dark matter at its expected density. We find that systematic errors are more important than observational errors and apply our technique to Hipparcos data to reassign realistic error bars to the local dark matter density.
    01/2010; DOI:10.1063/1.3458549
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We place constraints on the formation redshifts for blue globular clusters (BGCs), independent of the details of hydrodynamics and population III star formation. The observed radial distribution of BGCs in the Milky Way Galaxy suggests that they formed in biased dark matter halos at high redshift. As a result, simulations of a ~1 Mpc box up to z ~ 10 must resolve BGC formation in LambdaCDM. We find that most halo stars could be produced from destroyed BGCs and other low-mass clusters that formed at high redshift. We present a proof-of-concept simulation that captures the formation of globular-like star clusters.
    The Astrophysical Journal 11/2009; 706(1). DOI:10.1088/0004-637X/706/1/L192 · 6.28 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We place constraints on the formation redshifts for blue globular clusters (BGCs), independent of the details of hydrodynamics and population III star formation. The observed radial distribution of BGCs in the Milky Way Galaxy suggests that they formed in biased dark matter halos at high redshift. As a result, simulations of a ~1 Mpc box up to z~10 must resolve BGC formation in LCDM. We find that most halo stars could be produced from destroyed BGCs and other low-mass clusters that formed at high redshift. We present a proof-of-concept simulation that captures the formation of globular-like star clusters. Comment: Accepted for publication in ApJL
    The Astrophysical Journal 08/2009; · 6.28 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We trace the evolution of central galaxies in three ~10^13 M_sun galaxy groups simulated at high resolution in cosmological hydrodynamical simulations. The evolution in the group potential leads, at z=0, to central galaxies that are massive, gas-poor early-type systems supported by stellar velocity dispersion resembling either elliptical or S0 galaxies. Their z~2-2.5 main progenitors are massive M* ~ 3-10 x 10^10 M_sun, star forming (20-60 M_sun/yr) galaxies which host substantial reservoirs of cold gas (~5 x 10^9 M_sun) in extended gas disks. Our simulations thus show that star forming galaxies observed at z~2 are likely the main progenitors of central galaxies in galaxy groups at z=0. Their central stellar densities stay approximately constant from z~1.5 down to z=0. Instead, the galaxies grow inside-out, by acquiring a stellar envelope outside the innermost ~2 kpc. Consequently the density within the effective radius decreases by up to two orders of magnitude. Both major and minor mergers contribute to most of the mass accreted outside the effective radius and thus drive the evolution of the half-mass radii. In one of the three simulated groups the short central cooling time leads to a dramatic rejuvenation of the central group galaxy at z<1, affecting its morphology, kinematics and colors. This episode is eventually terminated by a group-group merger. Our simulations demonstrate that, in galaxy groups, the interplay between halo mass assembly, galaxy merging and gas accretion has a substantial influence on the star formation histories and z=0 morphologies of central galaxies.[Abridged] Comment: 28 pages, 23 figures, 9 tables, accepted to APJ (revised to match accepted version)
    The Astrophysical Journal 06/2009; 709(1). DOI:10.1088/0004-637X/709/1/218 · 6.28 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The fact that dark matter (DM), thus far, has revealed itself only on scales of galaxies and larger, again thrusts onto astrophysics the opportunity and the responsibility to confront the age old mystery "What is the nature of matter?" By deriving basic data on the nature of DM - e.g., mass of its particle(s), present mean temperature, distribution in galaxies and other structures in the universe, and capacity for dissipational collapse - we will be uncovering the properties of the dominant species of matter in the universe and significantly extending the standard models of particle physics. Determining the mass of the DM particle to an order of magnitude would help to sort out the particle family to which it (or they) belongs. Beyond mass, there are issues of stability. The DM particle may be unstable with a measurable half-life, or it may become unstable after absorbing a certain amount of energy from collisions. In both cases it would contribute to the present hot dark matter component. Some key parameters of DM can most accurately be measured in the very nearby universe because DM dominates the mass in the outer Milky Way (MW), in other galaxies in the Local Group, and in the Local Group in its entirety. The presence and amount of DM can be quantified by study of dynamical processes observable in fine detail within these entities. Precise measurements of 3-D velocities for stars, coherent star streams, and stars in satellite stellar systems out to the edge of the Galaxy can reveal "what is the shape, orientation, density law, and lumpiness of the dark matter halo" as well as "what is the total mass of the Galaxy?"
  • Source
    O Agertz · G Lake · B Moore · L Mayer · R Teyssier · A B Romeo
    [Show abstract] [Hide abstract]
    ABSTRACT: Not Available
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: As disc galaxies form in a hierarchical cosmology, massive merging satellites are preferentially dragged towards the disc plane. The material accreted from these satellites forms a dark matter disc that contributes 0.25–1.5 times the non-rotating halo density at the solar position. Here, we show the importance of the dark disc for indirect dark matter detection in neutrino telescopes. Previous predictions of the neutrino flux from WIMP annihilation in the Earth and the Sun have assumed that Galactic dark matter is spherically distributed with a Gaussian velocity distribution, the standard halo model. Although the dark disc has a local density comparable to the dark halo, its higher phase space density at low velocities greatly enhances capture rates in the Sun and Earth. For typical dark disc properties, the resulting muon flux from the Earth is increased by three orders of magnitude over the SHM, while for the Sun the increase is an order of magnitude. This significantly increases the sensitivity of neutrino telescopes to fix or constrain parameters in WIMP models. The flux from the Earth is extremely sensitive to the detailed properties of the dark disc, while the flux from the Sun is more robust. The enhancement of the muon flux from the dark disc puts the search for WIMP annihilation in the Earth on the same level as the Sun for WIMP masses ≲100 GeV.
    Physics Letters B 02/2009; 674(4-5-674):250-256. DOI:10.1016/j.physletb.2009.03.042 · 6.02 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Making robust predictions for the phase space distribution of dark matter at the solar neighbourhood is vital for dark matter direct detection experiments. To date, almost all such predictions have been based on simulations that model the dark matter alone. Here, we use three cosmological hydrodynamics simulations of bright, disc dominated galaxies to include the effects of baryonic matter self-consistently for the first time. We find that the addition of baryonic physics drastically alters the dark matter profile in the vicinity of the Solar neighbourhood. A stellar/gas disc, already in place at high redshift, causes merging satellites to be dragged preferentially towards the disc plane where they are torn apart by tides. This results in an accreted dark matter disc that contributes ~0.25 - 1.5 times the non-rotating halo density at the solar position. The dark disc, unlike dark matter streams, is an equilibrium structure that must exist in disc galaxies that form in a hierarchical cosmology. Its low rotation lag with respect to the Earth significantly boosts WIMP capture in the Earth and Sun, boosts the annual modulation signal, and leads to distinct variations in the flux as a function of recoil energy that allow the WIMP mass to be determined. Comment: Final version accepted for publication in MNRAS; only minor changes from previous version
    Monthly Notices of the Royal Astronomical Society 01/2009; 397(1). DOI:10.1111/j.1365-2966.2009.14757.x · 5.23 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Predicting the local flux of dark matter particles is vital for dark matter direct detection experiments. To date, such predictions have been based on simulations that model the dark matter alone. Here we include the influence of the baryonic matter for the first time. We use two different approaches. Firstly, we use dark matter only simulations to estimate the expected merger history for a Milky Way mass galaxy, and then add a thin stellar disc to measure its effect. Secondly, we use three cosmological hydrodynamic simulations of Milky Way mass galaxies. In both cases, we find that a stellar/gas disc at high redshift (z~1) causes merging satellites to be preferentially dragged towards the disc plane. This results in an accreted dark matter disc that contributes ~0.25 - 1 times the non-rotating halo density at the solar position. An associated thick stellar disc forms with the dark disc and shares a similar velocity distribution. If these accreted stars can be separated from those that formed in situ, future astronomical surveys will be able to infer the properties of the dark disc from these stars. The dark disc, unlike dark matter streams, is an equilibrium structure that must exist in disc galaxies that form in a hierarchical cosmology. Its low rotation lag with respect to the Earth significantly boosts WIMP capture in the Earth and Sun, increases the likelihood of direct detection at low recoil energy, boosts the annual modulation signal, and leads to distinct variations in the flux as a function of recoil energy that allow the WIMP mass to be determined (see contribution from T. Bruch this volume). Comment: To appear in Proceedings of Science, Identification of dark matter 2008, August 18-22, 2008 Stockholm, Sweden
    01/2009; DOI:10.1063/1.3458542

Publication Stats

7k Citations
446.51 Total Impact Points

Institutions

  • 2006–2014
    • University of Zurich
      • Institut für Theoretische Physik
      Zürich, Zurich, Switzerland
  • 1986–2008
    • University of Washington Seattle
      • Department of Astronomy
      Seattle, WA, United States
  • 2003–2004
    • Washington State University
      Pullman, Washington, United States
  • 1999
    • Seattle BioMed
      Seattle, Washington, United States
    • Durham University
      • Department of Physics
      Durham, England, United Kingdom
  • 1990
    • University of Oxford
      Oxford, England, United Kingdom
  • 1989
    • University of Texas at Austin
      Austin, Texas, United States
  • 1988
    • University of Toronto
      Toronto, Ontario, Canada
  • 1984
    • Rutgers, The State University of New Jersey
      New Brunswick, New Jersey, United States
    • University of Padova
      Padua, Veneto, Italy
  • 1980
    • University of California, Berkeley
      Berkeley, California, United States
  • 1977–1980
    • Princeton University
      • • Department of Astrophysical Sciences
      • • Department of Physics
      Princeton, New Jersey, United States