Program

Three-body interactions between stellar binaries and black holes in the core of a globular cluster can eject stars from the cluster at high velocities, much larger than the central velocity dispersion. This may be used to probe the central population of black holes and binaries, including the possible presence (or absence!) of yet unconfirmed intermediate-mass black holes. Proper motions from Gaia now allow us to systematically search for high-velocity stars from globular clusters and have a fresh look at this phenomenon, to complement the few previously identified high-velocity star candidates detected based on their radial velocity. In this talk, I will discuss our search for high-velocity stars around Milky Way globular clusters and what constraints these observations can place on the black hole content of these systems.

The Milky Way consists of a large number of star clusters, with over 150 being labelled as old, metal poor globular clusters and thousands being labelled as young, metal-rich open clusters. However these numbers are small compared to the total number of clusters that have ever existed in the Milky Way, as most have completely dissolved between their time of formation and the present day. Once a star cluster dissolves, all that is left behind is the thin stream of stars that once preceded and trailed the cluster as its tidal tails. In this talk, I will explore how the properties of tidal tails can be used to better understand the dissolved star cluster population of the Milky Way. The length and width of the tails, variation in the stellar mass function along the tails, and the properties of surviving planetary systems within the tails all depend on the properties of progenitor cluster.

Three-body interactions can eject stars from the core of a globular cluster, causing them to enter the Galactic halo as extra-tidal stars. While finding extra-tidal stars is imperative for understanding both star cluster and Galaxy evolution, connecting isolated extra-tidal field stars back to their birth cluster is extremely difficult. In this work, we present a new methodology consisting of high-dimensional data analysis and a particle spray code to identify extra-tidal stars of any Galactic globular cluster using M3 as a case study. Using the t-Stochastic Neighbour Embedding (t-SNE) and Uniform Manifold Approximation and Projection (UMAP) machine learning dimensionality reduction algorithms, we first identify a set of extra-tidal candidates in the APOGEE DR16 data catalogue with similar chemical abundances. To confirm each candidate's extra-tidal nature, we introduce Corespray — a new Python-based three-body particle spray code that simulates extra-tidal stars for any globular cluster in the Milky Way. By comparing the kinematics and photometry of our extra-tidal candidates to the simulated Corespray stars, we identify a sample of new extra-tidal stars of M3. Future applications of Corespray will yield better understandings of core dynamics, star formation histories and binary fractions in globular clusters throughout the Galaxy.

The distribution of stars and stellar remnants (white dwarfs, neutron stars, black holes) within globular clusters holds clues about their formation and long-term evolution, with important implications for their initial mass function and the formation of black hole mergers. In this work, we infer best-fit equilibrium multimass models for a large number of Milky Way globular clusters, which are compared to various datasets, including proper motions from Gaia EDR3 and HST, line-of-sight velocities from ground-based spectroscopy and stellar mass functions from HST. The speed of the distribution-function based models used allows for precise determinations of the model parameters via nested sampling, ensuring the closest fit to all observables. These models allow us to describe many aspects of a globular cluster's structural and dynamical makeup. In particular, based on constraints of the mass distribution of white dwarfs, we are able to infer the shape of the system's initial mass function at masses higher than the current turnoff mass and explore potential relationships with other cluster parameters such as metallicity.

An important component of globular clusters is their binaries. These binaries are dynamically affected by their interactions within the cluster, potentially resulting in their disruption or ionization. Ionization occurs preferentially to binaries with binding energy less than the average kinetic energy of a star in the cluster. As the average kinetic energy is related to the density of the cluster, this results in a higher ionization rate in the case of a dense globular cluster. Since the majority of binary disruptions occur early on in the clusters evolution when a cluster is at its densest, the current population of binaries and their period disruption can be used to infer back to the initial density of the globular cluster. Using this idea, this work implements hierarchial bayesian methods along with multi-epoch radial velocity measurements to put constraints on the intrinsic binary period distribution with particular emphasis on understanding the long period tail of the distribution. This has important consequences for our understanding of globular clusters and other astronomical topics such as black holes in globular clusters and gravitational waves.

I present spectroscopy of individual stars in 26 Magellanic Cloud (MC) star clusters with the aim of estimating dynamical masses and V-band mass-to-light (M/L_V) ratios over a wide range in age and metallicity. We obtained 3137 high-resolution stellar spectra with M2FS on the Magellan/Clay Telescope. Combined with 239 published spectroscopic results of comparable quality, we produced a final sample of 2787 stars with good quality spectra for kinematic analysis in the target clusters. Line-of-sight velocities measured from these spectra and stellar positions within each cluster were used in a customized expectation-maximization (EM) technique to estimate cluster membership probabilities. Using appropriate cluster structural parameters and corresponding single-mass dynamical models, this technique ultimately provides self-consistent total mass and M/L_V estimates for each cluster. Mean metallicities for the clusters were also obtained and tied to a scale based on calcium IR triplet metallicities. We present trends of the cluster M/LV values with cluster age, mass, and metallicity, and find that our results run about 40 per cent on average lower than the predictions of a set of simple stellar population (SSP) models. Modified SSP models that account for internal and external dynamical effects greatly improve agreement with our results, as can models that adopt a strongly bottom-light IMF. To the extent that dynamical evolution must occur, a modified IMF is not required to match data and models. In contrast, a bottom-heavy IMF is ruled out for our cluster sample as this would lead to higher predicted M/LV values, significantly increasing the discrepancy with our observations.

We present a Bayesian inference approach to estimating the mass profile and the phase space center of a globular cluster (GC) given the spatial and kinematic information of its stars based on the lowered-isothermal (LIMEPY) model. As a first step towards more realistic modelling of GC, we built a differentiable, accurate emulator of the LIMEPY distribution function using interpolation. The reliable gradient information provided by the emulator allows the use of Hamiltonian Monte Carlo methods to sample large Bayesian models with hundreds of parameters. We explore the use of hierarchical Bayesian models to address several issues encountered in observations of GC including the incomplete data, the measurement errors, and the selection bias. Our approach can incorporate these uncertainties and bias in a robust and statistically consistent way.

A gamma-ray excess observed from our Galactic Center may be produced by annihilating dark matter, or by many millisecond pulsars (MSPs). We simulated the spatial distribution of populations of MSPs, using two plausible kick distributions and the Galactic potential, and found that a population of MSPs produced by normal stellar evolution cannot explain the spatial distribution of the observed gamma-ray excess. The alternatives are that the gamma-rays are produced by MSPs generated by stellar interactions in the nuclear cluster (a somewhat different environment than globular clusters), that a large fraction of the original Galactic globular clusters have dissolved in the Galactic center, or that the gamma-ray excess is created by annihilating dark matter.

Currently, ultraluminous X-ray sources (ULXs) with globular cluster (GC) counterparts have been identified. This is exciting, as ULXs have been theorized as potential intermediate mass black holes. New black hole mergers detected by LIGO-Virgo may also be associated with GC’s, underscoring the importance of ULXs as a potential linkage between GC electromagnetic and gravitational wave source populations. GC ULXs show a diverse behaviour with regards to temporal variability, both on long (16 years) and short (~hours) timescales, in both the X-ray and optical wavelengths. They can switch on or off over the course of many years or remain at a constant luminosity. Some sources exhibit a long-term change in their luminosity with no discernible variability within the other observations, other sources show a stunning long-term variability while also demonstrating variability on the timescale of around four hours. I will undertake a comprehensive comparison of the temporal variability of the zoo of currently known GC ULXs, discuss the possible origins of some of the extreme variability observed, and how this informs on our knowledge of black hole populations in extragalactic globular clusters.

We present a comprehensive census of X-ray millisecond pulsars (MSPs) in 29 Galactic globular clusters (GCs), including 68 MSPs with confirmed X-ray luminosities and 107 MSPs with X-ray upper limits. We compile previous X-ray studies of GC MSPs, and add new analyses of six MSPs (PSRs J1326-4728A, J1326-4728B, J1518+0204C, J1717+4308A, J1737-0314A, and J1807-2459A) discovered in five GCs. Their X-ray spectra are well described by a single blackbody model, a single power-law model, or a combination of them, with X-ray luminosities ranging from 1.9 × 10^30 to 8.3 × 10^31 erg s^-1. We find that most detected X-ray MSPs have luminosities between ~10^30 and 3 × 10^31 erg s^-1. Redback pulsars are a relatively bright MSP population with X-ray luminosities of ~2 × 10^31-3 × 10^32 erg s^-1. Black widows show a bimodal distribution in X-ray luminosities, with eclipsing black widows between ~7 × 10^30 and 2 × 10^31 erg s^-1, while the two confirmed non-eclipsing black widows are much fainter, with L_X of 1.5-3 × 10^30 erg s^-1, suggesting an intrinsic difference in the populations. We estimate the total number of MSPs in 36 GCs by considering the correlation between the number of MSPs and stellar encounter rate in GCs, and suggest that between 600 and 1500 MSPs exist in these 36 GCs. Finally, we estimate the number of X-ray-detectable MSPs in the Galactic bulge, finding that 1-86 MSPs with L_X > 10^33 erg s^-1, and 20-900 MSPs with L_X > 10^32 erg s^-1 should be detectable there.

The early, simplistic picture of star clusters is that they consist of stars that formed in unison out of the same gas. However, we now know that clusters house multiple populations with varied chemical compositions. About half of the stars have the same abundances as field stars. The other half are enriched with the products of high-temperature hydrogen burning. No source of enrichment has yet been demonstrated to quantitatively match observations. With the majority of stars formed in clusters, our understanding of star formation is tethered to resolving the origin of multiple populations in clusters. We examine massive binaries as a potential enrichment source. We simulate a suite of massive interacting binaries using Modules for Experiments in Stellar Astrophysics (MESA) and calculate the composition of their ejected material. We analyze the parameter space of binaries expected to contribute to enrichment, in stellar masses and orbital period. Select binaries eject gas with the desired enhancements and depletions in the light elements.

It is well-established from observational and computational studies that binary stars play an important role in driving the long-term structural evolution of star clusters. On the other hand, the influence of binaries over the structure of much younger star clusters, in which stars remain embedded in their natal gas and star formation is still ongoing, is not well understood. Understanding the interplay between the physics of clustered star formation and of binary dynamics is essential, as stars are preferentially formed both in clustered environments and in binary systems. In this work, we investigate the impact of different populations of binary stars on the structure of embedded clusters over timescales shorter than the duration of star formation. To this end, we analyze a suite of coupled radiation hydrodynamics and direct N-body simulations of star cluster formation. Our simulations include star formation, stellar evolution, and stellar feedback in the form of winds, direct radiation pressure and ionizing radiation. We compare here the structural properties of the clusters formed in the presence of different populations of binaries.

In this talk, I discuss cloud-scale analytic models of star formation and the impact of different physical processes, including self-gravity, turbulence, magnetic fields, and stellar feedback, on the evolution of a star forming region. We use a suite of 3D hydrodynamical simulations of star-forming molecular clouds, and trace the role of these physical processes using a variety of metrics including the star formation rate (SFR), the gas density probability distribution function (PDF), and the rate at which gas is expanding and contracting. We find that magnetic fields and protostellar outflows both decrease the SFR, and that the PDF is not purely lognormal when outflows and self-gravity are considered. The simulation with outflows has a time-varying excess of diffuse gas compared to simulations without outflows, exhibits increased average sonic Mach number, and maintains a slower SFR over the entire duration of the run. We also investigate the rate at which the gas is collapsing or expanding, and find that gas under the influence of only self-gravity collapses at approximately the free-fall rate, while turbulence increases the mass flow rate at low densities and protostellar outflows produce small amounts of rapidly expanding gas. This work confirms the importance of stellar feedback in reducing the efficiency of star formation efficiency, and connects this to the dynamics of the gas in star forming regions.

Many stars form in stellar clusters that dissolve into the galactic field on timescales of tens to hundreds of millions of years. Planet formation takes place in a protoplanetary disk around a young star, and these disks have typical lifetimes of a few millions years. The process of planet formation thus typically takes place in a stellar cluster environment. Understanding the impact of this environment on protoplanetary disks, and subsequently on planet formation, requires multi-scale, multi-physics models. I will report on recent simulations of protoplanetary disk populations in young star forming regions. These simulations combine the collapse of a giant molecular cloud, the formation of stars, and stellar feedback (using the Torch model), with the evolution of protoplanetary disks around the newly formed stars. These disks evolve viscously, and are subject to truncation due to stellar encounters and to evaporation due to radiation from nearby massive stars. I will focus on the importance of the extinction of photoevaporating radiation due to intracluster gas on the early evolution of the protoplanetary disks.

An ideal computational star formation simulation would track an enormous spatial scale that included galactic dynamics and retained detail down to the physics of single stars. However, such ambitions are computationally prohibitive. We attempt to bridge this gap between galaxy simulations and simulations of individual star forming clouds. We extract GMC structures and dynamics from galaxy simulations using AREPO and use them as initial conditions for a more detailed simulation of star cluster formation using our software Torch which couples FLASH and AMUSE . This technique allows us to track the dynamics and feedback physics of individual stars in the context of a collapsing GMC that formed under self-consistent galactic conditions.

Most stars are formed in clustered groups inside Giant Molecular Clouds (GMCs) which involves the conversion of molecular gas into stars along dense filaments inside the GMC. This process results in the formation of small clustered groups of stars (subclusters) that can further accrete surrounding gas, form stars, and merge with other subclusters throughout their life in the GMC. Therefore, in order to model this formation properly, the inclusion of both stellar dynamics and hydrodynamics in numerical simulations is essential.
In this work, we focus on the mergers between gas-rich subclusters. Larger scale GMC simulations illustrate the importance of this process in the formation of massive star clusters, but they lack the resolution to fully model it on the scale of individual stars. Here, we take global characteristics of individual sub- clusters present in previous GMC simulations, resolve them into their stellar and gaseous components, and simulate their merger. We evolve both components simultaneously using N-body and Smoothed Particle Hydrodynamics solvers present in the AMUSE framework. This represents one step closer towards fully understanding the formation of the massive star clusters we see today.

Massive stars are important drivers of the interstellar medium due to their explosive deaths which enrich surrounding gas and push it outwards. Yet, these stars are also enshrouded in mystery due to their scarcity and the optically thick HII regions surrounding their birth sites. To learn more about these stars, it is important to look at multiple scales involved in the star formation process. The formation and environments of molecular clouds and star clusters will affect the final stars that form within them. We have created a test bed of the cluster scale as a first step to a galactic multi-scale simulation of star formation. From this I will present the results of analyzing cluster formation, fragmentation, and filamentary feeding. I will also present a breakdown of the properties of star clusters in a magnetized CNM environment, serving as a preliminary effort towards high resolution galaxy simulations zooming down to the cluster scale and beyond.

High spatial resolution surveys of star formation in our and nearby galaxies indicate that star cluster formation is strongly linked with the formation and evolution of giant molecular clouds that are highly filamentary. We have undertaken high resolution multiscale simulations of GMC and cluster formation within magnetized spiral galaxy disks using the RAMSES code. Our simulated disk galaxy undergoes supernova driven feedback processes that regulate the overall ISM and star formation dynamics providing a backdrop for our zoom in simulations. These descend from 10s of kpc down to 0.3 pc scales by focusing on particular patches of the galaxy. Filament formation is found on all scales. In particular GMCs arise as filamentary structures within a larger scale cold neutral medium, often at the intersection of expanding SN driven bubbles. By charting the gravitational stability of these filaments, we see that cluster formation is initiated by fragmentation when GMC filaments exceed a critical mass per unit length. The turbulent ISM in sheared galactic flows also produces rotating, flattened disk like structures on 100 pc scales that undergo Toomre instability resulting in spiral arm like filaments and their cluster fragments. We discuss many consequences of these simulations for observations of GMCs, cluster formation, and the associated structure of magnetic fields in these regions.

In this contribution, I will present a joint analysis of molecular clouds and star clusters in the nearby galaxy M33. Assuming all clouds eventually form stars means molecular clouds transition from a phase of no star formation to a phase where enough stars are formed to destroy the cloud. We can find the time from the beginning of cluster formation to the destruction by finding the fraction of young clusters associated with clouds in a galaxy. In M33, cluster positions, ages, masses, and radii were identified by citizen scientists in Johnson et al. (2022). These M33 clusters were identified from the HST-PHATTER survey (Williams et al. 2021). We identify molecular clouds from ALMA CO(2-1) maps using the SCIMES algorithm. We find that young clusters (<10Myr) are more likely to be associated with molecular clouds with an average separation of less than 100pc. We find no correlation between cluster mass and properties of molecular clouds such as mass, surface density, and pressure. There is also no strong preference for young clusters to drift in a particular direction from molecular clouds. Using these catalogues and assuming the fraction of spatial associations corresponds to the fraction of cloud lifetimes that are actively star-forming, we estimate the lifespan of molecular clouds in M33 to be ~30Myr.

Various studies show that star clusters have power-law shape mass functions with similar slope as those of giant molecular clouds (GMCs). This similarity hints that more massive GMCs tend to form more massive clusters. Howard+2018 show in their simulation that the maximal cluster mass that is formed in a GMC is tightly correlated with the GMC’s mass. In this study, we try to confirm this correlation in observations by matching the young massive star clusters (YMCs) with their host GMCs in 3 interactive systems, the Antennae, NGC 3256 and M51. We extract the star cluster catalog identified using HST data and GMC catalog identified using CO data with the CPROPS cloud-finding algorithm. We then corrected the astrometry error for the star cluster catalog and find the closest GMC for each YMC (age<10 Myr). We find a strong correlation between the maximal YMC mass and GMC mass, with Spearman coefficient of 0.76. This correlation coefficient also seems to be dependent on our age and distance (2D spatial distance between matched GMCs and YMCs) selection criterion. We find the optimal age cut of 6 Myr and distance cut of 2 GMC radius. In the next step, we will explore if this correlation reflects differences in galaxy global properties or a universal correlation between GMC and YMC mass.

Stellar cluster populations in the Local Universe show a wide range in properties, suggesting that these objects form via a unique physical channel, and that their demographics are shaped by their formation and evolution in an evolving cosmic environment. This scenario links the current cluster formation sites in the disks of the Antennae galaxies to the old GC population that mostly populates the halo of the Milky Way, implying that their evolution is tightly coupled to that of their host galaxy. To understand the observed cluster populations, it has become necessary the use of numerical simulations that can model the co-formation and evolution of stellar clusters alongside their galactic environments over a Hubble time. In this talk, I will introduce the EMP-Pathfinder simulations (Reina-Campos+ subm.). As a key ingredient, we include the description of the multiphase nature of the interstellar medium, allowing star formation only in the cold, dense gas. I will show that GC populations emerge self-consistently in this scenario after 10 Gyr of co-evolution with their host galaxies. Lastly, I will discuss how cluster demographics can be used as diagnosis tools for baryonic models that modify the cold, gas reservoir within galaxies in galaxy formation simulations.

The mass of galaxies’ globular cluster (GC) populations have been found to correlate strongly with their stellar mass and halo mass. While these relations are relatively well constrained in the high-mass regime, there exists much uncertainty in how these relations behave for low mass galaxies. The lowest mass galaxies, many of which do not host any GCs, are often excluded from models entirely due to observational limitations or the inability of simple linear models to incorporate zero values. We introduce an errors-in-variables Bayesian lognormal hurdle model to model the relation between galaxy stellar mass and GC system mass for the lowest mass galaxies. This Bayesian model contains a logarithmic portion to model the turnoff point at which galaxies are large enough to host clusters as well as a linear portion to model the mass relation in galaxies that do have cluster populations. The model incorporates measurement errors in galaxy luminosity and converts measured luminosities to masses within its hierarchical structure by sampling for individual M/L ratios based on a Chabrier IMF. This model has the ability to fully encompass the range of dwarf galaxies, both those that do and do not host GCs, to shed light on the conditions needed to form and retain cluster populations.

The prevailing assumptions in the use of globular clusters (GCs) to trace the origins of extragalactic systems are that they are both simple stellar populations and essentially clones of the GCs observed in detail in the Local Group. There is, however, accumulating evidence against both these assumptions. We do not know how or when GCs formed or what their evolutionary history was like. To answer these fundamental questions, we need a better understanding of their internal properties, such as detailed abundance patterns and stellar mass functions, and connect that understanding with broad properties of GC systems. As new facilities come online that will provide observations of increasingly distant GCs, we need to ensure that we have the tools in place to accurately interpret the integrated light of extragalactic GCs. In this talk I will summarize the work I have done, and am currently doing, in optimizing precision stellar population synthesis models for integrated spectroscopy of individual GCs. I will discuss the impact of these models for both understanding the formation pathways of "compact stellar systems" (e.g., GCs and ultra-compact dwarf galaxies) via measurement of their initial mass functions and in elucidating the assembly histories of distant galaxies.

This research proposes a new method of determining radial density distributions of globular cluster systems (GCS) using Voronoi tessellations, which can be also applied to a variety of other situations. Voronoi tessellations allow for finer-grained local information about densities and thus allows for GCs associated with distributions from satellite galaxies to be better identified and removed from the target galaxy. This methodology will provide more accurate estimates of GCS masses, especially for brightest cluster galaxies with multiple, significant satellite systems. The goal of this research is to apply this method to better constrain the relation between GCS mass and halo mass between galaxies, investigated by Harris et al. (2018).

In the talk, I will present the latest results from a spectroscopic program I am leading which surveys the star clusters in Milky Way's outer halo. I will discuss the chemical and orbital properties of these clusters and compare them with the dwarf galaxies, stellar streams, and other globular clusters in the Milky Way.

We introduce a new method for detecting ultra-diffuse galaxies by searching for over-densities in intergalactic globular cluster populations. Our approach is based on an application of the log-Gaussian Cox process, which is a commonly used model in the spatial statistics literature but rarely used in astronomy. This method is applied to the globular cluster data obtained from the PIPER survey, a Hubble Space Telescope imaging program targeting the Perseus cluster. We successfully detect all confirmed ultra-diffuse galaxies with known globular cluster populations in the survey. We also identify a potential galaxy that has no detected diffuse stellar content. Preliminary analysis shows that it is unlikely to be merely an accidental clump of globular clusters or other objects. If confirmed, this system would be the first of its kind.

I present Hubble Space Telescope photometry in optical (F475X, 475 nm) and near- infrared (F110W, 1.1 μm) bands of the globular cluster (GC) system of the inner haloes of a sample of 15 brightest cluster galaxies (BCGs). I also present a quantitative model of the relation between (F475X - F110W) colour and cluster metallicity, using simulated GCs. The sample comprises massive elliptical galaxies in a range of environments, from sparsely populated groups to dense clusters. Because the material available for large galaxies to accrete varies with environment and GC systems of such galaxies are built up through accretion, I expect the metallicity distribution of the GC systems in my sample to vary with galaxy environment. GC systems in massive elliptical galaxies tend to follow a bimodal colour distribution, with two subpopulations of blue (metal-poor) and red (metal-rich) clusters. The photometry is used to create a completeness-corrected metallicity histogram for each galaxy in my sample, and to fit a double Gaussian curve to each histogram in order to model the two subpopulations. Finally, the properties of the GC metallicity distribution are correlated against each BCG environment. I found that almost no GCS properties and host galaxy environmental properties are correlated, with the exception of weak but consistent correlations between number of GCs and nth- nearest neighbour surface density and between blue fraction and nth-nearest neighbour surface density.

I will present photometry of the globular cluster systems around 26 distant, giant ETGs, using raw data from the HST and a strictly homogeneous set of procedures witf Dolphot. Metallicity distribution functions are derived from the color indices through a combination of SSP models and spectroscopic calibration. Bimodal-Gaussian fits to the MDFs are remarkably accurate, but the nonlinear shift from color to metallicity makes a significant change to previous views about the proportions of red and blue clusters. Those proportions also show interesting trends with galaxy mass, in line with current models for GCS formation.