This contribution presents recent advances in identifying the stellar upper mass limit using simulations of UV radiative feedback during the star formation process. Generally, due to computational costs and a focus on au to parsec scales, simulations of massive star formation do not trace the flow of material to distances closer than a few au from the forming star. However, UV line-acceleration acts directly on accreting material in the sub-au circumstellar region, thereby efficiently ablating the surface layers off the protostellar disk. For stars on the order of a few hundred solar masses, this disk destruction rate exceeds the accretion rate, destroying the disk faster than it is replenished, and setting a maximum stellar mass as a function of metallicity that can be attained by single star formation channels.

We have measured the IMF of the inner Galaxy using ~3000 OGLE-III microlensing events. Each event’s timescale depends on both the lens mass, and the velocities and distances of the lens and source. New dynamical models were used provide the distribution of distances and velocities, and thereby measure the lens mass distribution. Using a power-law or log-normal parameterisation the resultant IMF is indistinguishable from local measurements by Kroupa or Chabrier respectively. The lenses lie in the inner Galaxy where the stars are mostly ~10 Gyr old and formed on a fast α-element enhanced timescale thereby constraining IMF variability with the properties of the collapsing gas cloud. Furthermore microlensing measures the stellar mass budget, including dark remnants, to low mass. Stars contribute most of the mass in the inner Galaxy with a low fraction remaining for dark matter. Reconciling this with local dark matter estimates requires a core or shallow cusp in its profile.

The upper mass limit of stars remains an open question in astrophysics. Here we discuss observations of the most massive stars (greater than 100 solar masses) in the local universe and how the observations fit in with theoretical predictions. In particular, the Large Magellanic Cloud plays host to numerous very massive stars, making it an ideal template to study the roles that environment, metallicity, and multiplicity play in the formation and evolution of the most massive stars. We will discuss the work that is instrumental in laying the groundwork for interpreting future observations by James Webb of starburst regions in the high redshift universe.

We present the results of binary population simulations of carbon- and nitrogen-enhanced metal-poor (CEMP and NEMP) stars. We show that the observed paucity of very nitrogen-rich stars puts strong constraints on possible modifications of the initial mass function at low metallicity.

It has been observed that the ratio of Hα to FUV luminosity (LHα/LFUV) is lower in low surface brightness galaxies. This behaviour has been attributed to systematic variations of the upper mass end and/or the slope of the Initial Mass Function (IMF) Meurer et al. (2009) and Lee et al. (2009)). However these hypotheses do not explain the observed scatter in luminosity ratio (LHα/LFUV). We present a model for the total LHα and LFUV luminosity arising from a randomly populated IMF following the Salpeter power law and the clustering law of Oey & Clarke (2007).

We use images derived from collapsing, turbulent molecular cloud simulations without sinks to explore the effects of finite image angular resolution and noise on the derived clump mass function. These effects randomly perturb the clump masses, producing a lognormal clump mass function with a Salpeter-like high mass end. We show that the characteristic break mass of the simulated clump mass functions changes with the angular resolution of the images in a way that is entirely consistent with the observations. We also present some cautionary tales regarding sink particles and highlight the need to ensure that sinks actually correspond to distinct collapsing objects. We test several popular numerical sink criteria in the literature and compare to converged, non-sink results.

We present simulations of supersonic collisions between molecular clouds of mass 500 M⊙ and radius 2.24 pc. The simulations are performed with the SEREN SPH code. The code treats the energy equation and the associated transport of heating and cooling radiation. The formation of protostars is captured by introducing sink particles. Low velocity collisions form a shock-compressed layer which fragments to form stars. For high-velocity collisions, υrel ⪆ 5 km s−1, the non-linear thin shell instability strongly disrupts the shock-compressed layer, and may inhibit the formation of stars.

The formation of the first stars out of metal-free gas appears to result in stars at least an order of magnitude more massive than in the present-day case. We here consider what controls the transition from a primordial to a modern initial mass function. We study the influence of low levels of metal enrichment and different initial conditions on the cooling and collapse of initially ionized gas in small protogalactic halos using three-dimensional, smoothed particle hydrodynamics simulations. We argue that fragmentation at moderate density depends on the initial conditions for star formation more than on the metal abundances present.

Extremely metal-poor (EMP) stars are thought to be formed in the low-mass protogalaxies as building blocks of the Milky Way and can be probes to investigate the early stage of galaxy formation and star formation in the early universe. We study the formation history of EMP stars in the Milky Way halo using a new model of chemical evolution based on the hierarchical theory of the galaxy formation. We construct the merging history of the Milky Way halo based on the extended Press-Schechter formalism, and follow the star formation and chemical evolution along the merger tree. The abundance trends and number of low-mass stars predicted in our model are compared with those of observed EMP stars. Additionally, in order to clarify the origin of hyper metal poor stars, we investigate the change of the surface metal abundances of stars by accretion of interstellar matter. We also investigate the pre-enrichment of intergalactic matter by the first supernovae.

We compared deep images acquired with the Advanced Camera for Surveys on board of the Hubble Space Telescope with mid-IR Spitzer Space Telescope images and University College London Echelle Spectrograph spectra of NGC 346 and NGC 602, two of the youngest star clusters in the Small Magellanic Cloud. Our multi-wavelength approach allowed us to infer very different origins for the clusters: while NGC 346 is likely the result of the hierarchical collapse of a giant molecular cloud, NGC 602 is probably the result of the collision and consequent interaction of two H I shells of gas.

The stellar initial mass function (IMF) is a very important question in modern astrophysics. Globular clusters (GCs) are good samples for studying the IMF, but the Galactic GCs can provide only one time-scale evolutionary stage. The Large Magellanic Cloud (LMC) is an ideal environment for studying the IMF because it contains compact clusters at different evolutionary stages. By studying the IMF at different evolutionary stages, we can see how the mass function evolves with time.

We present the results of binary population simulations of carbon-enhanced metal-poor (CEMP) stars. We show that nitrogen and fluorine are useful tracers of the origin of CEMP stars, and conclude that the observed paucity of very nitrogen-rich stars puts strong constraints on possible modifications of the initial mass function at low metallicity. The large number fraction of CEMP stars may instead require much more efficient dredge-up from low-metallicity asymptotic giant branch stars.

Two aspects of our recent N-body studies of star clusters are presented:

1. 1) What impact does mass segregation and selective mass loss have on integrated photometry?

2. 2) How well do results compare from N-body simulations using NBODY4 and STARLAB/KIRA?

We present a study of the structure of the Galactic Bulge in the region −6° < b < 3°,−17° < l < 17° using 2MASS archive data. More than 100 fields are used. We make use of the red clump method to derive the distance of the Bulge in the studied regions. We derive a position angle of the Galactic Bulge going from 42° ± 11 to 35° ± 10 depending on the adopted bulge mass distribution.

Basing on recent microlensing observations, we analyse the mass spectrum of the Galactic bulge stellar population and study the slope of the initial mass function.

The combination of stellar population synthesis models and photoionization models allows a better understanding of the spectral features of HII regions and HII galaxies. In this work we show that sampling effects in the initial mass function (IMF) are very important in the low cluster mass case. To this aim, we compute photoionization models ionized by realistic clusters made up of various combinations of individual stars and clusters made up with a synthesis model. We discuss the differences in the position on diagnostic diagrams and their implications.

While theoretical understanding remains to be clarified regarding the mechanisms that may or may not limit stellar masses, it is possible to empirically evaluate the existence of an upper-mass limit. ZAMS masses of the most massive stars have been estimated for a range of environments in our local Milky Way neighborhood and the Magellanic Clouds. Various statistical techniques demonstrate the existence of an upper-mass limit in this stellar sample.

Near infrared dust extinction mapping is opening a new window on molecular cloud research. Applying a straightforward technique to near infrared large scale data of nearby molecular complexes one can easily construct density maps with dynamic ranges in column density covering, 3σ~ 0.5 < AV< 50 mag or 1021<N<1023 cm−2. These maps are unique in capturing the low column density distribution of gas in molecular cloud complexes, where most of the mass resides, and at the same time allow the identification of dense cores (n~104cm−3) which are the precursors of stars. For example, the application of this technique to the nearby Pipe Nebula complex revealed the presence of 159 dense cores (the largest sample of such object in one single complex) whose mass spectrum presents the first robust evidence for a departure from a single power-law. The form of this mass function is surprisingly similar in shape to the stellar IMF but scaled to a higher mass by a factor of about 3. This suggests that the distribution of stellar birth masses (IMF) is the direct product of the dense core mass function and a uniform star formation efficiency of 30%±10%, and that the stellar IMF may already be fixed during or before the earliest stages of core evolution. We are now extending this technique to extra-galactic mapping of Giant molecular Clouds (GMCs), and although a much less straightforward task, preliminary results indicate that the GMC mass spectrum in M83 and Centaurus A is a power-law characterized by α~−2 unlike CO results which suggest α~−1.

It has been recently shown by several authors that fragmentation of pre-stellar gas (i.e. at densities from 104 to 1010 particles cm−3 and temperatures of order 10-30 K) depends on the gas thermodynamics much more than it was anticipated in earlier studies, in which only an isothermal behaviour has been assumed for the gas. Here we review the results of a number of numerical hydrodynamic simulations (e.g. Li et al. 2003, Jappsen et al. 2005, Bonnell et al. 2006) in which departure from isothermality has been attempted by employing a polytropic equation of state (eos) with exponent different from unity. In particular, in these studies it has been shown that the dominant fragmentation scale of pre-stellar gas, and hence the peak of the initial mass function (IMF), depends on a polytropic exponent that changes value, from below to above unity, at a critical density (Larson 2005). Furthermore, this piecewise polytropic eos depends on the gas metallicity and fundamental constants. Therefore, the peak of the IMF depends, in turn, also on the gas metallicity and fundamental constants rather than on initial conditions, as it has been previously suggested (e.g. Larson 1995). Hence, we are for the first time in a position to infer theoretically the notion of a universal IMF (at least for its low-mass end).

We also present two test cases in which a non-isothermal eos has been used in the context of smoothed particle hydrodynamic (SPH) numerical simulations. In the first case star formation is triggered by means of low-mass clump collisions. These calculations have shown that clump collisions can be a relatively efficient mechanism for the formation of solar-mass protostars and their lower-mass companions (efficiency greater or of order 20-25%; Kitsionas & Whitworth 2006). We have also found that in such collisions protostars form mainly by fragmentation of dense filaments along which it is likely that pairs of protostars capture each other in close binaries surrounded by circumbinary discs. In the second case, the use of a polytropic eos with a varying exponent appropriate for the metallicity of starburst regions (Spaans & Silk 2000, 2005) is shown to be sufficient to obtain a top heavy IMF similar to that observed e.g. in the Galactic centre (Klessen, Spaans & Jappsen 2006). These are preliminary results in the direction of revisiting earlier isothermal calculations that were resolving all densities up to the opacity limit for fragmentation (e.g. Bate et al. 2002ab, 2003), this time also taking into account the thermal properties of the gas in the density range between 104 and 1010 particles cm−3. The next step would be to include self-consistent radiation transport in the calculations, the first attempts for which are already in the making (e.g. Whitehouse & Bate 2004).

Any predictive theory of star formation must explain observed variations (or lack thereof) in the initial mass function. Recent work suggests that we might expect quantitative variations in the IMF as a function of metallicity (Larson 2005) or magnetic field strength (Shu et al. 2004). We summarize results from several on-going studies attempting to constrain the ratio of high to low mass stars, as well as stars to sub- stellar objects, in a variety of different environments, all containing high mass stars.

First, we examine the ratio of stars to sub–stellar objects in the nearby Mon R2 region utilizing NICMOS/HST data. We compare our results to the IMF by Kroupa (2002) and to the observed ratios for IC 348 and Orion. Second, we present preliminary results for the ratio of high to low mass stars in W51, the most luminous HII region in the galaxy. Based on ground–based multi–colour images of the cluster obtained with the MMT adaptive optics system, we derive a lower limit to the ratio of high-mass to low-mass stars and compare it to the ratios for nearby clusters. Finally, we present the derived IMF for the R136 region in the LMC where the metallicity is 1/4 solar using HST/NICMOS data. We find that the IMF is consistent with that characterizing the field (Chabrier 2003), as well as nearby star–forming regions, down to 1.0 M$_\odot$ outside 2 pc. Whereas the results for both Mon R2 and R136 are consistent with the nearby clusters, the ratio of high to low mass stars in W51 tentatively indicates a lack of low–mass objects.