We present preliminary results of a GPU porting of all main Gadget3 modules (gravity computation, SPH density computation, SPH hydrodynamic force, and thermal conduction) using OpenACC directives. Here we assign one GPU to each MPI rank and exploit both the host and accellerator capabilities by overlapping computations on the CPUs and GPUs: while GPUs asynchronously compute interactions between particles within their MPI ranks, CPUs perform tree-walks and MPI communications of neighbouring particles. We profile various portions of the code to understand the origin of our speedup, where we find that a peak speedup is not achieved because of time-steps with few active particles. We run a hydrodynamic cosmological simulation from the Magneticum project, with $2\cdot10^{7}$ particles, where we find a final total speedup of $\approx 2.$ We also present the results of an encouraging scaling test of a preliminary gravity-only OpenACC porting, run in the context of the EuroHack17 event, where the prototype of the porting proved to keep a constant speedup up to $1024$ GPUs.

We present a novel and fast application to generate glass-like initial conditions for Lagrangian hydrodynamic schemes (e.g. Smoothed Particle Hydrodynamics (SPH)) following arbitrary density models based on weighted Voronoi tessellations and combine it with improved initial configurations and an additional particle reshuffling scheme. We show our application's ability to sample different kinds of density features and to converge properly towards the given model density as well as a glass-like particle configuration. We analyse convergence with iterations as well as with varying particle number. Additionally, we demonstrate the versatility of the implemented algorithms by providing an extensive test suite for standard (magneto-) hydrodynamic test cases as well as a few common astrophysical applications. We indicate the potential to bridge further between observational astronomy and simulations as well as applicability to other fields of science by advanced features such as describing a density model using gridded data for exampling from an image file instead of an analytic model.

GANDALF is a new hydrodynamics and N-body dynamics code designed for investigating planet formation, star formation and star cluster problems. GANDALF is written in C++, parallelised with both OpenMP and MPI and contains a python library for analysis and visualisation. The code has been written with a fully object-oriented approach to easily allow user-defined implementations of physics modules or other algorithms. The code currently contains implementations of Smoothed Particle Hydrodynamics, Meshless Finite-Volume and collisional N-body schemes, but can easily be adapted to include additional particle schemes. We present in this paper the details of its implementation, results from the test suite, serial and parallel performance results and discuss the planned future development. The code is freely available as an open source project on the code-hosting website github at https://github.com/gandalfcode/gandalf and is available under the GPLv2 license.

We present the results of smoothed particle hydrodynamic simulations investigating the evolution and fragmentation of filaments that are accreting from a turbulent medium. We show that the presence of turbulence, and the resulting inhomogeneities in the accretion flow, play a significant role in the fragmentation process. Filaments which experience a weakly turbulent accretion flow fragment in a two-tier hierarchical fashion, similar to the fragmentation pattern seen in the Orion Integral Shaped Filament. Increasing the energy in the turbulent velocity field results in more sub-structure within the filaments, and one sees a shift from gravity-dominated fragmentation to turbulence-dominated fragmentation. The sub-structure formed in the filaments is elongated and roughly parallel to the longitudinal axis of the filament, similar to the fibres seen in observations of Taurus, and suggests that the fray and fragment scenario is a possible mechanism for the production of fibres. We show that the formation of these fibre-like structures is linked to the vorticity of the velocity field inside the filament and the filament's accretion from an inhomogeneous medium. Moreover, we find that accretion is able to drive and sustain roughly sonic levels of turbulence inside the filaments, but is not able to prevent radial collapse once the filaments become supercritical. However, the supercritical filaments which contain fibre-like structures do not collapse radially, suggesting that fibrous filaments may not necessarily become radially unstable once they reach the critical line-density.

We perform SPH simulations to study precession and changes in alignment between the circumprimary disc and the binary orbit in misaligned binary systems. We find that the precession process can be described by the rigid-disc approximation, where the disc is considered as a rigid body interacting with the binary companion only gravitationally. Precession also causes change in alignment between the rotational axis of the disc and the spin axis of the primary star. This type of alignment is of great important for explaining the origin of spin-orbit misaligned planetary systems. However, we find that the rigid-disc approximation fails to describe changes in alignment between the disc and the binary orbit. This is because the alignment process is a consequence of interactions that involve the fluidity of the disc, such as the tidal interaction and the encounter interaction. Furthermore, simulation results show that there are not only alignment processes, which bring the components towards alignment, but also anti-alignment processes, which tend to misalign the components. The alignment process dominates in systems with misalignment angle near 90 degrees, while the anti-alignment process dominates in systems with the misalignment angle near 0 or 180 degrees. This means that highly misaligned systems will become more aligned but slightly misaligned systems will become more misaligned.

We use smoothed particle hydrodynamic simulations to investigate the growth of perturbations in infinitely long, initially sub-critical but accreting filaments. The growth of these perturbations leads to filament fragmentation and the formation of cores. Most previous work on this subject has been confined to the growth and fragmentation of equilibrium filaments and has found that there exists a preferential fragmentation length scale which is roughly 4 times the filament's diameter. Our results show a more complicated dispersion relation with a series of peaks linking perturbation wavelength and growth rate. These are due to gravo-acoustic oscillations along the longitudinal axis during the sub-critical phase of growth. The positions of the peaks in growth rate have a strong dependence on both the mass accretion rate onto the filament and the temperature of the gas. When seeded with a multi-wavelength density power spectrum there exists a clear preferred core separation equal to the largest peak in the dispersion relation. Our results allow one to estimate a minimum age for a filament which is breaking up into regularly spaced fragments, as well as a maximum accretion rate. We apply the model to observations of filaments in Taurus by Tafalla & Hacar (2015) and find accretion rates consistent with those estimated by Palmeirim et al. (2013).

We use Smoothed Particle Hydrodynamics to explore the circumstances under which an isolated very-low-mass prestellar core can be formed by colliding turbulent flows and collapse to form a brown-dwarf. Our simulations suggest that the flows need not be very fast, but do need to be very strongly convergent, i.e. the gas must flow in at comparable speeds from all sides, which seems rather unlikely. We therefore revisit the object Oph-B11, which Andre, Ward-Thompson and Greaves (2012) have identified as a prestellar core with mass between $\sim 0.020\,\mathrm{M_\odot}$ and $\sim 0.030\,\mathrm{M_\odot}$. We reanalyse the observations using a Markov-chain Monte Carlo method that allows us (i) to include the uncertainties on the distance, temperature and dust mass opacity, and (ii) to consider different Bayesian prior distributions of the mass. We estimate that the posterior probability that Oph-B11 has a mass below the hydrogen burning limit at $\sim 0.075\,\mathrm{M_\odot}$, is between 0.66 and 0.86. We conclude that, if Oph-B11 is destined to collapse, it probably will form a brown dwarf. However, the flows required to trigger this appear to be so contrived that it is difficult to envisage this being the only way, or even a major way, of forming isolated brown dwarfs. Moreover, Oph-B11 could easily be a transient, bouncing, prolate core, seen end-on; there could, indeed should, be many such objects masquerading as very low-mass prestellar cores.

Disc fragmentation provides an important mechanism for producing low mass stars in prestellar cores. Here, we describe Smoothed Particle Hydrodynamics simulations which show how populations of prestellar cores evolve into stars. We find the observed masses and multiplicities of stars can be recovered under certain conditions. First, protostellar feedback from a star must be episodic. The continuous accretion of disc material on to a central protostar results in local temperatures which are too high for disc fragmentation. If, however, the accretion occurs in intense outbursts, separated by a downtime of $\sim10^4\,\mathrm{years}$, gravitational instabilities can develop and the disc can fragment. Second, a significant amount of the cores' internal kinetic energy should be in solenoidal turbulent modes. Cores with less than a third of their kinetic energy in solenoidal modes have insufficient angular momentum to form fragmenting discs. In the absence of discs, cores can fragment but results in a top heavy distribution of masses with very few low mass objects.

Ionising feedback from massive stars dramatically affects the interstellar medium local to star forming regions. Numerical simulations are now starting to include enough complexity to produce morphologies and gas properties that are not too dissimilar from observations. The comparison between the density fields produced by hydrodynamical simulations and observations at given wavelengths relies however on photoionisation/chemistry and radiative transfer calculations. We present here an implementation of Monte Carlo radiation transport through a Voronoi tessellation in the photoionisation and dust radiative transfer code MOCASSIN. We show for the first time a synthetic spectrum and synthetic emission line maps of an hydrodynamical simulation of a molecular cloud affected by massive stellar feedback. We show that the approach on which previous work is based, which remapped hydrodynamical density fields onto Cartesian grids before performing radiative transfer/photoionisation calculations, results in significant errors in the temperature and ionisation structure of the region. Furthermore, we describe the mathematical process of tracing photon energy packets through a Voronoi tessellation, including optimisations, treating problematic cases and boundary conditions. We perform various benchmarks using both the original version of MOCASSIN and the modified version using the Voronoi tessellation. We show that for uniform grids, or equivalently a cubic lattice of cell generating points, the new Voronoi version gives the same results as the original Cartesian-grid version of MOCASSIN for all benchmarks. For non-uniform initial conditions, such as using snapshots from Smoothed Particle Hydrodynamics simulations, we show that the Voronoi version performs better than the Cartesian grid version, resulting in much better resolution in dense regions.

We present the results of SPH simulations in which two clouds, each having mass $M_{_{\rm{o}}}\!=\!500\,{\rm M}_{_\odot}$ and radius $R_{_{\rm{o}}}\!=\!2\,{\rm pc}$, collide head-on at relative velocities of $\Delta v_{_{\rm{o}}} =2.4,\;2.8,\;3.2,\;3.6\;{\rm and}\;4.0\,{\rm km}\,{\rm s}^{-1}$. There is a clear trend with increasing $\Delta v_{_{\rm{o}}}$. At low $\Delta v_{_{\rm{o}}}$, star formation starts later, and the shock-compressed layer breaks up into an array of predominantly radial filaments; stars condense out of these filaments and fall, together with residual gas, towards the centre of the layer, to form a single large-$N$ cluster, which then evolves by competitive accretion, producing one or two very massive protostars and a diaspora of ejected (mainly low-mass) protostars; the pattern of filaments is reminiscent of the hub and spokes systems identified recently by observers. At high $\Delta v_{_{\rm{o}}}$, star formation occurs sooner and the shock-compressed layer breaks up into a network of filaments; the pattern of filaments here is more like a spider's web, with several small-$N$ clusters forming independently of one another, in cores at the intersections of filaments, and since each core only spawns a small number of protostars, there are fewer ejections of protostars. As the relative velocity is increased, the {\it mean} protostellar mass increases, but the {\it maximum} protostellar mass and the width of the mass function both decrease. We use a Minimal Spanning Tree to analyse the spatial distributions of protostars formed at different relative velocities.

StarBench is a project focused on benchmarking and validating different star-formation and stellar feedback codes. In this first StarBench paper we perform a comparison study of the D-type expansion of an HII region. The aim of this work is to understand the differences observed between the twelve participating numerical codes against the various analytical expressions examining the D-type phase of HII region expansion. To do this, we propose two well-defined tests which are tackled by 1D and 3D grid- and SPH- based codes. The first test examines the `early phase' D-type scenario during which the mechanical pressure driving the expansion is significantly larger than the thermal pressure of the neutral medium. The second test examines the `late phase' D-type scenario during which the system relaxes to pressure equilibrium with the external medium. Although they are mutually in excellent agreement, all twelve participating codes follow a modified expansion law that deviates significantly from the classical Spitzer solution in both scenarios. We present a semi-empirical formula combining the two different solutions appropriate to both early and late phases that agrees with high-resolution simulations to $\lesssim2\%$. This formula provides a much better benchmark solution for code validation than the Spitzer solution. The present comparison has validated the participating codes and through this project we provide a dataset for calibrating the treatment of ionizing radiation hydrodynamics codes.

Massive clumps within the swept-up shells of bubbles, like that surrounding the galactic HII region RCW 120, have been interpreted in terms of the Collect and Collapse (C&C) mechanism for triggered star formation. The cold, dusty clumps surrounding RCW 120 are arranged in an almost spherical shell and harbour many young stellar objects. By performing high-resolution, three-dimensional SPH simulations of HII regions expanding into fractal molecular clouds, we investigate whether the formation of massive clumps in dense, swept-up shells necessarily requires the C&C mechanism. In a second step, we use RADMC-3D to compute the synthetic dust continuum emission from our simulations, in order to compare them with observations of RCW 120 made with APEX-LABOCA at 870 micron. We show that a distribution of clumps similar to the one seen in RCW 120 can readily be explained by a non-uniform initial molecular cloud structure. Hence, a shell-like configuration of massive clumps does not imply that the C&C mechanism is at work. Rather, we find a hybrid form of triggering, which combines elements of C&C and Radiatively Driven Implosion (RDI). In addition, we investigate the reliability of deriving clump masses from their 870 micron emission. We find that for clumps with more than 100 M_sun the observational estimates are accurate to within a factor of two and that, even at these long wavelengths, it is important to account for the radiative heating from triggered, embedded protostars.

We present the results of an ensemble of SPH simulations that follow the evolution of prestellar cores for $0.2\,{\rm Myr}$. All the cores have the same mass, and start with the same radius, density profile, thermal and turbulent energy. Our purpose is to explore the consequences of varying the fraction of turbulent energy, $\delta_\mathrm{sol}$, that is solenoidal, as opposed to compressive; specifically we consider $\delta_\mathrm{sol}=1,\,2/3,\,1/3,\,1/9\;{\rm and}\;0$. For each value of $\delta_\mathrm{sol}$, we follow ten different realisations of the turbulent velocity field, in order also to have a measure of the stochastic variance blurring any systematic trends. With low $\delta_\mathrm{sol}(<\!1/3)$ filament fragmentation dominates and delivers relatively high mass stars. Conversely, with high values of $\delta_\mathrm{sol}(>\!1/3)$ disc fragmentation dominates and delivers relatively low mass stars. There are no discernible systematic trends in the multiplicity statistics obtained with different $\delta_\mathrm{sol}$.

Massive stars shape the surrounding ISM by emitting ionizing photons and ejecting material through stellar winds. To study the impact of the momentum from the wind of a massive star on the surrounding neutral or ionized material, we implemented a new HEALPix-based momentum conserving wind scheme in the Smoothed Particle Hydrodynamics (SPH) code SEREN. A qualitative study of the impact of the feedback from an O7.5-like star on a self gravitating sphere shows that, on its own, the transfer of momentum from a wind onto cold surrounding gas has both a compressing and dispersing effect. It mostly affects gas at low and intermediate densities. When combined with a stellar source's ionizing UV radiation, we find the momentum driven wind to have little direct effect on the gas. We conclude that, during a massive star's main sequence, the UV ionizing radiation is the main feedback mechanism shaping and compressing the cold gas. Overall, the wind's effects on the dense gas dynamics and on the triggering of star formation are very modest. The structures formed in the ionization-only simulation and in the combined feedback simulation are remarkably similar. However, in the combined feedback case, different SPH particles end up being compressed. This indicates that the microphysics of gas mixing differ between the two feedback simulations and that the winds can contribute to the localized redistribution and reshuffling of gas.

Lomax et al. have constructed an ensemble of 60 prestellar cores having masses, sizes, projected shapes, temperatures and non-thermal radial velocity dispersions that match, statistically, the cores in Ophiuchus; and have simulated the evolution of these cores using SPH. Each core has been evolved once with no radiative feedback from stars, once with continuous radiative feedback, and once with episodic radiative feedback. Here we analyse the multiplicity statistics from these simulations. With episodic radiative feedback, (i) the multiplicity frequency is ~60% higher than in the field; (ii) the multiplicity frequency and the mean semi-major axis both increase with primary mass; (iii) one third of multiple systems are hierarchical systems with more than two components; (iv) in these hierarchical systems the inner pairings typically have separations of a few au and mass ratios concentrated towards unity, whereas the outer pairings have separations of order 100 au and a flatter distribution of mass ratios. The binary statistics are compatible with observations of young embedded populations, and -- if wider orbits are disrupted preferentially by external perturbations -- with observations of mature field populations. With no radiative feedback, the results are similar to those from simulations with episodic feedback. With continuous radiative feedback, brown dwarfs are under-produced, the number of multiple systems is too low, and the statistical properties of multiple systems are at variance with observation. This suggests that star formation in Ophiuchus may only be representative of global star formation if accretion onto protostars, and hence radiative feedback, is episodic.

Most stars form in a clustered environment. Therefore, it is important to assess how this environment influences the evolution of protoplanetary discs around young stars. In turn, this affects their ability to produce planets and ultimately life. We present here for the first time 3D SPH/N-body simulations that include both the hydrodynamical evolution of the discs around their natal stars, as well as the dynamics of the stars themselves. The discs are viscously evolving, accreting mass onto the central star and spreading. We find penetrating encounters to be very destructive for the discs as in previous studies, although the frequency of such encounters is low. We also find, however, that encounter influence the disc radii more strongly than other disc properties such as the disc mass. The disc sizes are set by the competition between viscous spreading and the disruptive effect of encounters. As discs spread, encounters become more and more important. In the regime of rapid spreading encounters simply truncate the discs, stripping the outer portions. In the opposite regime, we find that the effect of many distant encounters is able to limit the disc size. Finally, we predict from our simulations that disc sizes are limited by encounters at stellar densities exceeding $\sim 2-3 \times 10^3 \ \mathrm{pc}^{-2}$.

We have simulated star formation in prestellar cores, using SPH and initial conditions informed by observations of the cores in Ophiuchus. Because the observations are limited to two spatial dimensions plus radial velocity, we cannot infer initial conditions for the collapse of a particular core. However, with a minimum of assumptions (isotropic turbulence with a power-law spectrum, a thermal mix of compressive and solenoidal modes, a critical Bonnor-Ebert density profile) we can generate initial conditions that match, in a statistical sense, the distributions of mass, projected size and aspect ratio, thermal and non-thermal one-dimensional velocity dispersion, observed in Ophiuchus. The time between core-core collisions in Ophiuchus is sufficiently long, that we can simulate single cores evolving is isolation, and therefore we are able to resolve masses well below the opacity limit. We generate an ensemble of 100 cores, and evolve them with no radiative feedback from the stars formed, then with continuous radiative feedback, and finally with episodic radiative feedback. With no feedback the simulations produce too many brown dwarfs, and with continuous feedback too few. With episodic radiative feedback, both the peak of the protostellar mass function (at ~ 0.2 M_sun) and the ratio of H-burning stars to brown dwarfs are consistent with observations. The mass of a star is not strongly related to the mass of the core in which it forms. Low-mass cores (M ~ 0.1 M_sun) tend to collapse into single objects, whereas high-mass cores (1 > M_sun) usually fragment into several objects with a broad mass range.

Infrared shells and bubbles are ubiquitous in the Galaxy and can generally be associated with HII regions formed around young, massive stars. In this paper, we use high-resolution 3D SPH simulations to explore the effect of a single O7 star emitting photons at 10^49 1/s and located at the centre of a molecular cloud with mass 10^4 M_sun and radius 6.4 pc; the internal structure of the cloud is characterised by its fractal dimension, D (with 2.0 <= D <= 2.8), and the variance of its (log-normal) density distribution, sigma_0^2 (with 0.36 <= sigma_0^2 <= 1.42). Our study focuses on the morphology of the swept-up cold gas and the distribution and statistics of the resulting star formation. If the fractal dimension is low, the border of the HII region is dominated by extended shell-like structures, and these break up into a small number of massive high-density clumps which then spawn star clusters; star formation occurs relatively quickly, and delivers somewhat higher stellar masses. Conversely, if the fractal dimension is high, the border of the HII region is dominated by a large number of pillars and cometary globules, which contain compact dense clumps and tend to spawn single stars or individual multiple systems; star formation occurs later, the stellar masses are somewhat lower, and the stars are more widely distributed.

We compare the results for a set of hydrodynamical tests performed with the AMR finite volume code, MG and the SPH code, SEREN. The test suite includes shock tube tests, with and without cooling, the non-linear thin-shell instability and the Kelvin-Helmholtz instability. The main conclusions are : (i) the two methods converge in the limit of high resolution and accuracy in most cases. All tests show good agreement when numerical effects (e.g. discontinuities in SPH) are properly treated. (ii) Both methods can capture adiabatic shocks and well-resolved cooling shocks perfectly well with standard prescriptions. However, they both have problems when dealing with under-resolved cooling shocks, or strictly isothermal shocks, at high Mach numbers. The finite volume code only works well at 1st order and even then requires some additional artificial viscosity. SPH requires either a larger value of the artificial viscosity parameter, alpha_AV, or a modified form of the standard artificial viscosity term using the harmonic mean of the density, rather than the arithmetic mean. (iii) Some SPH simulations require larger kernels to increase neighbour number and reduce particle noise in order to achieve agreement with finite volume simulations. However, this is partly due to the need to reduce noise that can corrupt the growth of small-scale perturbations. In contrast, instabilities seeded from large-scale perturbations do not require more neighbours and hence work well with standard SPH formulations and converge with the finite volume simulations. (iv) For purely hydrodynamical problems, SPH simulations take an order of magnitude longer to run than finite volume simulations when running at equivalent resolutions, i.e. when they both resolve the underlying physics to the same degree. This requires about 2-3 times as many particles as the number of cells.

Numerical simulations of star formation frequently rely on the implementation of sink particles, (a) to avoid expending computational resource on the detailed internal physics of individual collapsing protostars, (b) to derive mass functions, binary statistics and clustering kinematics (and hence to make comparisons with observation), and (c) to model radiative and mechanical feedback; sink particles are also used in other contexts, for example to represent accreting black holes in galactic nuclei. We present a new algorithm for creating and evolving sink particles in SPH simulations, which appears to represent a significant improvement over existing algorithms {\refrpt -- particularly in situations where sinks are introduced after the gas has become optically thick to its own cooling radiation and started to heat up by adiabatic compression}. (i) It avoids spurious creation of sinks. (ii) It regulates the accretion of matter onto a sink so as to mitigate non-physical perturbations in the vicinity of the sink. (iii) Sinks accrete matter, but the associated angular momentum is transferred back to the surrounding medium. With the new algorithm -- and modulo the need to invoke sufficient resolution to capture the physics preceding sink formation -- the properties of sinks formed in simulations are essentially independent of the user-defined parameters of sink creation, or the number of SPH particles used.

We present a new hybrid Smoothed Particle Hydrodynamics (SPH)/N-body method for modelling the collisional stellar dynamics of young clusters in a live gas background. By deriving the equations of motion from Lagrangian mechanics we obtain a formally conservative combined SPH/N-body scheme. The SPH gas particles are integrated with a 2nd order Leapfrog, and the stars with a 4th order Hermite scheme. Our new approach is intended to bridge the divide between the detailed, but expensive, full hydrodynamical simulations of star formation, and pure N-body simulations of gas-free star clusters. We have implemented this hybrid approach in the SPH code SEREN (Hubber et al. 2011) and perform a series of simple tests to demonstrate the fidelity of the algorithm and its conservation properties. We investigate and present resolution criteria to adequately resolve the density field and to prevent strong numerical scattering effects. Future developments will include a more sophisticated treatment of binaries.

Protostars grow in mass by accreting material through their discs, and this accretion is initially their main source of luminosity. The resulting radiative feedback heats the environments of young protostars, and may thereby suppress further fragmentation and star formation. There is growing evidence that the accretion of material onto protostars is episodic rather than continuous; most of it happens in short bursts that last up to a few hundred years, whereas the intervals between these outbursts of accretion could be thousands of years. We have developed a model to include the effects of episodic accretion in simulations of star formation. Episodic accretion results in episodic radiative feedback, which heats and temporarily stabilises the disc, suppressing the growth of gravitational instabilities. However, once an outburst has been terminated, the luminosity of the protostar is low, and the disc cools rapidly. Provided that there is enough time between successive outbursts, the disc may become gravitationally unstable and fragment. The model suggests that episodic accretion may allow disc fragmentation if (i) the time between successive outbursts is longer than the dynamical timescale for the growth of gravitational instabilities (a few kyr), and (ii) the quiescent accretion rate onto the protostar is sufficiently low (at most a few times 1e-7 Msun/yr). We also find that after a few protostars form in the disc, their own episodic accretion events shorten the intervals between successive outbursts, and sup- press further fragmentation, thus limiting the number of objects forming in the disc. We conclude that episodic accretion moderates the effect of radiative feedback from young protostars on their environments, and, under certain conditions, allows the formation of low-mass stars, brown dwarfs, and planetary-mass objects by fragmentation of protostellar discs.

The role of feedback from massive stars is believed to be a key element in the evolution of molecular clouds. We use high-resolution 3D SPH simulations to explore the dynamical effects of a single O7 star located at the centre of a molecular cloud with mass 10^4M_sun and radius 6.4pc. The initial internal structure of the cloud is characterised by its fractal dimension, D=2.0 - 2.8, and its log-normal density PDF. (i) As regards star formation, in the short term ionising feedback is positive, in the sense that star formation occurs much more quickly in gas that is compressed by the high pressure of the ionised gas. However, in the long term ionising feedback is negative, in the sense that most of the cloud is dispersed with an outflow rate of up to ~0.01M_sun/yr, on a timescale comparable with the sound-crossing time for the ionised gas (~1-2Myr), and triggered star formation is therefore limited to a few percent of the cloud's mass. (ii) As regards the morphology of the ionisation fronts (IFs) bounding the HII region and the systematics of outflowing gas, we distinguish two regimes. For low D<=2.2, the initial cloud is dominated by large-scale structures, so the neutral gas tends to be swept up into a few extended coherent shells, and the ionised gas blows out through a few large holes between these shells; we term these HII regions "shell-dominated". Conversely, for high D>=2.6, the initial cloud is dominated by small-scale structures, and these are quickly overrun by the advancing IF, thereby producing neutral pillars whilst the ionised gas blows out through a large number of small holes between the pillars; we term these HII regions "pillar-dominated". (iii) As regards the injection of bulk kinetic energy, by ~1Myr, the expansion of the HII region has delivered a rms velocity of ~6km/s; this represents less than 0.1% of the total energy radiated by the O7 star.

It is speculated that the accretion of material onto young protostars is episodic. We present a computational method to include the effects of episodic accretion in radiation hydrodynamic simulations of star formation. We find that during accretion events protostars are "switched on", heating and stabilising the discs around them. However, these events typically last only a few hundred years, whereas the intervals in between them may last for a few thousand years. During these intervals the protostars are effectively "switched off", allowing gravitational instabilities to develop in their discs and induce fragmentation. Thus, episodic accretion promotes disc frag- mentation, enabling the formation of low-mass stars, brown dwarfs and planetary-mass objects. The frequency and the duration of episodic accretion events may be responsible for the low-mass end of the IMF, i.e. for more than 60% of all stars.

We present simulations of initially stable isothermal clouds exposed to ionising radiation from a discrete external source, and identify the conditions that lead to radiatively driven implosion and star formation. We use the Smoothed Particle Hydrodynamics code SEREN and an HEALPix-based photo-ionisation algorithm to simulate the propagation of the ionising radiation and the resulting dynamical evolution of the cloud. We find that the incident ionising flux, $\Phi_{_{\rm LyC}}$, is the critical parameter determining the cloud evolution. At moderate fluxes, a large fraction of the cloud mass is converted into stars. As the flux is increased, the fraction of the cloud mass that is converted into stars and the mean masses of the individual stars both decrease. Very high fluxes simply disperse the cloud. Newly-formed stars tend to be concentrated along the central axis of the cloud (i.e. the axis pointing in the direction of the incident flux). For given cloud parameters, the time, $t_{_\star}$, at which star formation starts is proportional to $\Phi_{_{\rm LyC}}^{-1/3}$. The pattern of star formation found in the simulations is similar to that observed in bright-rimmed clouds.

A star acquires much of its mass by accreting material from a disc. Accretion is probably not continuous but episodic. We have developed a method to include the effects of episodic accretion in simulations of star formation. Episodic accretion results in bursts of radiative feedback, during which a protostar is very luminous, and its surrounding disc is heated and stabilised. These bursts typically last only a few hundred years. In contrast, the lulls between bursts may last a few thousand years; during these lulls the luminosity of the protostar is very low, and its disc cools and fragments. Thus, episodic accretion enables the formation of low-mass stars, brown dwarfs and planetary-mass objects by disc fragmentation. If episodic accretion is a common phenomenon among young protostars, then the frequency and duration of accretion bursts may be critical in determining the low-mass end of the stellar initial mass function.

We present SEREN, a new hybrid Smoothed Particle Hydrodynamics and N-body code designed to simulate astrophysical processes such as star and planet formation. It is written in Fortran 95/2003 and has been parallelised using OpenMP. SEREN is designed in a flexible, modular style, thereby allowing a large number of options to be selected or disabled easily and without compromising performance. SEREN uses the conservative `grad-h' formulation of SPH, but can easily be configured to use traditional SPH or Godunov SPH. Thermal physics is treated either with a barotropic equation of state, or by solving the energy equation and modelling the transport of cooling radiation. A Barnes-Hut tree is used to obtain neighbour lists and compute gravitational accelerations efficiently, and an hierarchical time-stepping scheme is used to reduce the number of computations per timestep. Dense gravitationally bound objects are replaced by sink particles, to allow the simulation to be evolved longer, and to facilitate the identification of protostars and the compilation of stellar and binary properties. At the termination of a hydrodynamical simulation, SEREN has the option of switching to a pure N-body simulation, using a 4th-order Hermite integrator, and following the ballistic evolution of the sink particles (e.g. to determine the final binary statistics once a star cluster has relaxed). We describe in detail all the algorithms implemented in SEREN and we present the results of a suite of tests designed to demonstrate the fidelity of SEREN and its performance and scalability. Further information and additional tests of SEREN can be found at the web-page http://www.astro.group.shef.ac.uk/seren.

We present simulations of initially stable isothermal clouds exposed to ionizing radiation from a discrete external source, and identify the conditions that lead to radiatively driven implosion and star formation. We use the Smoothed Particle Hydrodynamics code SEREN (Hubber et al. 2010) and the HEALPix-based photoionization algorithm described in Bisbas et al. (2009). We find that the incident ionizing flux is the critical parameter determining the evolution: high fluxes simply disperse the cloud, whereas low fluxes trigger star formation. We find a clear connection between the intensity of the incident flux and the parameters of star formation.

We describe a new algorithm for including the dynamical effects of ionizing radiation in SPH simulations, and we present several examples of how the algorithm can be applied to problems in star formation. We use the HEALPix software to tessellate the sky and to solve the equation of ionization equilibrium along a ray towards each of the resulting tesserae. We exploit the hierarchical nature of HEALPix to make the algorithm adaptive, so that fine angular resolution is invoked only where it is needed, and the computational cost is kept low. We present simulations of (i) the spherically symmetric expansion of an HII region inside a uniform-density, non--self-gravitating cloud; (ii) the spherically symmetric expansion of an HII region inside a uniform-density, self-gravitating cloud; (iii) the expansion of an off-centre HII region inside a uniform-density, non--self-gravitating cloud, resulting in rocket acceleration and dispersal of the cloud; and (iv) radiatively driven compression and ablation of a core overrun by an HII region. The new algorithm provides the means to explore and evaluate the role of ionizing radiation in regulating the efficiency and statistics of star formation.

We suggest that low-mass hydrogen-burning stars like the Sun should sometimes form with massive extended discs; and we show, by means of radiation hydrodynamic simulations, that the outer parts of such discs (R>100 AU) are likely to fragment on a dynamical timescale (10^3 to $10^4 yr), forming low-mass companions: principally brown dwarfs (BDs), but also very low-mass hydrogen-burning stars and planetary-mass objects. A few of the BDs formed in this way remain attached to the primary star, orbiting at large radii. The majority are released into the field, by interactions amongst themselves; in so doing they acquire only a low velocity dispersion (<2 km/s), and therefore they usually retain small discs, capable of registering an infrared excess and sustaining accretion. Some BDs form close BD/BD binaries, and these binaries can survive ejection into the field. This BD formation mechanism appears to avoid some of the problems associated with the `embryo ejection' scenario, and to answer some of the questions not yet answered by the `turbulent fragmentation' scenario.

We examine the spatial distribution of brown dwarfs produced by the decay of small-N stellar systems as expected from the embryo ejection scenario. We model a cluster of several hundred stars grouped into 'cores' of a few stars/brown dwarfs. These cores decay, preferentially ejecting their lowest-mass members. Brown dwarfs are found to have a wider spatial distribution than stars, however once the effects of limited survey areas and unresolved binaries are taken into account it can be difficult to distinguish between clusters with many or no ejections. A large difference between the distributions probably indicates that ejections have occurred, however similar distributions sometimes arise even with ejections. Thus the spatial distribution of brown dwarfs is not necessarily a good discriminator between ejection and non-ejection scenarios.

Jeans showed analytically that, in an infinite uniform-density isothermal gas, plane-wave perturbations collapse to dense sheets if their wavelength, $\lambda$, satisfies $\lambda > \lambda_{_{\rm JEANS}} = (\pi a^2 / G \rho_{_0})^{1/2}$ (where $a$ is the isothermal sound speed and $\rho_{_0}$ is the unperturbed density); in contrast, perturbations with smaller $\lambda$ oscillate about the uniform density state. Here we show that Smoothed Particle Hydrodynamics reproduces these results well, even when the diameters of the SPH particles are twice the wavelength of the perturbation. Our simulations are performed in 3-D with initially settled (i.e. non-crystalline) distributions of particles. Therefore there exists the seed noise for artificial fragmentation, but it does not occur. We conclude that, although there may be -- as with any numerical scheme -- `skeletons in the SPH cupboard', a propensity to fragment artificially is evidently not one of them.

We present a simple model of binary star formation based on the assumption that rotating prestellar cores collapse to form rings and these rings then fragment into protostars. We assume that each ring spawns a small number (N <= 6) of protostars, and that the condensation of the protostars is sufficiently rapid that they can subsequently be treated as point masses. We simulate the dynamical evolution of a ring of N stars and record the properties of the single stars, binaries and higher multiples that form as a result of the dissolution of the ring. The masses of the individual stars in a ring are drawn from a log-normal distribution with dispersion \sigma. This part of the model is perfomed for many different realizations of the ring, to obtain good statistics. The second part of the model is to convolve the results of these dimensionless simulations, first with the distribution of core masses, which yields the distributions of multiplicity,mass ratio and eccentricity, as a function of primary mass; and second with the distribution of core angular momenta, which yields the distributions of semi-major axis and period, again as a function of primary mass. Using the observed distribution of core masses, and the observed distribution of core angular momenta, our model is able to reproduce the observed IMF, the observed high multiplicity frequency of pre-Main Sequence stars, the observed distribution of separations, and -- for long-period systems -- the observed distributions of eccentricity and mass-ratio, provided we invoke N = 4 or 5 and $\sigma = 0.6$. We presume that the reduced multiplicity frequency in the field, compared with young clusters, is the result of dynamical interactions between stars formed in different cores but the same cluster, following ring dissolution.