Peter D. Mitchell

A unified multiwavelength model of galaxy formation

We present a new version of the GALFORM semi-analytical model of galaxy formation. This brings together several previous developments of GALFORM into a single unified model, including a different initial mass function (IMF) in quiescent star formation and in starbursts, feedback from active galactic nuclei supressing gas cooling in massive halos, and a new empirical star formation law in galaxy disks based on their molecular gas content. In addition, we have updated the cosmology, introduced a more accurate treatment of dynamical friction acting on satellite galaxies, and updated the stellar population model. The new model is able to simultaneously explain both the observed evolution of the K-band luminosity function and stellar mass function, and the number counts and redshift distribution of sub-mm galaxies selected at 850μm. This was not previously achieved by a single physical model within the ΛCDM framework, but requires having an IMF in starbursts that is somewhat top-heavy. The new model is tested against a wide variety of observational data covering wavelengths from the far-UV to sub-mm, and redshifts from z = 0 to z = 6, and is found to be generally successful. These observations include the optical and near-IR luminosity functions, HI mass function, fraction of early type galaxies, Tully-Fisher, metallicity-luminosity and size-luminosity relations at z = 0, as well as far-IR number counts, and far-UV luminosity functions at z ∼ 3 − 6. Discrepancies are however found in galaxy sizes and metallicities at low luminosities, and in the abundance of low mass galaxies at high-z, suggesting the need for a more sophisticated model of supernova feedback.

How well can we really estimate the stellar masses of galaxies from broad-band photometry?

The estimated stellar masses of galaxies are widely used to characterize how the galaxy population evolves over cosmic time. If stellar masses can be estimated in a robust manner, free from any bias, global diagnostics such as the stellar mass function can be used to constrain the physics of galaxy formation. We explore how galaxy stellar masses, estimated by fitting broad-band spectral energy distributions (SEDs) with stellar population models, can be biased as a result of commonly adopted assumptions for the star formation and chemical enrichment histories, recycled fractions and dust attenuation curves of galaxies. We apply the observational technique of broad-band SED fitting to model galaxy SEDs calculated by the theoretical galaxy formation model GALFORM, isolating the effect of each of these assumptions. We find that, averaged over the entire galaxy population, the common assumption of exponentially declining star formation histories does not, by itself, adversely affect stellar mass estimation. However, we also show that this result does not hold when considering galaxies that have undergone a recent burst of star formation. We show that fixing the metallicity in SED fitting or using sparsely sampled metallicity grids can introduce mass-dependent systematics into stellar mass estimates. We find that the common assumption of a star–dust geometry corresponding to a uniform foreground dust screen can cause the stellar masses of dusty model galaxies to be significantly underestimated. Finally, we show that stellar mass functions recovered by applying SED fitting to model galaxies at high redshift can differ significantly in both shape and normalization from the intrinsic mass functions predicted by a given model. In particular, the effects of dust can reduce the normalization at the high-mass end by up to 0.6 dex in some cases. Given these differences, our methodology of using stellar masses estimated from model galaxy SEDs offers a new, self-consistent way to compare model predictions with observations. We conclude that great care should be taken when comparing theoretical galaxy formation models to observational results based on the estimated stellar masses of high-redshift galaxies.

A unified mutiwavelength model of galaxy formation.

We present a new version of the GALFORM semi-analytical model of galaxy formation. This brings together several previous developments of GALFORM into a single unified model, including a different initial mass function (IMF) in quiescent star formation and in starbursts, feedback from active galactic nuclei supressing gas cooling in massive halos, and a new empirical star formation law in galaxy disks based on their molecular gas content. In addition, we have updated the cosmology, introduced a more accurate treatment of dynamical friction acting on satellite galaxies, and updated the stellar population model. The new model is able to simultaneously explain both the observed evolution of the K-band luminosity function and stellar mass function, and the number counts and redshift distribution of sub-mm galaxies selected at 850μm. This was not previously achieved by a single physical model within the ΛCDM framework, but requires having an IMF in starbursts that is somewhat top-heavy. The new model is tested against a wide variety of observational data covering wavelengths from the far-UV to sub-mm, and redshifts from z = 0 to z = 6, and is found to be generally successful. These observations include the optical and near-IR luminosity functions, HI mass function, fraction of early type galaxies, Tully-Fisher, metallicity-luminosity and size-luminosity relations at z = 0, as well as far-IR number counts, and far-UV luminosity functions at z ∼ 3 − 6. Discrepancies are however found in galaxy sizes and metallicities at low luminosities, and in the abundance of low mass galaxies at high-z, suggesting the need for a more sophisticated model of supernova feedback.

Constraining the properties of AGN host galaxies with spectral energy distribution modelling

Detailed studies of the spectral energy distribution (SED) of normal galaxies have increasingly been used to understand the physical mechanism dominating their integrated emission, mainly owing to the availability of high quality multi-wavelength data from the UV to the far-infrared (FIR). However, systems hosting dust-enshrouded nuclear starbursts and/or an accreting supermassive black hole (an active galactic nucleus or AGN) are especially challenging to study. This is due to the complex interplay between the heating by massive stars and the AGN, the absorption and emission of radiation from dust, as well as the presence of the underlying old stellar population. We used the latest release of CIGALE, a fast state-of-the-art galaxy SED-fitting model relying on energy balance, to study the influence of an AGN in a self consistent manner in estimating both the star formation rate (SFR) and stellar mass in galaxies, as well as to calculate the contribution of the AGN to the power output of the host. Using the semi-analytical galaxy formation model galform, we created a suite of mock galaxy SEDs using realistic star formation histories (SFH). We also added an AGN of Type-1, Type-2, or intermediate-type whose contribution to the bolometric luminosity can be variable. We performed an SED-fitting of these catalogues with CIGALE, assuming three different SFHs: a single-exponentially-decreasing (1τ-dec), a double-exponentially-decreasing (2τ-dec), and a delayed SFH. Constraining the overall contribution of an AGN to the total infrared luminosity (fracAGN) is very challenging for fracAGN< 20%, with uncertainties of ~5–30% for higher fractions depending on the AGN type, while FIR and sub-mm are essential. The AGN power has an impact on the estimation of M∗ in Type-1 and intermediate-type AGNs but has no effect on galaxies hosting Type-2 AGNs. We find that in the absence of AGN emission, the best estimates of M∗ are obtained using the 2τ-dec model but at the expense of realistic ages of the stellar population. The delayed SFH model provides good estimates of M∗ and SFR, with a maximum offset of 10% as well as better estimates of the age. Our analysis shows that the under-estimation of the SFR increases with fracAGN for Type-1 systems, as well as for low contributions of an intermediate AGN type, but it is quite insensitive to the emission of Type-2 AGNs up to fracAGN ~ 45%. A lack of sampling the FIR, or sub-mm domain systematically over-estimates the SFR (<20%), independent of the contribution of the AGN. Similarly, the UV emission is critical in accurately retrieving both the M∗ for Type-1 and intermediate- type AGN and the SFR of all three AGN types. We show that the presence of AGN emission introduces a scatter to the SFR-M∗ main sequence relation derived from SED-fitting, which is driven by the uncertainties on M∗. Finally, we used our mock catalogues to test the popular IR SED-fitting code DecompIR and show that fracAGN is under-estimated but that the SFR is recovered well for Type-1 and intermediate-types of AGN. The fracAGN, SFR, and LIR estimates of Type-2 AGNs are more problematic owing to a FIR emission disagreement between predicted and observed models.

Galaxies in the EAGLE hydrodynamical simulation and in the Durham and Munich semi-analytical models

We compare global predictions from the eagle hydrodynamical simulation, and two semi-analytic (SA) models of galaxy formation, l-galaxies and galform. All three models include the key physical processes for the formation and evolution of galaxies and their parameters are calibrated against a small number of observables at z ≈ 0. The two SA models have been applied to merger trees constructed from the eagle dark matter only simulation. We find that at z ≤ 2, both the galaxy stellar mass functions for stellar masses M* 109.5 M⊙ differ in some instances by an order of magnitude, while the stellar mass–size relation in eagle is a factor of ≈2 tighter than for the two SA models. Our results suggest the need for a revision of how SA models treat the effect of baryonic self-gravity on the underlying dark matter. The treatment of gas flows in the models needs to be revised based on detailed comparison with observations to understand in particular the evolution of the stellar mass–metallicity relation.

The evolution of the star-forming sequence in hierarchical galaxy formation models

It has been argued that the specific star formation rates of star forming galaxies inferred from observational data decline more rapidly below z = 2 than is predicted by hierarchical galaxy formation models. We present a detailed analysis of this problem by comparing predictions from the GALFORM semi-analytic model with an extensive compilation of data on the average star formation rates of star-forming galaxies. We also use this data to infer the form of the stellar mass assembly histories of star forming galaxies. Our analysis reveals that the currently available data favours a scenario where the stellar mass assembly histories of star forming galaxies rise at early times and then fall towards the present day. In contrast, our model predicts stellar mass assembly histories that are almost flat below z = 2 for star forming galaxies, such that the predicted star formation rates can be offset with respect to the observational data by factors of up to 2 3. This disagreement can be explained by the level of coevolution between stellar and halo mass assembly that exists in contemporary galaxy formation models. In turn, this arises because the standard implementations of star formation and supernova feedback used in the models result in the efficiencies of these process remaining approximately constant over the lifetime of a given star forming galaxy. We demonstrate how a modification to the timescale for gas ejected by feedback to be reincorporated into galaxy haloes can help to reconcile the model predictions with the data.

A new methodology to test galaxy formation models using the dependence of clustering on stellar mass.

We present predictions for the two-point correlation function of galaxy clustering as a function of stellar mass, computed using two new versions of the GALFORM semi-analytic galaxy formation model. These models make use of a high resolution, large volume N-body simulation, set in the 7-year Wilkinson Microwave Anisotropy Probe cosmology. One model uses a universal stellar initial mass function (IMF), while the other assumes different IMFs for quiescent star formation and bursts. Particular consideration is given to how the assumptions required to estimate the stellar masses of observed galaxies (such as the choice of IMF, stellar population synthesis model, and dust extinction) influence the perceived dependence of galaxy clustering on stellar mass. Broad-band spectral energy distribution fitting is carried out to estimate stellar masses for the model galaxies in the same manner as in observational studies. We show clear differences between the clustering signals computed using the true and estimated model stellar masses. As such, we highlight the importance of applying our methodology to compare theoretical models to observations. We introduce an alternative scheme for the calculation of the merger time-scales for satellite galaxies in GALFORM, which takes into account the dark matter subhalo information from the simulation. This reduces the amplitude of small-scale clustering. The new merger scheme offers improved or similar agreement with observational clustering measurements, over the redshift range 0 < z < 0.7. We find reasonable agreement with clustering measurements from the Galaxy and Mass Assembly Survey, but find larger discrepancies for some stellar mass ranges and separation scales with respect to measurements from the Sloan Digital Sky Survey and the VIMOS Public Extragalactic Redshift Survey, depending on the GALFORM model used.

The evolution of the stellar mass versus halo mass relationship

We present an analysis of the predictions made by the GALFORM semi-analytic galaxy formation model for the evolution of the relationship between stellar mass and halo mass. We show that for the standard implementations of supernova feedback and gas reincorporation used in semi-analytic models, this relationship is predicted to evolve weakly over the redshift range 0 < z < 4. Modest evolution in the median stellar mass versus halo mass (SHM) relationship implicitly requires that, at fixed halo mass, the efficiency of stellar mass assembly must be almost constant with cosmic time. We show that in our model, this behaviour can be understood in simple terms as a result of a constant efficiency of gas reincorporation, and an efficiency of SNe feedback that is, on average, constant at fixed halo mass. We present a simple explanation of how feedback from active galactic nuclei (AGN) acts in our model to introduce a break in the SHM relation whose location is predicted to evolve only modestly. Finally, we show that if modifications are introduced into the model such that, for example, the gas reincorporation efficiency is no longer constant, the median SHM relation is predicted to evolve significantly over 0 < z < 4. Specifically, we consider modifications that allow the model to better reproduce either the evolution of the stellar mass function or the evolution of average star formation rates inferred from observations.

The MUSE Hubble Ultra Deep Field Survey. VIII. Extended Lyman-α haloes around high-z star-forming galaxies

We report the detection of extended Ly α haloes around 145 individual star-forming galaxies at redshifts 3 ≤ z ≤ 6 in the Hubble Ultra Deep Field observed with the Multi-Unit Spectroscopic Explorer (MUSE) at ESO-VLT. Our sample consists of continuum-faint (− 15 ≥ M UV ≥ −22) Ly α emitters (LAEs). Using a 2D, two-component (continuum-like and halo) decomposition of Ly α emission assuming circular exponential distributions, we measure scale lengths and luminosities of Ly α haloes. We find that 80% of our objects having reliable Ly α halo measurements show Ly α emission that is significantly more extended than the UV continuum detected by HST (by a factor ≈4 to >20). The median exponential scale length of the Ly α haloes in our sample is ≈4.5 kpc with a few haloes exceeding 10 kpc. By comparing the maximal detected extent of the Ly α emission with the predicted dark matter halo virial radii of simulated galaxies, we show that the detected Ly α emission of our selected sample of Ly α emitters probes a significant portion of the cold circum-galactic medium of these galaxies (>50% in average). This result therefore shows that there must be significant HI reservoirs in the circum-galactic medium and reinforces the idea that Ly α haloes are ubiquitous around high-redshift Ly α emitting galaxies. Our characterization of the Ly α haloes indicates that the majority of the Ly α flux comes from the halo (≈65%) and that their scale lengths seem to be linked to the UV properties of the galaxies (sizes and magnitudes). We do not observe a significant Ly α halo size evolution with redshift, although our sample for z > 5 is very small. We also explore the diversity of the Ly α line profiles in our sample and we find that the Ly α lines cover a large range of full width at half maximum (FWHM) from 118 to 512 km s -1 . While the FWHM does not seem to be correlated to the Ly α scale length, most compact Ly α haloes and those that are not detected with high significance tend to have narrower Ly α profiles ( -1 ). Finally, we investigate the origin of the extended Ly α emission but we conclude that our data do not allow us to disentangle the possible processes, i.e. scattering from star-forming regions, fluorescence, cooling radiation from cold gas accretion, and emission from satellite galaxies.

Comparing galaxy formation in semi-analytic models and hydrodynamical simulations.

It is now possible for hydrodynamical simulations to reproduce a representative galaxy population. Accordingly, it is timely to assess critically some of the assumptions of traditional semi-analytic galaxy formation models. We use the EAGLE simulations to assess assumptions built into the GALFORM semi-analytic model, focusing on those relating to baryon cycling, angular momentum and feedback. We show that the assumption in GALFORM that newly formed stars have the same specific angular momentum as the total disc leads to a significant overestimate of the total stellar specific angular momentum of disc galaxies. In EAGLE, stars form preferentially out of low-specific angular momentum gas in the interstellar medium due to the assumed gas density threshold for stars to form, leading to more realistic galaxy sizes. We find that stellar mass assembly is similar between GALFORM and EAGLE but that the evolution of gas properties is different, with various indications that the rate of baryon cycling in EAGLE is slower than is assumed in GALFORM. Finally, by matching individual galaxies between EAGLE and GALFORM, we find that an artificial dependence of active galactic nucleus feedback and gas infall rates on halo mass-doubling events in GALFORM drives most of the scatter in stellar mass between individual objects. Put together our results suggest that the GALFORM semi-analytic model can be significantly improved in light of recent advances.