Simulating subhaloes at high redshift: merger rates, counts and types

Galaxies are believed to be in one-to-one correspondence with simulated dark matter subhaloes. We use high-resolution N -body simulations of cosmological volumes to calculate the statistical properties of subhalo (galaxy) major mergers at high redshift ( z = 0.6–5). We measure the evolution of the galaxy merger rate, finding that it is much shallower than the merger rate of dark matter host haloes at   z > 2.5  , but roughly parallels that of haloes at   z < 1.6  . We also track the detailed merger histories of individual galaxies and measure the likelihood of multiple mergers per halo or subhalo. We examine satellite merger statistics in detail: 15–35 per cent of all recently merged galaxies are satellites, and satellites are twice as likely as centrals to have had a recent major merger. Finally, we show how the differing evolution of the merger rates of haloes and galaxies leads to the evolution of the average satellite occupation per halo, noting that for a fixed halo mass, the satellite halo occupation peaks at   z ∼ 2.5  .     

The clustering and host haloes of galaxy mergers at high redshift

High-resolution simulations of cosmological structure formation indicate that dark matter substructure in dense environments, such as groups and clusters, may survive for a long time. These dark matter subhaloes are the likely hosts of galaxies. We examine the small-scale spatial clustering of subhalo major mergers at high redshift using high-resolution N -body simulations of cosmological volumes. Recently merged, massive subhaloes exhibit enhanced clustering on scales  ∼100–300  h ⁻¹ kpc  , relative to all subhaloes of the same infall mass, for a short time after a major merger (<500 Myr). The small-scale clustering enhancement is smaller for lower mass subhaloes, which also show a deficit on scales just beyond the excess. Haloes hosting recent subhalo mergers tend to have more subhaloes; for massive subhaloes, the excess is stronger and it tends to increase for the most massive host haloes. The subhalo merger fraction is independent of halo mass for the scales we probe. In terms of satellite and central subhaloes, the merger increase in small-scale clustering for massive subhaloes arises from recently merged massive central subhaloes having an enhanced satellite population. Our mergers are defined via their parent infall mass ratios. Subhaloes experiencing major mass gains also exhibit a small-scale clustering enhancement, but these correspond to two-body interactions leading to two final subhaloes, rather than subhalo coalescence.     

The concentration–velocity dispersion relation in galaxy groups

Based on results from cold dark matter N -body simulations, we develop a dynamical model for the evolution of subhaloes within group-sized host haloes. Only subhaloes more massive than 5 × 10⁸ M_⊙ are considered, because they are massive enough to possibly host luminous galaxies. On their orbits within a growing host potential the subhaloes are subject to tidal stripping and dynamical friction. At the present time  ( z = 0)  , all model hosts have equal mass  ( M _vir= 3.9 × 10¹³ M_⊙)  but different concentrations associated with different formation times. We investigate the variation of subhalo (or satellite galaxy) velocity dispersion with host concentration and/or formation time. In agreement with the Jeans equation, the velocity dispersion of subhaloes increases with the host concentration. Between concentrations of ∼5 and ∼20, the subhalo velocity dispersions increase by a factor of ∼1.25. By applying a simplified tidal disruption criterion, that is, rejection of all subhaloes with a tidal truncation radius below 3  kpc at   z = 0  , the central velocity dispersion of the 'surviving' subhalo sample increases substantially for all concentrations. The enhanced central velocity dispersions in the surviving subhalo samples are caused by a lack of slow tangential motions. Additionally, we present a fitting formula for the anisotropy parameter which does not depend on concentration if the group-centric distances are scaled by r _s, the characteristic radius of the Navarro, Frenk & White profile. Since the expected loss of subhaloes and galaxies due to tidal disruption increases the velocity dispersion of surviving galaxies, the observed galaxy velocity dispersion can substantially overestimate the virial mass.     

The fate of substructures in cold dark matter haloes

We use the Millennium Simulation, a large, high-resolution N -body simulation of the evolution of structure in a Λ cold dark matter cosmology, to study the properties and fate of substructures within a large sample of dark matter haloes. We find that the subhalo mass function departs significantly from a power law at the high-mass end. We also find that the radial and angular distributions of substructures depend on subhalo mass. In particular, high-mass subhaloes tend to be less radially concentrated and to have angular distributions closer to the direction perpendicular to the spin of the host halo than their less massive counterparts. We find that mergers between subhaloes occur. These tend to be between substructures that were already dynamically associated before accretion into the main halo. For subhaloes larger than 0.001 times the mass of the host halo, it is more likely that the subhalo will merge with the central or main subhalo than with another subhalo larger than itself. For lower masses, subhalo–subhalo mergers become equally likely to mergers with the main subhalo. Our results have implications for the variation of galaxy properties with environment and for the treatment of mergers in galaxy formation models.     

Analytical approach to subhalo population in dark matter haloes

In the standard model of cosmic structure formation, dark matter haloes form by gravitational instability. The process is hierarchical: smaller systems collapse earlier, and later merge to form larger haloes. The galaxy clusters, hosted by the largest dark matter haloes, are at the top of this hierarchy and representing the largest as well as the last structures formed in the Universe, while the smaller and first haloes are those Earth-sized dark subhaloes that have been both predicted by theoretical considerations and found in numerical simulations, though there do not exist any observational hints of their existence. The probability that a halo of mass m at redshift z will be part of a larger halo of mass M at the present time can be described in the frame of the extended Press & Schecter theory making use of the progenitor (conditional) mass function. Using the progenitor mass function, we calculate analytically, at redshift zero, the distribution of subhaloes in mass, formation epoch and rarity of the peak of the density field at the formation epoch. That is done for a Milky Way size system, assuming both a spherical and an ellipsoidal collapse model. Our calculation assumes that small progenitors do not lose mass due to dynamical processes after entering the parent halo, and that they do not interact with other subhaloes. For a Λ cold dark matter power spectrum, we obtain a subhalo mass function  d n /d m   proportional to   m ^−α  with a model-independent  α∼ 2  . Assuming that the dark matter is a weakly interacting massive particle, the inferred distributions are used to test the feasibility of an indirect detection in the γ-ray energy band of such a population of subhaloes with a Gamma-ray Large Area Space Telescope like satellite.     

The distribution of ejected subhaloes and its implication for halo assembly bias

Using a high-resolution cosmological N -body simulation, we identify the ejected population of subhaloes, which are haloes at redshift   z = 0  but were once contained in more massive 'host' haloes at high redshifts. The fraction of the ejected subhaloes in the total halo population of the same mass ranges from 9 to 4 per cent for halo masses from  ∼10¹¹  to  ∼10¹²  h ⁻¹ M_⊙  . Most of the ejected subhaloes are distributed within four times the virial radius of their hosts. These ejected subhaloes have distinct velocity distribution around their hosts in comparison to normal haloes. The number of subhaloes ejected from a host of given mass increases with the assembly redshift of the host. Ejected subhaloes in general reside in high-density regions, and have a much higher bias parameter than normal haloes of the same mass. They also have earlier assembly times, so that they contribute to the assembly bias of dark matter haloes seen in cosmological simulations. However, the assembly bias is not dominated by the ejected population, indicating that large-scale environmental effects on normal haloes are the main source for the assembly bias.     

Intrinsic correlation of galaxy shapes: implications for weak lensing measurements

Weak gravitational lensing is now established as a powerful method to measure mass fluctuations in the universe. It relies on the measurement of small coherent distortions of the images of background galaxies. Even low-level correlations in the intrinsic shapes of galaxies could however produce a significant spurious lensing signal. These correlations are also interesting in their own right, since their detection would constrain models of galaxy formation. Using     haloes found in N -body simulations, we compute the correlation functions of the intrinsic ellipticity of spiral galaxies assuming that the disc is perpendicular to the angular momentum of the dark matter halo. We also consider a simple model for elliptical galaxies, in which the shape of the dark matter halo is assumed to be the same as that of the light. For deep lensing surveys with median redshifts ∼1, we find that intrinsic correlations of ∼10⁻⁴ on angular scales     are generally below the expected lensing signal, and contribute only a small fraction of the excess signals reported on these scales. On larger scales we find limits to the intrinsic correlation function at a level ∼10⁻⁵, which gives a (model-dependent) range of separations for which the intrinsic signal is about an order of magnitude below the ellipticity correlation function expected from weak lensing. Intrinsic correlations are thus negligible on these scales for dedicated weak lensing surveys. For wider but shallower surveys such as SuperCOSMOS, APM and SDSS, we cannot exclude the possibility that intrinsic correlations could dominate the lensing signal. We discuss how such surveys could be used to calibrate the importance of this effect, as well as study spin–spin correlations of spiral galaxies.     

The JVAS/CLASS search for 6-arcsec to 15-arcsec image separation lensing

The Jodrell Bank–VLA Astrometric Survey (JVAS) and the Cosmic Lens All Sky Survey (CLASS) have been systematically searched for multiple gravitational imaging of sources with image separations between 6 arcsec and 15 arcsec, associated with galaxy group and cluster lensing masses. The radio and optical follow-up observations of all candidates are presented. From a total of ∼15 000 sources only one weak candidate remains and this is not contained in the statistically complete sample of flat-spectrum JVAS/CLASS sources of 11 670 sources. A simple Press–Schechter analysis is performed. For singular isothermal sphere lenses the lack of multiple image systems is inconsistent with the currently favoured cosmologies with     at the 4.2 σ level. Cored isothermal lenses reduce the expected number of lens systems and we suggest that the most probable interpretation of our results is that the surface mass density of groups and clusters of galaxies is not high enough to cause multiple imaging and the presence of the mass concentrations associated with individual galaxies is required to produce image separations such as those in B0957+561.     

Two new large-separation gravitational lenses from SDSS

We present discovery images, together with follow-up imaging and spectroscopy, of two large-separation gravitational lenses found by our survey for wide arcs [the CAmbridge Sloan Survey Of Wide ARcs in the skY (CASSOWARY)]. The survey exploits the multicolour photometry of the Sloan Digital Sky Survey to find multiple blue components around red galaxies. CASSOWARY 2 (or 'the Cheshire Cat') is composed of two massive early-type galaxies at   z = 0.426  and 0.432, respectively, lensing two background sources, the first a star-forming galaxy at   z = 0.97  and the second a high -redshift galaxy  ( z > 1.4)  . There are at least three images of the former source and probably four or more of the latter, arranged in two giant arcs. The mass enclosed within the larger arc of radius ∼11 arcsec is  ∼33 × 10¹² M_⊙  . CASSOWARY 3 comprises an arc of three bright images of a   z = 0.725  source, lensed by a foreground elliptical at   z = 0.274  . The radius of the arc is ∼4 arcsec and the enclosed mass is  ∼2.5 × 10¹² M_⊙  . Together with earlier discoveries like the Cosmic Horseshoe and the 8 o'clock Arc, these new systems, with separations intermediate between the arcsecond-separation lenses of typical strong galaxy lensing and arcminute-separation cluster lenses, probe the very high end of the galaxy mass function.     

The dynamics of subhaloes in warm dark matter models

We present a comparison of the properties of substructure haloes ( subhaloes ) orbiting within host haloes that form in cold dark matter (CDM) and warm dark matter (WDM) cosmologies. Our study focuses on selected properties of these subhaloes, namely their anisotropic spatial distribution within the hosts; the existence of a 'backsplash' population; the age–distance relation; the degree to which they suffer mass loss; and the distribution of relative (infall) velocities with respect to the hosts. We find that the number density of subhaloes in our WDM model is suppressed relative to that in the CDM model, as we would expect. Interestingly, our analysis reveals that backsplash subhaloes exist in both the WDM and CDM models. Indeed, there are no statistically significant differences between the spatial distributions of subhaloes in the CDM and WDM models. There is evidence that subhaloes in the WDM model suffer enhanced mass loss relative to their counterparts in the CDM model, reflecting their lower central densities. We note also a tendency for the (infall) velocities of subhaloes in the WDM model to be higher than in the CDM model. Nevertheless, we conclude that observational tests based on either the spatial distribution or the kinematics of the subhalo population are unlikely to help us to differentiate between the CDM model and our adopted WDM model.     

The growth of central and satellite galaxies in cosmological smoothed particle hydrodynamics simulations

We examine the accretion and merger histories of central and satellite galaxies in a smoothed particle hydrodynamics (SPH) cosmological simulation that resolves galaxies down to  7 × 10⁹ M_⊙  . Most friends-of-friends haloes in the simulation have a distinct central galaxy, typically 2–5 times more massive than the most massive satellite. As expected, satellites have systematically higher assembly redshifts than central galaxies of the same baryonic mass, and satellites in more massive haloes form earlier. However, contrary to the simplest expectations, satellite galaxies continue to accrete gas and convert it to stars; the gas accretion declines steadily over a period of 0.5–1 Gyr after the satellite halo merges with a larger parent halo. Satellites in a cluster mass halo eventually begin to lose baryonic mass. Typically, satellites in our simulation are 0.1–0.2 mag bluer than in models that assume no gas accretion on to satellites after a halo merger. Since   z = 1  , 27 per cent of central galaxies (above  3 × 10¹⁰ M_⊙  ) and 22 per cent of present-day satellite galaxies have merged with a smaller system above a 1:4 mass ratio; about half of the satellite mergers occurred after the galaxy became a satellite and half before. In effect, satellite galaxies can remain 'central' objects of halo substructures, with continuing accretion and mergers, making the transition in assembly histories and physical properties a gradual one. Implementing such a gradual transformation in semi-analytic models would improve their agreement with observed colour distributions of satellite galaxies in groups and with the observed colour dependence of galaxy clustering.     

Testing tidal-torque theory – I. Spin amplitude and direction

We evaluate the success of linear tidal-torque theory (TTT) in predicting galactic-halo spin using a cosmological N -body simulation with thousands of well-resolved haloes. The protohaloes are identified by tracing today's haloes back to the initial conditions. The TTT predictions for the protohaloes match, on average, the spin amplitudes of the virialized haloes of today, if linear growth is assumed until ∼ t ₀/3, or  55–70  per cent of the halo effective turn-around time. This makes it a useful qualitative tool for understanding certain average properties of galaxies, such as total spin and angular momentum distribution within haloes, but with a random scatter of the order of the signal itself. Non-linear changes in spin direction cause a mean error of ∼50° in the TTT prediction at t ₀, such that the linear spatial correlations of spins on scales ≥1  h ⁻¹ Mpc are significantly weakened by non-linear effects. This questions the usefulness of TTT for predicting intrinsic alignments in the context of gravitational lensing. We find that the standard approximations made in TTT, including a second-order expansion of the Zel'dovich potential and a smoothing of the tidal field, provide close-to-optimal results.     

Dynamical friction and galaxy merging time-scales

The time-scale for galaxies within merging dark matter haloes to merge with each other is an important ingredient in galaxy formation models. Accurate estimates of merging time-scales are required for predictions of astrophysical quantities such as black hole binary merger rates, the build-up of stellar mass in central galaxies and the statistical properties of satellite galaxies within dark matter haloes. In this paper, we study the merging time-scales of extended dark matter haloes using N -body simulations. We compare these results to standard estimates based on the Chandrasekhar theory of dynamical friction. We find that these standard predictions for merging time-scales, which are often used in semi-analytic galaxy formation models, are systematically shorter than those found in simulations. The discrepancy is approximately a factor of 1.7 for M _sat/ M _host≈ 0.1 and becomes larger for more disparate satellite-to-host mass ratios, reaching a factor of ∼3.3 for M _sat/ M _host≈ 0.01. Based on our simulations, we propose a new, easily implementable fitting formula that accurately predicts the time-scale for an extended satellite to sink from the virial radius of a host halo down to the halo's centre for a wide range of M _sat/ M _host and orbits. Including a central bulge in each galaxy changes the merging time-scale by ≲10 per cent. To highlight one concrete application of our results, we show that merging time-scales often used in the literature overestimate the growth of stellar mass by satellite accretion by ≈40 per cent, with the extra mass gained in low mass ratio mergers.     

Dark matter subhaloes in numerical simulations

We use cosmological Λ cold dark matter (CDM) numerical simulations to model the evolution of the substructure population in 16 dark matter haloes with resolutions of up to seven million particles within the virial radius. The combined substructure circular velocity distribution function (VDF) for hosts of 10¹¹ to  10¹⁴ M_⊙  at redshifts from zero to two or higher has a self-similar shape, is independent of host halo mass and redshift, and follows the relation  d n /d v = (1/8)( v _cmax/ v _cmax,host)⁻⁴  . Halo to halo variance in the VDF is a factor of roughly 2 to 4. At high redshifts, we find preliminary evidence for fewer large substructure haloes (subhaloes). Specific angular momenta are significantly lower for subhaloes nearer the host halo centre where tidal stripping is more effective. The radial distribution of subhaloes is marginally consistent with the mass profile for   r ≳ 0.3 r _vir  , where the possibility of artificial numerical disruption of subhaloes can be most reliably excluded by our convergence study, although a subhalo distribution that is shallower than the mass profile is favoured. Subhalo masses but not circular velocities decrease towards the host centre. Subhalo velocity dispersions hint at a positive velocity bias at small radii. There is a weak bias towards more circular orbits at lower redshift, especially at small radii. We additionally model a cluster in several power-law cosmologies of   P ∝ kⁿ   , and demonstrate that a steeper spectral index, n , results in significantly less substructure.     

Halo masses for optically selected and for radio-loud AGN from clustering and galaxy–galaxy lensing

We compute two-point correlation functions and measure the shear signal due to galaxy–galaxy lensing for 80 000 optically identified and 5700 radio-loud active galactic nuclei (AGN) from Data Release 4 of the Sloan Digital Sky Survey. Halo occupation models are used to estimate halo masses and satellite fractions for these two types of AGN. The large sample size allows us to separate AGN according to the stellar mass of their host galaxies. We study how the halo masses of optical and radio AGN differ from those of the parent population at fixed   M _*  . Halo masses deduced from clustering and from lensing agree satisfactorily. Radio AGN are found in more massive haloes than optical AGN: in our samples, their mean halo masses are  1.6 × 10¹³  and  8 × 10¹¹  h ⁻¹ M_⊙  , respectively. Optical AGN follow the same relation between stellar mass and halo mass as galaxies selected without regard to nuclear properties, but radio-loud AGN deviate significantly from this relation. The dark matter haloes of radio-loud AGN are about twice as massive as those of control galaxies of the same stellar mass. This boost is independent of radio luminosity, and persists even when our analysis is restricted to field galaxies. The large-scale gaseous environment of the galaxy clearly plays a crucial role in producing observable radio emission. The dark matter halo masses that we derive for the AGN in our two samples are in good agreement with recent models in which feedback from radio AGN becomes dominant in haloes where gas cools quasi-statically.     

Global environmental effects versus galaxy interactions

We explore properties of close galaxy pairs and merging systems selected from the Sloan Digital Sky Survey Data Release 4 in different environments with the aim to assess the relative importance of the role of interactions over global environmental processes. For this purpose, we perform a comparative study of galaxies with and without close companions as a function of local density and host halo mass, carefully removing sources of possible biases. We find that at low- and high-local-density environments, colours and concentration indices of close galaxy pairs are very similar to those of isolated galaxies. At intermediate densities, we detect significant differences, indicating that close pairs could have experienced a more rapid transition on to the red sequence than isolated galaxies. The presence of a correlation between concentration index and colours indicates that the physical mechanism responsible for the colour transformation also operates in the transformation of the luminous matter distribution. At fixed local densities, we find a dependence of the red galaxy fraction on dark matter halo mass for galaxies with or without a close companion. This suggests the action of host halo mass related effects. Regardless of dark matter halo mass, we show that the percentage of red galaxies in close pairs and in the control sample are comparable at low- and high-local-density environments. However, at intermediate local densities, the gap in the red fraction between close pairs and the control galaxies increases from ∼10 per cent in low-mass haloes up to ∼50 per cent in the most massive ones. Interestingly, we also detect that 50 per cent of merging systems populate the intermediate local environments, with a large fraction of them being extremely red and bulge dominated. Our findings suggest that in intermediate-density environments galaxies are efficiently pre-processed by close encounters and mergers before entering higher local density regions.     

Probing galactic dark matter in dense environments: on the strong lensing efficiency of galaxies in rich clusters

The recent detection by Limousin et al. of five new strong lensing events dominated by galaxy cluster members in Abell 1689, and outside the critical regime of the cluster itself, offers a way to obtain constraints on the cluster mass distribution in a region inaccessible to standard lensing analysis. In addition, modelling such systems will provide another window on the dark matter haloes of galaxies in very dense environments. Here, it is shown that the boost in image separation due to the external shear and convergence from a smooth cluster component means that more numerous, less massive galaxies have the potential to create multiple images with detectable separations, relative to isolated field galaxies. This comes in addition to a potential increase in their lensing (source plane) cross-section. To gain insight into the factors involved and as a precursor to a numerical study using N -body simulations, a simple analytic model of a cluster at   z = 0.3  lensing background galaxies at   z = 2  is considered here. The fiducial model has cluster members with isothermal density profiles and luminosities L , distributed in a Schechter function (faint-end slope  ν=−1.25  ), related to their velocity dispersions σ via the Faber–Jackson scaling L ∝σ⁴. Just outside the critical regime of the cluster, the scale of galaxy-dominated image separations is significantly increased. Folding in the fact that less massive galaxies present a lower lensing cross-section, and that the cross-section can itself be enhanced in an external field leads to a factor of a few times more detected events relative to field galaxies. These values will be higher closer to the critical curve. Given that the events in Abell 1689 were detected over a very small region of the cluster where ACS data were available, this motivates the search for such events in other clusters.     

Galactic halo cusp–core: tidal compression in mergers

We explain in simple terms how the build-up of dark haloes by merging compact satellites, as in the cold dark matter (CDM) cosmology, inevitably leads to an inner cusp of density profile  ρ∝ r ^−α  with  α≳ 1  , as seen in cosmological N -body simulations. A flatter halo core with  α < 1  exerts on the satellites tidal compression in all directions, which prevents the deposit of stripped satellite material in the core region. This makes the satellite orbits decay from the radius where  α∼ 1  to the halo centre with no local tidal mass transfer, and thus causes a rapid steepening of the inner profile to  α > 1  . These tidal effects, the resultant steepening of the profile to a cusp, and the stability of this cusp to tandem mergers with compact satellites are demonstrated using N -body simulations. The transition at  α∼ 1  is then addressed using toy models in the limiting cases of impulse and adiabatic approximations and using tidal radii for satellites on radial and circular orbits. In an associated paper, we address the subsequent slow convergence from either side to an asymptotic stable cusp with  α≳ 1  . Our analysis thus implies that an inner cusp is enforced when small haloes are typically more compact than larger haloes, as in the CDM scenario, such that enough satellite material makes it intact into the inner halo and is deposited there. We conclude that a necessary condition for maintaining a flat core, as indicated by observations, is that the inner regions of the CDM satellite haloes be puffed up by about 50 per cent such that when they merge into a larger halo they would be disrupted outside the halo core. This puffing up could be due to baryonic feedback processes in small haloes, which may be stimulated by the tidal compression in the halo cores.     

Effects of cluster galaxies on arc statistics

We present the results of a set of numerical simulations evaluating the effect of cluster galaxies on arc statistics.
We perform a first set of gravitational lensing simulations using three independent projections for each of nine different galaxy clusters obtained from N -body simulations. The simulated clusters consist of dark matter only. We add a population of galaxies to each cluster, mimicking the observed luminosity function and the spatial galaxy distribution, and repeat the lensing simulations including the effects of cluster galaxies, which themselves act as individual lenses. Each galaxy is represented by a spherical Navarro, Frenk & White density profile.
We consider the statistical distributions of the properties of the gravitational arcs produced by our clusters with and without galaxies. We find that the cluster galaxies do not introduce perturbations strong enough to significantly change the number of arcs and the distributions of lengths, widths, curvature radii and length-to-width ratios of long arcs. We find some changes to the distribution of short-arc properties in the presence of cluster galaxies. The differences appear in the distribution of curvature radii for arc lengths smaller than 12 arcsec, while the distributions of lengths, widths and length-to-width ratios are significantly changed only for arcs shorter than 4 arcsec.     

Strong-lensing optical depths in a ΛCDM universe – II. The influence of the stellar mass in galaxies

We investigate how strong gravitational lensing in the concordance ΛCDM cosmology is affected by the stellar mass in galaxies. We extend our previous studies, based on ray tracing through the Millennium Simulation, by including the stellar components predicted by galaxy formation models. We find that the inclusion of these components greatly enhances the probability for strong lensing compared to a 'dark matter only' universe. The identification of the 'lenses' associated with strong-lensing events reveals that the stellar mass of galaxies (i) significantly enhances the strong-lensing cross-sections of group and cluster haloes and (ii) gives rise to strong lensing in smaller haloes, which would not produce noticeable effects in the absence of the stars. Even if we consider only image splittings ≳10 arcsec, the luminous matter can enhance the strong-lensing optical depths by up to a factor of 2.