In the absence of H 2 molecules, the primordial gas in early dark matter haloes with virial temperatures just above T vir 10 4 K cools by collisional excitation of atomic H. Although it cools efficiently, this gas remains relatively hot, at a temperature near T ∼ 8000 K, and consequently might be able to avoid fragmentation and collapse directly into a supermassive black hole. In order for H 2 formation and cooling to be strongly suppressed, the gas must be irradiated by a sufficiently intense ultraviolet (UV) flux. We performed a suite of threedimensional hydrodynamical adaptive mesh refinement (AMR) simulations of gas collapse in three different protogalactic haloes with T vir 10 4 K, irradiated by a UV flux with various intensities and spectra. We determined the critical specific intensity, J crit 21 , required to suppress H 2 cooling in each of the three haloes. For a hard spectrum representative of metal-free stars, we find (in units of 10 −21 erg s −1 Hz −1 sr −1 cm −2 ) 10 4 < J crit 21 < 10 5 , while for a softer spectrum, which is characteristic of a normal stellar population, and for which H − dissociation is important, we find 30 < J crit 21 < 300. These values are a factor of 3-10 lower than previous estimates. We attribute the difference to the higher, more accurate H 2 collisional dissociation rate we adopted. The reduction in J crit 21 exponentially increases the number of rare haloes exposed to supercritical radiation. When H 2 cooling is suppressed, gas collapse starts with a delay, but it ultimately proceeds more rapidly. The infall velocity is near the increased sound speed, and an object as massive as M ∼ 10 5 M may form at the centre of these haloes, compared to the M ∼ 10 2 M stars forming when H 2 cooling is efficient.

The power spectrum of cosmic infrared background (CIB) anisotropies is sensitive to the connection between star formation and dark matter haloes over the entire cosmic star formation history. Here we develop a model that associates star‐forming galaxies with dark matter haloes and their subhaloes. The model is based on a parametrized relation between the dust‐processed infrared luminosity and (sub)halo mass. By adjusting three free parameters, we attempt to simultaneously fit the four frequency bands of the Planck measurement of the CIB anisotropy power spectrum. To fit the data, we find that the star formation efficiency must peak on a halo mass scale of ≈5 × 1012 M⊙ and the infrared luminosity per unit mass must increase rapidly with redshift. By comparing our predictions with a well‐calibrated phenomenological model for shot noise, and with a direct observation of source counts, we show that the mean duty cycle of the underlying infrared sources must be near unity, indicating that the CIB is dominated by long‐lived quiescent star formation, rather than intermittent short ‘starbursts’. Despite the improved flexibility of our model, the best simultaneous fit to all four Planck channels remains relatively poor. We discuss possible further extensions to alleviate the remaining tension with the data. Our model presents a theoretical framework for a future joint analysis of both background anisotropy and source count measurements.

Accurate mass determination of clusters of galaxies is crucial if they are to be used as cosmological probes. However, there are some discrepancies between cluster masses determined based on gravitational lensing and X-ray observations assuming strict hydrostatic equilibrium (i.e., the equilibrium gas pressure is provided entirely by thermal pressure). Cosmological simulations suggest that turbulent gas motions remaining from hierarchical structure formation may provide a significant contribution to the equilibrium pressure in clusters. We analyze a sample of massive clusters of galaxies drawn from high-resolution cosmological simulations and find a significant contribution (20%-45%) from non-thermal pressure near the center of relaxed clusters, and, in accord with previous studies, a minimum contribution at about 0.1 R vir , growing to about 30%-45% at the virial radius, R vir. Our results strongly suggest that relaxed clusters should have significant non-thermal support in their core region. As an example, we test the validity of strict hydrostatic equilibrium in the well-studied massive galaxy cluster A1689 using the latest high-resolution gravitational lensing and X-ray observations. We find a contribution of about 40% from non-thermal pressure within the core region of A1689, suggesting an alternate explanation for the mass discrepancy: the strict hydrostatic equilibrium is not valid in this region.

Energetic feedback processes during the formation of galaxy clusters may have heated and ionized the majority of the intergalactic gas in protocluster regions. When such a highly ionized superbubble falls along the sight line to a background quasar, it would be seen as a large void with little or no absorption in the Ly forest. We examine the spectra of 137 quasars in the Sloan Digital Sky Survey to search for such voids and find no clear evidence of their existence. The size distribution of voids in the range 5 8 P Ák P 70 8 (corresponding to physical sizes of 3 h À1 P L P 35 h À1 comoving Mpc) is consistent with the standard model for the Ly forest without additional ionized bubbles. We adapt a physical model for H ii bubble growth during cosmological reionization to describe the expected size distribution of ionized superbubbles at z $ 3. This model incorporates the conjoining of bubbles around individual neighboring galaxies. Using the nondetection of voids, we find that models in which the volume filling factor of ionized bubbles exceeds $20% at z $ 3 can be ruled out, primarily because they overproduce the number of large (40Y50 8) voids. We conclude that any preheating mechanism that explains galaxy cluster observations must avoid heating the low-density gas in the protocluster regions, either by operating relatively recently (z P3) or by increasing entropy primarily in high-density regions. Subject headingg s: intergalactic medium -large-scale structure of universe -methods: data analysisquasars: absorption lines

The gravitational waves (GWs) emitted by inspiralling binary black holes, expected to be detected by the Laser Interferometer Space Antenna (LISA), could be used to determine the luminosity distance to these sources with the unprecedented precision of 1 per cent. We study cosmological parameter constraints from such standard sirens, in the presence of gravitational lensing by large-scale structure. Lensing introduces magnification with a probability distribution function (PDF) whose shape has significant skewness and kurtosis, and depends on cosmological parameters. We use Monte Carlo simulations to generate mock samples of standard sirens, including a small intrinsic scatter, as well as the additional, larger scatter from lensing, in their inferred distances. We derive constraints on cosmological parameters, by simultaneously fitting the mean and the distribution of the residuals on the distance versus redshift (d L -z) Hubble diagram. We find that for standard sirens at redshift z ≈ 1, the sensitivity to a single cosmological parameter, such as the matter density m , or the dark energy equation of state w, is ∼50-80 per cent tighter when the lensing PDF is used, compared to the sensitivity derived from a Gaussian PDF with the same variance. When these two parameters are constrained simultaneously, the non-Gaussian shape yields a further enhanced improvement (by ∼120 per cent), owing to the correlation between the parameters. The sensitivity to the amplitude of the matter power spectrum, σ 8 from the cosmological dependence of the PDF alone, however, is ∼20 per cent worse than that from the Gaussian PDF. The improvements for m and w arise purely from the non-Gaussian shape of the lensing PDF and can be attributed specifically to the sharpness of the peak of this PDF (i.e. to a finite kurtosis); the dependence of the PDF on these parameters does not improve constraints relative to those available from the mean d L -z relation. At higher redshifts, the PDF resembles a Gaussian more closely, and the effects of the non-Gaussianities become less prominent. These results highlight the importance of obtaining an accurate and reliable PDF of the lensing convergence, in order to realize the full potential of standard sirens as cosmological probes.