arXiv subjects
Publications and source records attributed to A. Dekel.
Using high resolution N-body simulations with hydrodynamics and star formation, we investigate interactions and the resulting starbursts in galaxies with properties typical of $z\sim 3$. We apply spectral population models to produce mock-HST images, and discuss the observed magnitude, color, and morphological appearance of our simulated galaxies in both the rest-UV and rest-visual bands.
We present an improved POTENT method for reconstructing the velocity and mass density fields from radial peculiar velocities, test it with mock catalogs, and apply it to the Mark III Catalog. Method improvments: (a) inhomogeneous Malmquist bias is reduced by grouping and corrected in forward or inverse analyses of inferred distances, (b) the smoothing into a radial velocity field is optimized to reduce window and sampling biases, (c) the density is derived from the velocity using an improved nonlinear approximation, and (d) the computational errors are made negligible. The method is tested and optimized using mock catalogs based on an N-body simulation that mimics our cosmological neighborhood, and the remaining errors are evaluated quantitatively. The Mark III catalog, with ~3300 grouped galaxies, allows a reliable reconstruction with fixed Gaussian smoothing of 10-12 Mpc/h out to ~60 Mpc/h. We present maps of the 3D velocity and mass-density fields and the corresponding errors. The typical systematic and random errors in the density fluctuations inside 40 Mpc/h are \pm 0.13 and \pm 0.18. The recovered mass distribution resembles in its gross features the galaxy distribution in redshift surveys and the mass distribution in a similar POTENT analysis of a complementary velocity catalog (SFI), including the Great Attractor, Perseus-Pisces, and the void in between. The reconstruction inside ~40 Mpc/h is not affected much by a revised calibration of the distance indicators (VM2, tailored to match the velocities from the IRAS 1.2Jy redshift survey). The bulk velocity within the sphere of radius 50 Mpc/h about the Local Group is V_50=370 \pm 110 km/s (including systematic errors), and is shown to be mostly generated by external mass fluctuations. With the VM2 calibration, V_50 is reduced to 305 \pm 110 km/s.
The galaxy density field as extracted from the IRAS 1.2 Jy redshift survey is compared to the mass density field as reconstructed by the POTENT method from the Mark III catalog of peculiar velocities. The reconstruction is done with Gaussian smoothing of radius 12 h^{-1}Mpc, and the comparison is carried out within volumes of effective radii 31-46 h^{-1}Mpc, containing approximately 10-26 independent samples. Random and systematic errors are estimated from multiple realizations of mock catalogs drawn from a simulation that mimics the observed density field in the local universe. The relationship between the two density fields is found to be consistent with gravitational instability theory in the mildly nonlinear regime and a linear biasing relation between galaxies and mass. We measure beta = Omega^{0.6}/b_I = 0.89 \pm 0.12 within a volume of effective radius 40 h^{-1}Mpc, where b_I is the IRAS galaxy biasing parameter at 12 h^{-1}Mpc. This result is only weakly dependent on the comparison volume, suggesting that cosmic scatter is no greater than \pm 0.1. These data are thus consistent with Omega=1 and b_I\approx 1. If b_I>0.75, as theoretical models of biasing indicate, then Omega>0.33 at 95% confidence. A comparison with other estimates of beta suggests scale-dependence in the biasing relation for IRAS galaxies.
The power spectrum (PS) of mass density fluctuations, independent of `biasing', is estimated from the Mark III catalog of peculiar velocities using Bayesian statistics. A parametric model is assumed for the PS, and the free parameters are determined by maximizing the probability of the model given the data. The method has been tested using detailed mock catalogs. It has been applied to generalized CDM models with and without COBE normalization. The robust result for all the models is a relatively high PS, with $P(k) Ω^{1.2} = (4.8 \pm 1.5) \times 10^3 (Mpc/h)^3$ at $k=0.1 h/Mpc$. An extrapolation to smaller scales using the different CDM models yields $σ_8 Ω^{0.6} = 0.88 \pm 0.15$. The peak is weakly constrained to the range $0.02 \leq k \leq 0.06 h/Mpc$. These results are consistent with a direct computation of the PS (Kolatt & Dekel 1996). When compared to galaxy-density surveys, the implied values for $β$ ($\equiv Ω^{0.6}/b$) are of order unity to within 25%. The parameters of the COBE-normalized, flat CDM model are confined by a 90% likelihood contour of the sort $Ωh_{50}^μn^ν= 0.8 \pm 0.2$, where $μ= 1.3$ and $ν= 3.4, 2.0$ for models with and without tensor fluctuations respectively. For open CDM the powers are $μ= 0.95$ and $ν= 1.4$ (no tensor fluctuations). A $Γ$-shape model free of COBE normalization yields only a weak constraint: $Γ= 0.4 \pm 0.2$.
The mass density field as extracted from peculiar velocities in our cosmological neighborhood is mapped back in time to the CMB in two ways. First, the density power spectrum ($P_k$) is translated into a temperature angular power spectrum of sub-degree resolution ($C_l$) and compared to observations. Second, the local density field is translated into a temperature map in a patch on the last-scattering surface of a distant observer. A likelihood analysis of the Mark III peculiar velocity data have constrained the range of parameters for $P_k$ within the family of COBE-normalized CDM models (Zaroubi et al 1996), favoring a slight tilt in the initial spectrum, $n<1$. The corresponding range of $C_l$'s is plotted against current observations, indicating that the CMB data can tighten the constraints further: only models with ``small'' tilt ($n\sim 0.9$) and ``high'' baryonic content ($Ω_b \sim 0.1$) could survive the two data sets simultaneously. The local mass density field that has been recovered from the velocities via a Wiener method is convolved with a Boltzmann calculation to recover $10'$ resolution temperature maps as viewed from different directions. The extent of the CMB patch and the amplitude of fluctuations depend on the choice of cosmological parameters, e.g., the local $100\hmpc$ sphere corresponds to $90'$ to $30'$ at the CMB for $Ω$ between 1 and 0 respectively. The phases of the temperature map are correlated with those of the density field, contrary to the contribution of the Sachs-Wolfe effect alone. This correlation suggests the possibility of an inverse reconstruction of the underlying density field from CMB data with interesting theoretical implications.
We compare the predicted Abell/ACO cluster density and velocity fields, using linear theory, with that based on the MarkIII catalog and the POTENT machinery within 7000 km/sec. Both density & velocity fields are in good qualitative agreement indicating that apart from a linear biasing factor the optical Abell/ACO clusters do trace the underlying mass distribution quite well.
The power spectrum (PS) of {\it mass} density fluctuations, independent of ``biasing", is estimated from the Mark3 Catalog of Peculiar Velocities of galaxies using Bayesian statistics. A parametric model is assumed for the PS, and the free parameters are determined by maximizing the posterior probability of the model given the data. The method has been tested using detailed mock catalogs. We use generalized CDM models with and without COBE normalization. The robust result for all the models tested is $P(k) Ω^{1.2} = (4.1 \pm 0.7)x10^3 (Mpc h^{-1})^3$ at $k=0.1 h Mpc^{-1}$, with the peak constrained to the range $0.03 < k < 0.06 h Mpc^{-1}$. It is consistent with a direct computation of the PS (Kolatt \& Dekel 1996). When compared to galaxy-density surveys, the implied values for $β$ ($\equiv Ω^{0.6}/b$) are of order unity to within 25\%. A $Γ$-shape model, free of COBE normalization, is constrained by the velocity data to $Γ=0.5\pm 0.15$ and $σ_8 Ω^{0.6}=0.85\pm 0.1$. Within the family of COBE-normalized CDM models, the best tilted model ($Ω=1$, $n \approx 0.84 h_{50}^{-0.65}$) and the best open model ($n=1$, $Ω\approx 0.75 h_{50}^{-1.0}$) are more likely than the best $Λ$ model ($n=1$, $Λ=1-Ω$, $Ω\approx 0.70 h_{50}^{-1.2}$). The most likely CDM model with $Ω\leq 1$ is found to be of $Ω=1$, $h\approx 0.75$, and a {\it tilted} spectrum of $n= 0.8\pm 0.02$ with tensor fluctuations.The tightest constraint for the tilted-$Λ$ models is of the sort $Ωh_{50}^{1.2} n^ν= 0.7 \pm 0.08$, with $ν=3.8$ and $1.85$ with and without tensor fluctuations respectively.
The mass density field in the local universe, recovered by the POTENT method from peculiar velocities of $\sim$3000 galaxies, is compared with the density field of optically-selected galaxies. Both density fields are smoothed with a Gaussian filter of radius 12 $h^{-1}$ Mpc. Under the assumptions of gravitational instability and a linear biasing parameter $b\sbo$ between optical galaxies and mass, we obtain $β\sbo \equiv \om^{0.6}/b\sbo = 0.74 \pm 0.13$. This result is obtained from a regression of POTENT mass density on optical density after correcting the mass density field for systematic biases in the velocity data and POTENT method. The error quoted is just the $1σ$ formal error estimated from the observed scatter in the density--density scatterplot; it does not include the uncertainty due to cosmic scatter in the mean density or in the biasing relation. We do not attempt a formal analysis of the goodness of fit, but the scatter about the fit is consistent with our estimates of the uncertainties.
We propose a method for measuring the cosmological density parameter $Ω$ from the statistics of the divergence field, $θ\equiv H^{-1} ÷v$, the divergence of peculiar velocity, expressed in units of the Hubble constant, $H \equiv 100 h km/s/Mpc$. The velocity field is spatially smoothed over $\sim 10 h^{-1} Mpc$ to remove strongly nonlinear effects. Assuming weakly-nonlinear gravitational evolution from Gaussian initial fluctuations, and using second-order perturbative analysis, we show that $ <θ^3> \propto -Ω^{-0.6} <θ^2>^2$. The constant of proportionality depends on the smoothing window. For a top-hat of radius R and volume-weighted smoothing, this constant is $26/7-γ$, where $γ=-d\log <θ^2> / d\log R$. If the power spectrum is a power law, $P(k)\propto k^n$, then $γ=3+n$. A Gaussian window yields similar results. The resulting method for measuring $Ω$ is independent of any assumed biasing relation between galaxies and mass. The method has been successfully tested with numerical simulations. A preliminary application to real data, provided by the POTENT recovery procedure from observed velocities favors $Ω\sim 1$. However, because of an uncertain sampling error, this result should be treated as an assessment of the feasibility of our method rather than a definitive measurement of $Ω$.
We challenge a widely accepted assumption of observational cosmology: that successful reconstruction of observed galaxy density fields from measured galaxy velocity fields (or vice versa), using the methods of gravitational instability theory, implies that the observed large-scale structures and large-scale flows were produced by the action of gravity. This assumption is false, in that there exist non-gravitational theories that pass the reconstruction tests and gravitational theories with certain forms of biased galaxy formation that fail them. Gravitational instability theory predicts specific correlations between large-scale velocity and mass density fields, but the same correlations arise in any model where (a) structures in the galaxy distribution grow from homogeneous initial conditions in a way that satisfies the continuity equation, and (b) the present-day velocity field is irrotational and proportional to the time-averaged velocity field. We demonstrate these assertions using analytical arguments and N-body simulations of gravitational and non-gravitational models. We also show examples of gravitational and non-gravitational models that {\it fail} reconstruction tests because galaxy formation is modulated (``biased'') in a way that violates the continuity equation. We discuss the relation between the value of $Ω$ inferred from velocity-density comparisons and the true cosmological value.
We test the hypothesis that the velocity field derived from Tully-Fisher measurements of spiral galaxies, and that derived independently from Dn-sigma measurements of ellipticals and S0s, are noisy versions of the same underlying velocity field. The radial velocity fields are derived using tensor Gaussian smoothing of radius 1200 km/s. They are compared at grid points near which the sampling by both types of galaxies is proper. This requirement defines a volume of ~(50 Mpc/h)^3, containing ~10 independent subvolumes, mostly limited by the available ellipticals. The two fields are compared using a correlation statistic, whose distribution is determined via Monte-Carlo simulations. We find that the data is consistent with the hypothesis, at the 10% level. We demonstrate that the failure to reject the correlation is not just a result of the errors being big, by using the same method to rule out complete independence between the fields at the 99.8% level. The zero points of the two distance indicators are matched by maximizing the correlation between the two velocity fields. There is a marginal hint that the ellipticals tend to stream slower than the spirals by ~8%. The correlation reinforced here is consistent with the common working hypotheses that (a) the derived large-scale velocity field is real, (b) it has a gravitational origin, and (c) the large-scale velocities of spirals and ellipticals are hardly biased relative to each other. On the other hand, it does not rule out any alternative to gravity where objects of all types obtain similar large-scale velocities.
We recover the one-point probability distribution function of the {\it initial} density fluctuations (\ipdf) from the quasi-linear galaxy density field of the 1.2 Jy \iras\ redshift survey smoothed by $10\hmpc$. The recovery, using the laminar, Eulerian, Zel'dovich approximation of Nusser and Dekel, is independent of $Ω$ and is relatively insensitive to linear galaxy biasing in the range $0.5 \leq b \leq 2$. Errors due to discrete sampling, the limited survey volume, and the method of recovering the \ipdf\ are evaluated by comparing the \ipdf\ determined from the \iras\ data with Monte-Carlo \iras-like catalogs ``observed" from $N$-body simulations of CDM models with Gaussian initial conditions. Eight sensitive statistical tests used in this comparison find the \iras\ \ipdf\ to be consistent with Gaussian. We provide observational constraints on possible deviations from Gaussianity, which should be obeyed by any theoretical model.