Search for: All records

We implement and explore high-dimensional generalized dark matter (HDGDM) with an arbitrary equation of state as a function of redshift as an extension to Λ cold dark matter. Exposing this model to cosmic microwave background, baryon acoustic oscillations, and supernova data, we demonstrate that the use of marginalized posterior distributions can easily lead to misleading conclusions on the viability of a high-dimensional model such as this one. We discuss such pitfalls and corresponding mitigation strategies, which can be used to search for an observationally favored region of the parameter space. We further show that the HDGDM model in particular does show promise in terms of its ability to ease the Hubble tension once such techniques are employed, and we find interesting features in the best-fitting equation of state that can serve as an inspiration for future model building. 

We seek to clarify the origin of constraints on the dark energy equation of state parameter from CMB lensing tomography, that is the combination of galaxy clustering and the cross-correlation of galaxies with CMB lensing in a number of redshift bins. We focus on the analytic understanding of the origin of the constraints. Dark energy information in these data arises from the influence of three primary relationships: distance as a function of redshift (geometry), the amplitude of the power spectrum as a function of redshift (growth), and the power spectrum as a function of wavenumber (shape). We find that the effects from geometry and growth play a significant role and partially cancel each other out, while the shape effect is unimportant. We also show that Dark Energy Task Force figure of merit forecasts from the combination of LSST galaxies and CMB-S4 lensing are comparable to the forecasts from cosmic shear in the absence of the CMB lensing map, thus providing an important independent check. Compared to the forecasts with the LSST galaxies alone, combining CMB lensing and LSST clustering information increases the FoM by roughly a factor of 3–4 in the optimistic scenario where systematics are fully under control. We caution that achieving these forecasts will likely require a full analysis of higher-order biasing, photometric redshift uncertainties, and stringent control of other systematic limitations, which are outside the scope of this work, whose primary purpose is to elucidate the physical origin of the constraints.

 

ABSTRACT Emission from the interstellar medium can be a significant contaminant of measurements of the intensity and polarization of the cosmic microwave background (CMB). For planning CMB observations, and for optimizing foreground-cleaning algorithms, a description of the statistical properties of such emission can be helpful. Here, we examine a machine learning approach to inferring the statistical properties of dust from observational data. In particular, we apply a type of neural network called a variational autoencoder (VAE) to maps of the intensity of emission from interstellar dust as inferred from Planck sky maps and demonstrate its ability to (i) simulate new samples with similar summary statistics as the training set, (ii) provide fits to emission maps withheld from the training set, and (iii) produce constrained realizations. We find VAEs are easier to train than another popular architecture: that of generative adversarial networks, and are better suited for use in Bayesian inference.