The Fisher Information Matrix formalism is extended to cases where the data is divided into two parts (X,Y), where the expectation value of Y depends on X according to some theoretical model, and X and Y both have errors with arbitrary covariance. In the simplest case, (X,Y) represent data pairs of abscissa and ordinate, in which case the analysis deals with the case of data pairs with errors in both coordinates, but X can be any measured quantities on which Y depends. The analysis applies for arbitrary covariance, provided all errors are gaussian, and provided the errors in X are small, both in comparison with the scale over which the expected signal Y changes, and with the width of the prior distribution. This generalises the Fisher Matrix approach, which normally only considers errors in the `ordinate' Y. In this work, we include errors in X by marginalising over latent variables, effectively employing a Bayesian hierarchical model, and deriving the Fisher Matrix for this more general case. The methods here also extend to likelihood surfaces which are not gaussian in the parameter space, and so techniques such as DALI (Derivative Approximation for Likelihoods) can be generalised straightforwardly to include arbitrary gaussian data error covariances. For simple mock data and theoretical models, we compare to Markov Chain Monte Carlo experiments, illustrating the method with cosmological supernova data. We also include the new method in the Fisher4Cast software.

A new non-parametric method based on Gaussian Processes was proposed recently to measure the Hubble constant $H_0$. The freedom in this approach comes in the chosen covariance function, which determines how smooth the process is and how nearby points are correlated. We perform coverage tests with a thousand mock samples within the LCDM model in order to determine what covariance function provides the least biased results. The function Matern(5/2) is the best with sligthly higher errors than other covariance functions, although much more stable when compared to standard parametric analyses.

An intriguing discrepancy emerging in the concordance model of cosmology is the tension between the locally measured value of the Hubble rate, and the 'global' value inferred from the cosmic microwave background (CMB). This could be due to systematic uncertainties when measuring $H_0$ locally, or it could be that we live in a highly unlikely Hubble bubble, or other exotic scenarios. We point out that the global $H_0$ can be found by extrapolating $H(z)$ data points at high-$z$ down to $z=0$. By doing this in a Bayesian non-parametric way we can find a model-independent value for $H_0$. We apply this to 19 measurements based on differential age of passively evolving galaxies as cosmic chronometers. Using Gaussian processes, we find $H_0=64.9 \pm 4.2$ km s$^{-1}$ Mpc$^{-1}$ $(1\sigma)$, in agreement with the CMB value, but reinforcing the tension with the local value. An analysis of possible sources of systematic errors shows that the stellar population synthesis model adopted may change the results significantly, being the main concern for subsequent studies. Forecasts for future data show that distant $H(z)$ measurements can be a robust method to determine $H_0$, where a focus in precision and a careful assessment of systematic errors are required.

The standard concordance model of the Universe is based on the cosmological constant as the driver of accelerating expansion. This concordance model is being subjected to a growing range of inter-locking observations. In addition to using generic observational tests, one can also design tests that target the specific properties of the cosmological constant. These null tests do not rely on parametrisations of observables, but focus on quantities that are constant only if dark energy is a cosmological constant. We use supernova data in null tests that are based on the luminosity distance. In order to extract derivatives of the distance in a model-independent way, we use Gaussian Processes. We find that the concordance model is compatible with the Union 2.1 data, but the error bars are fairly large. Simulated datasets are generated for the DES supernova survey and we show that this survey will allow for a sharper null test of the cosmological constant if we assume the Universe is flat. Allowing for spatial curvature degrades the power of the null test.

Gaussian processes are a fully Bayesian smoothing technique that allows for the reconstruction of a function and its derivatives directly from observational data, without assuming a specific model or choosing a parameterization. This is ideal for constraining dark energy because physical models are generally phenomenological and poorly motivated. Model-independent constraints on dark energy are an especially important alternative to parameterized models, as the priors involved have an entirely different source so can be used to check constraints formulated from models or parameterizations. A critical prior for Gaussian process reconstruction lies in the choice of covariance function. We show how the choice of covariance function affects the result of the reconstruction, and present a choice which leads to reliable results for present day supernovae data. We also introduce a method to quantify deviations of a model from the Gaussian process reconstructions.

We test the isotropy of the expansion of the Universe by estimating the hemispherical anisotropy of supernova type Ia (SN Ia) Hubble diagrams at low redshifts (z<0.2). We compare the best fit Hubble diagrams in pairs of hemispheres and search for the maximal asymmetric orientation. For an isotropic Universe, we expect only a small asymmetry due to noise and the presence of nearby structures. This test does not depend on the assumed content of the Universe, the assumed model of gravity, or the spatial curvature of the Universe. The expectation for possible fluctuations due to large scale structure is evaluated for the \Lambda cold dark matter (\Lambda CDM) model and is compared to the supernova data from the Constitution set for four different light curve fitters, thus allowing a study of the systematic effects. The expected order of magnitude of the hemispherical asymmetry of the Hubble expansion agrees with the observed one. The direction of the Hubble asymmetry is established at 95% confidence level (C.L.) using both, the MLCS2k2 and the SALT II light curve fitter. The highest expansion rate is found towards (l, b) ~ (-35{\deg},-19{\deg}), which agrees with directions reported by other studies. Its amplitude is not in contradiction to expectations from the \Lambda CDM model. The measured Hubble anisotropy is \Delta H/H ~ 0.026. With 95% C.L. the expansion asymmetry is \Delta H/H<0.038.

We analyse the possibility that our Universe could be described by the model recently proposed by Melia & Shevchuk (2012), where the Hubble scale R_h=c/H is at all times equal to the distance ct that light has travelled since the Big Bang. In such a model, the scale factor is proportional to cosmic time and there is neither acceleration nor deceleration of the expansion. We first point out problems with the very foundations of the model and its consequences for the evolution of the Universe. Next, we compare predictions of the model with observational data. As probes of the expansion we use distance data of supernovae type Ia, as well as Hubble rate data obtained from cosmic chronometers and radial baryon acoustic oscillations. We analyse the redshift evolution of the Hubble parameter and its redshift derivatives, together with the so-called O_m diagnostic and the deceleration parameter. To reliably estimate smooth functions and their derivatives from discrete data, we use the recently developed Gaussian Processes in Python package (GaPP). Our general conclusion is that the discussed model is strongly disfavoured by observations, especially at low redshifts (z<0.5). In particular, it predicts specific constant values for the deceleration parameter and for redshift derivatives of the Hubble parameter, which is ruled out by the data.

Direct observations of the Hubble rate, from cosmic chronometers and the radial baryon acoustic oscillation scale, can out-perform supernovae observations in understanding the expansion history, because supernovae observations need to be differentiated to extract H(z). We use existing H(z) data and smooth the data using a new Gaussian Processes package, GaPP, from which we can also estimate derivatives. The obtained Hubble rate and its derivatives are used to reconstruct the equation of state of dark energy and to perform consistency tests of the LCDM model, some of which are newly devised here. Current data is consistent with the concordance model, but is rather sparse. Future observations will provide a dramatic improvement in our ability to constrain or refute the concordance model of cosmology. We produce simulated data to illustrate how effective H(z) data will be in combination with Gaussian Processes.

An important issue in cosmology is reconstructing the effective dark energy equation of state directly from observations. With few physically motivated models, future dark energy studies cannot only be based on constraining a dark energy parameter space, as the errors found depend strongly on the parameterisation considered. We present a new non-parametric approach to reconstructing the history of the expansion rate and dark energy using Gaussian Processes, which is a fully Bayesian approach for smoothing data. We present a pedagogical introduction to Gaussian Processes, and discuss how it can be used to robustly differentiate data in a suitable way. Using this method we show that the Dark Energy Survey - Supernova Survey (DES) can accurately recover a slowly evolving equation of state to sigma_w = +-0.04 (95% CL) at z=0 and +-0.2 at z=0.7, with a minimum error of +-0.015 at the sweet-spot at z~0.14, provided the other parameters of the model are known. Errors on the expansion history are an order of magnitude smaller, yet make no assumptions about dark energy whatsoever. A code for calculating functions and their first three derivatives using Gaussian processes has been developed and is available for download at http://www.acgc.uct.ac.za/~seikel/GAPP/index.html .

We present a short (and necessarily incomplete) review of the evidence for the accelerated expansion of the Universe. The most direct probe of acceleration relies on the detailed study of supernovae (SN) of type Ia. Assuming that these are standardizable candles and that they fairly sample a homogeneous and isotropic Universe, the evidence for acceleration can be tested in a model- and calibration-independent way. Various light-curve fitting procedures have been proposed and tested. While several fitters give consistent results for the so-called Constitution set, they lead to inconsistent results for the recently released SDSS SN. Adopting the SALT fitter and relying on the Union set, cosmic acceleration is detected by a purely kinematic test at 7 sigma when spatial flatness is assumed and at 4 sigma without assumption on the spatial geometry. A weak point of the described method is the local set of SN (at z < 0.2), as these SN are essential to anchor the Hubble diagram. These SN are drawn from a volume much smaller than the Hubble volume and could be affected by local structure. Without the assumption of homogeneity, there is no evidence for acceleration, as the effects of acceleration are degenerate with the effects of inhomogeneities. Unless we sit in the centre of the Universe, such inhomogeneities can be constrained by SN observations by means of tests of the isotropy of the Hubble flow.

As the Einstein equations are non-linear, spatial averaging and temporal evolution do not commute. Therefore, the evolution of the averaged universe is affected by inhomogeneities. It is, however, highly controversial how large these cosmological backreaction effects are. We use the supernova data of the Constitution set up to a redshift of 0.1 in order to analyse to what extent the measurement of the Hubble constant is affected. The size of the effect depends on the size of the volume that is averaged over. The observational results are then compared to the theory of the backreaction mechanism.

In this work, we consider a braneworld model with a timelike extra-dimension. There are strong constraints to the parameter values of such a model resulting from the claim that there must be a physical solution to the Friedmann equation at least between now and the time of recombination. We fitted the model to supernova type Ia data and checked the consistency of the result with other observations. For parameter values that are consistent with observations, the braneworld model is indistinguishable from a LambdaCDM universe as far as the considered cosmological tests are concerned.

We present a calibration-independent test of the accelerated expansion of the universe using supernova type Ia data. The test is also model-independent in the sense that no assumptions about the content of the universe or about the parameterization of the deceleration parameter are made and that it does not assume any dynamical equations of motion. Yet, the test assumes the universe and the distribution of supernovae to be statistically homogeneous and isotropic. A significant reduction of systematic effects, as compared to our previous, calibration-dependent test, is achieved. Accelerated expansion is detected at significant level (4.3 sigma in the 2007 Gold sample, 7.2 sigma in the 2008 Union sample) if the universe is spatially flat. This result depends, however, crucially on supernovae with a redshift smaller than 0.1, for which the assumption of statistical isotropy and homogeneity is less well established.

The present standard model of cosmology states that the known particles carry only a tiny fraction of total mass and energy of the Universe. Rather, unknown dark matter and dark energy are the dominant contributions to the cosmic energy budget. We review the logic that leads to the postulated dark energy and present an alternative point of view, in which the puzzle may be solved by properly taking into account the influence of cosmic structures on global observables. We illustrate the effect of averaging on the measurement of the Hubble constant.

We test the present expansion of the universe using supernova type Ia data without making any assumptions about the matter and energy content of the universe or about the parameterization of the deceleration parameter. We assume the cosmological principle to apply in a strict sense. The result strongly depends on the data set, the light-curve fitting method and the calibration of the absolute magnitude used for the test, indicating strong systematic errors. Nevertheless, in a spatially flat universe there is at least a 5 sigma evidence for acceleration which drops to 1.8 sigma in an open universe.