Filipe B. Abdalla

Upper bound of 0.28 eV on neutrino masses from the largest photometric redshift survey.

We present a new limit of ∑m(v) ≤ 0.28 (95% CL) on the sum of the neutrino masses assuming a flat ΛCDM cosmology. This relaxes slightly to ∑m(ν) ≤ 0.34 and ∑m(v) ≤ 0.47 when quasinonlinear scales are removed and w≠ -1, respectively. These are derived from a new photometric catalogue of over 700,000 luminous red galaxies (MegaZ DR7) with a volume of 3.3  (Gpc h(-1))(3) and redshift range 0.45 < z < 0.65. The data are combined with WMAP 5-year CMB, baryon acoustic oscillations, supernovae, and a Hubble Space Telescope prior on h. When combined with WMAP these data are as constraining as adding all supernovae and baryon oscillation data available. The upper limit is one of the tightest constraints on the neutrino from cosmology or particle physics. Further, if these bounds hold, they all predict that current-to-next generation neutrino experiments, such as KATRIN, are unlikely to obtain a detection.

The WiggleZ Dark Energy Survey: direct constraints on blue galaxy intrinsic alignments at intermediate redshifts

Correlations between the intrinsic shapes of galaxy pairs, and between the intrinsic shapes of galaxies and the large-scale density field, may be induced by tidal fields. These correlations, which have been detected at low redshifts (z < 0.35) for bright red galaxies in the Sloan Digital Sky Survey (SDSS), and for which upper limits exist for blue galaxies at z similar to 0.1, provide a window into galaxy formation and evolution, and are also an important contaminant for current and future weak lensing surveys. Measurements of these alignments at intermediate redshifts (z similar to 0.6) that are more relevant for cosmic shear observations are very important for understanding the origin and redshift evolution of these alignments, and for minimizing their impact on weak lensing measurements. We present the first such intermediate-redshift measurement for blue galaxies, using galaxy shape measurements from SDSS and spectroscopic redshifts from the WiggleZ Dark Energy Survey. Our null detection allows us to place upper limits on the contamination of weak lensing measurements by blue galaxy intrinsic alignments that, for the first time, do not require significant model-dependent extrapolation from the z similar to 0.1 SDSS observations. Also, combining the SDSS and WiggleZ constraints gives us a long redshift baseline with which to constrain intrinsic alignment models and contamination of the cosmic shear power spectrum. Assuming that the alignments can be explained by linear alignment with the smoothed local density field, we find that a measurement of Sigma(8) in a blue-galaxy dominated, CFHTLS-like survey would be contaminated by at most +0.02(-0.03) (95 per cent confidence level, SDSS and WiggleZ) or +/- 0.03 (WiggleZ alone) due to intrinsic alignments. We also allow additional power-law redshift evolution of the intrinsic alignments, due to (for example) effects like interactions and mergers that are not included in the linear alignment model, and find that our constraints on cosmic shear contamination are not significantly weakened if the power-law index is less than similar to 2. The WiggleZ sample (unlike SDSS) has a long enough redshift baseline that the data can rule out the possibility of very strong additional evolution.

Galaxy Zoo: reproducing galaxy morphologies via machine learning★

We present morphological classifications obtained using machine learning for objects in the Sloan Digital Sky Survey DR6 that have been classified by Galaxy Zoo into three classes, namely early types, spirals and point sources/artefacts. An artificial neural network is trained on a subset of objects classified by the human eye, and we test whether the machine-learning algorithm can reproduce the human classifications for the rest of the sample. We find that the success of the neural network in matching the human classifications depends crucially on the set of input parameters chosen for the machine-learning algorithm. The colours and parameters associated with profile fitting are reasonable in separating the objects into three classes. However, these results are considerably improved when adding adaptive shape parameters as well as concentration and texture. The adaptive moments, concentration and texture parameters alone cannot distinguish between early type galaxies and the point sources/artefacts. Using a set of 12 parameters, the neural network is able to reproduce the human classifications to better than 90 per cent for all three morphological classes. We find that using a training set that is incomplete in magnitude does not degrade our results given our particular choice of the input parameters to the network. We conclude that it is promising to use machine-learning algorithms to perform morphological classification for the next generation of wide-field imaging surveys and that the Galaxy Zoo catalogue provides an invaluable training set for such purposes.

Cross-correlation of 2MASS and WMAP 3: implications for the integrated Sachs-Wolfe effect

We perform a cross-correlation of the cosmic microwave background using the third year Wilkinson Microwave Anisotropy Probe (WMAP) data with the Two Micron All Sky Survey (2MASS) galaxy map (about 828 000 galaxies with median redshift z ≈ 0.07). One motivation is to detect the integrated Sachs‐Wolfe (ISW) effect, expected if the cosmic gravitational potential is time dependent; for example, as it is in a flat universe with a dark energy component. The measured spherical harmonic cross-correlation signal favours the ISW signal expected in the concordance Lambda cold dark matter (� CDM) model over that of zero correlation, although

Foreground Removal using FastICA: A Showcase of LOFAR-EoR

We introduce a new implementation of the fastica algorithm on simulated Low Frequency Array Epoch of Reionization data with the aim of accurately removing the foregrounds and extracting the 21-cm reionization signal. We find that the method successfully removes the foregrounds with an average fitting error of 0.5 per cent and that the 2D and 3D power spectra are recovered across the frequency range. We find that for scales above several point spread function scales, the 21-cm variance is successfully recovered though there is evidence of noise leakage into the reconstructed foreground components. We find that this blind independent component analysis technique provides encouraging results without the danger of prior foreground assumptions.

The scale of the problem: recovering images of reionization with Generalized Morphological Component Analysis

The accurate and precise removal of 21-cm foregrounds from Epoch of Reionization redshifted 21-cm emission data is essential if we are to gain insight into an unexplored cosmological era. We apply a non-parametric technique, Generalized Morphological Component Analysis or GMCA, to simulated LOFAR-EoR data and show that it has the ability to clean the foregrounds with high accuracy. We recover the 21-cm 1D, 2D and 3D power spectra with high accuracy across an impressive range of frequencies and scales. We show that GMCA preserves the 21-cm phase information, especially when the smallest spatial scale data is discarded. While it has been shown that LOFAR-EoR image recovery is theoretically possible using image smoothing, we add that wavelet decomposition is an efficient way of recovering 21-cm signal maps to the same or greater order of accuracy with more flexibility. By comparing the GMCA output residual maps (equal to the noise, 21-cm signal and any foreground fitting errors) with the 21-cm maps at one frequency and discarding the smaller wavelet scale information, we find a correlation coefficient of 0.689, compared to 0.588 for the equivalently smoothed image. Considering only the central 50% of the maps, these coefficients improve to 0.905 and 0.605 respectively and we conclude that wavelet decomposition is a significantly more powerful method to denoise reconstructed 21-cm maps than smoothing.

Excess Clustering on Large Scales in the MegaZ DR7 Photometric Redshift Survey

We observe a large excess of power in the statistical clustering of luminous red galaxies in the photometric SDSS galaxy sample called MegaZ DR7. This is seen over the lowest multipoles in the angular power spectra C_{ℓ} in four equally spaced redshift bins between 0.45≤z≤0.65. However, it is most prominent in the highest redshift band at ∼4σ and it emerges at an effective scale k≲0.01  h Mpc(-1). Given that MegaZ DR7 is the largest cosmic volume galaxy survey to date (3.3(Gpch(-1))(3)) this implies an anomaly on the largest physical scales probed by galaxies. Alternatively, this signature could be a consequence of it appearing at the most systematically susceptible redshift. There are several explanations for this excess power that range from systematics to new physics. We test the survey, data, and excess power, as well as possible origins.

Photometric redshifts for the Dark Energy Survey and VISTA and implications for large-scale structure

We conduct a detailed analysis of the photometric redshift requirements for the proposed Dark Energy Survey (DES) using two sets of mock galaxy simulations and an artificial neural network code –annz. In particular, we examine how optical photometry in the DES grizY bands can be complemented with near-infrared photometry from the planned VISTA Hemisphere Survey (VHS) in the JHKs bands. We find that the rms scatter on the photometric redshift estimate over 1 < z < 2 is σz= 0.2 from DES alone and σz= 0.15 from DES + VISTA, i.e. an improvement of more than 30 per cent. We draw attention to the effects of galaxy formation scenarios such as reddening on the photo-z estimate and using our neural network code, calculate the extinction, Av for these reddened galaxies. We also look at the impact of using different training sets when calculating photometric redshifts. In particular, we find that using the ongoing DEEP2 and VVDS-Deep spectroscopic surveys to calibrate photometric redshifts for DES, will prove effective. However, we need to be aware of uncertainties in the photometric redshift bias that arise when using different training sets as these will translate into errors in the dark energy equation of state parameter, w. Furthermore, we show that the neural network error estimate on the photometric redshift may be used to remove outliers from our samples before any kind of cosmological analysis, in particular for large-scale structure experiments. By removing all galaxies with a neural network photo-z error estimate of greater than 0.1 from our DES + VHS sample, we can constrain the galaxy power spectrum out to a redshift of 2 and reduce the fractional error on this power spectrum by ∼15–20 per cent compared to using the entire catalogue. Output tables of spectroscopic redshift versus photometric redshift used to produce the results in this paper can be found at http://www.star.ucl.ac.uk/~mbanerji/DESdata.

Constraining Modified Gravity and Growth with Weak Lensing

The idea that we live in a Universe undergoing a period of acceleration is a new, yet strongly held, notion in cosmology. As this can, potentially, be explained with a modification to general relativity, we look at current cosmological data with the purpose of testing gravity. First, we constrain a phenomenological model [modified Dvali Gabadadze Porrati (mDGP)] motivated by a possible extra dimension. This is characterized by a parameter a which interpolates between alpha = 0 [lambda cold dark matter (LCDM)] and alpha = 1 (the DGP model). In addition, we analyse general signatures of modified gravity given by the growth parameter gamma and power spectrum parameter Sigma. We utilize large angular scale (theta > 30 arcmin) weak lensing data (Canada-France-Hawaii Telescope Legacy Survey wide) in order to work in the more linear regime and then add, in combination, baryon acoustic oscillations (B Lambda Os) and supernovae. We subsequently show that current weak-lensing data are not yet capable of constraining either model in isolation. However, we demonstrate that even at present this probe is highly beneficial, for in combination with BAOs and Supernovae we obtain alpha < 0.58 and 0.91 at 1 sigma and 2 sigma, respectively. Without the lensing data, no constraint is possible. This corresponds to a disfavouring of the flat DGP braneworld model at over 2 sigma. We highlight these are insensitive to potential systematics in the lensing data such as an underestimation of the shear at high redshift. For the growth signature gamma, we show that, in combination, these probes do not yet have sufficient constraining power. Finally, we look beyond these present capabilities and demonstrate that Euclid, a future weak-lensing survey, will deeply probe the nature of gravity. A 1 sigma error of 0.104 is found for alpha (l(max) = 500) whereas for the general modified signatures we forecast 1 sigma errors of 0.045 for gamma and 0.25 for Sigma(0)(l(max) = 500), which is further tightened to 0.038 for gamma and 0.069 for Sigma(0)(l(max) = 10 000).

The Scale of the Problem : Recovering Images of Reionization with GMCA

The accurate and precise removal of 21-cm foregrounds from Epoch of Reionization redshifted 21-cm emission data is essential if we are to gain insight into an unexplored cosmological era. We apply a non-parametric technique, Generalized Morphological Component Analysis or GMCA, to simulated LOFAR-EoR data and show that it has the ability to clean the foregrounds with high accuracy. We recover the 21-cm 1D, 2D and 3D power spectra with high accuracy across an impressive range of frequencies and scales. We show that GMCA preserves the 21-cm phase information, especially when the smallest spatial scale data is discarded. While it has been shown that LOFAR-EoR image recovery is theoretically possible using image smoothing, we add that wavelet decomposition is an efficient way of recovering 21-cm signal maps to the same or greater order of accuracy with more flexibility. By comparing the GMCA output residual maps (equal to the noise, 21-cm signal and any foreground fitting errors) with the 21-cm maps at one frequency and discarding the smaller wavelet scale information, we find a correlation coefficient of 0.689, compared to 0.588 for the equivalently smoothed image. Considering only the central 50% of the maps, these coefficients improve to 0.905 and 0.605 respectively and we conclude that wavelet decomposition is a significantly more powerful method to denoise reconstructed 21-cm maps than smoothing.