1612.00752
Non-linear shrinkage estimation of large-scale structure covariance
Joachimi
In many astrophysical settings covariance matrices of large datasets have to be determined empirically from a finite number of mock realizations. The resulting noise degrades inference and precludes it completely if there are fewer realizations than data points. This work applies a recently proposed non-linear shrinkage estimator of covariance to a realistic example from large-scale structure cosmology. After optimizing its performance for the usage in likelihood expressions, the shrinkage estimator yields subdominant bias and variance comparable to that of the standard estimator with a factor ~50 less realizations. This is achieved without any prior information on the properties of the data or the structure of the covariance matrix, at negligible computational cost.
1512.00770
Unequal-time correlates for cosmology
Kitching, Heavens
Measurements of the power spectrum from large-scale structure surveys have to date assumed an equal-time approximation, where the full cross-correlation power spectrum of the matter density field evaluated at different times (or distances) has been approximated either by the power spectrum at fixed time, or in an improved fashion, by a geometric mean P(k; r1, r2) = [P(k;r1)P(k;r2)]^1/2. In this paper, investigate the expected impact of the geometric mean ansatz, and present an application in assessing the impact on weak gravitational lensing cosmological parameter inference, using an perturbative unequal-time correlation. As one might expect, find that the impact of this assumption is greatest at large separations in redshift Delta z > 0.3 where the change in the amplitude of the matter power spectrum can be as much as 10 percent for k>5h/Mpc. However, of more concern is that the corrections for small separations, where the clustering is not close to zero, may not be negligibly small. In particular, find that for a Euclid- or LSST-like weak lensing experiment the assumption of dual-time correlators may result in biased predictions of the cosmic shear power spectrum, and that the impact is strongly dependent on the amplitude of the intrinsic alignment signal. To compute uneuqal-time correlations to sufficient accuracy will require advances in either perturbation theory to high k-modes, or extensive use of simulations.
1612.00839
The 2-degree field lensing survey: photometric redshifts from a large new training sample to r<19.5
Wolf, et al
Present a new training set for estimating empirical photometric redshifts of galaxies, which was created as part of the 2dFLenS project. This training set is located in a 700 sq deg area of the KiDS South field and is randomly selected and nearly complete at r<19.5. Investigate the photometric z performance obtained with ugriz photometry from VST-ATLAS and W1/W2 from WISE, based on several empirical and template methods. The best redshift errors are obtained with kernel-density estimation, as are the lowest biases, which are consistent with zero within statistical noise. The 68th percentiles of the redshift scatter for magnitude-limited samples at r<(15.5, 17.5, 19.5) are (0.014, 0.017, 0.028). In this magnitude range, there are no known ambiguities in the color-redshift map, consistent with a small rate of redshift outliers. In the fainter regime, the KDE method produces p(z) estimates per galaxy that represent unbiased and accurate redshift frequency expectations. The p(z) sum over any subsample is consistent with the true redshift frequency plus Poisson noise. Further improvements in redshift precision at r<20 would mostly be expected from filter sets with narrower passbands to increase the sensitivity of course to small changes in redshift.
1612.00847
Data-driven, interpretable photometric redshifts trained on heterogeneous and unrepresentative data
Leistedt, Hogg
Present a new method for inferring photo-z in deep galaxy and quasar surveys, based on a data driven model of latent SEDs and a physical model of photometric fluxes as a function of redshift. This conceptually novel approach combines the advantages of both machine-learning and template-fitting methods by building template SEDs directly from the training data. This is made computationally tractable with gaussian Processes operating in flux-z space, encoding the physics of redshift and the projection of galaxy SEDs onto photometric band passes. This method alleviates the need of acquiring representative training data or constructing detailed galaxy SED models; it requires only that the photometric band passes and calibrations be known or have parameterized unknowns. The training data can consist of a combination of spectroscopic and deep many-band photometric data, which do not need to entirely spatially overlap with the target survey of interest or even involve the same photometric bands. Showase the method on the i-magnitude-selected, spectroscopically-confirmed galaxies in the COSMOS field. The model is trained on the deepest bands (from SUBARU and HST) and photo-zs are derived using the shallower SDSS optical bands only. Demonstrate that accurate z point estimates and probability distributions are obtained despite the training and target sets having very different z distributions, noise properties, and even photometric bands. The model can also be used to predict missing photometric fluxes, or to simulated populations of galaxies with realistic fluxes and z, for example. This method opens a new era in which photo-z for large photometric surveys are derived using a flexible yet physical model of the data trained on all available surveys (spectroscopic and photometric).
1612.00852
On the validity of the Born approximation for beyond-Gaussian weak lensing observables
Petri, Haiman, May
Accurate forward modeling of WL observables from cosmo parameters is necessary for upcoming galaxy surveys. Because WL probes structures in the non-linear regime, analytical forward modeling is very challenging, if not impossible. Numerical sims of WL features rely on ray-tracing through the outputs of N-body sims, which requires knowledge of the gravitational potential and accurate solvers for light ray trajectories. A less accurate procedure, based on the Born approximation, only requires knowledge of the density field, and can be implemented more efficiently and at a lower computational cost. In this work, use sims to show that deviations of the Born-approximated convergence power spectrum, skewness and kurtosis from their fully ray-traced counterparts are consistent with the smallest non-trivial post-Born corrections (so-called geodesic and lens-lens terms). Find, however, that the perturbative approach for the geodesic correction breaks down at higher orders. Also find that cosmo parameter biases induced by the Born approximation are negligible even for an LSST-like analyses. Using the LensTools software suite, show that the Born approximation saves a factor of 4 in computing time with respect to the full ray-tracing in reconstructing the convergence.
No comments:
Post a Comment