Search

Craters are among the most studied geomorphic features in the Solar System because they yield important information about the past and present geological processes and provide information about the relative ages of observed geologic formations. We present a method for automatic crater detection using advanced machine learning to deal with the large amount of satellite imagery collected. The challenge of automatically detecting craters comes from their is complex surface because their shape erodes over time to blend into the surface. Bandeira provided a seminal dataset that embodied this challenge that is still an unsolved pattern recognition problem to this day. There has been work to solve this challenge based on extracting shape and contrast features and then applying classification models on those features. The limiting factor in this existing work is the use of hand crafted filters on the image such as Gabor or Sobel filters or Haar features. These hand crafted methods rely on domain knowledge to construct. We would like to learn the optimal filters and features based on training examples. In order to dynamically learn filters and features we look to Convolutional Neural Networks (CNNs) which have shown their dominance in computer vision. The power of CNNs is that they can learn image filters which generate features for high accuracy classification., Comment: 2 Pages. Submitted to 47th Lunar and Planetary Science Conference (LPSC 2016)

We quantify the Fisher information content of the cosmic shear survey two-point function as a function of noise and resolution. The two point information of dark matter saturates at the trans-linear scale. We investigate the impact of non-linear non-Gaussianity on the information content for lensing, which probes the same dark matter. To do so we heavily utilize N-body simulations in order to probe accurately the non-linear regime. While we find that even in a perfect survey, there is no clear saturation scale, we observe that non-linear growth induced non-Gaussianity could lead to a factor of ~4 reduction for the common Dark Energy figure of merit. This effect is however mitigated by realistic levels of shot noise and we find that for future surveys, the effect is closer to a factor of 1.5. To do so, we develop a new scheme to compute the relevant covariant matrix. It leads us to claim an unbiased estimator with an order of magnitude improvement in accuracy with only twice more simulations than previously used. Finally, we evaluate the error on the errors using bootstrap methods., Comment: submitted to PRD

Wide area Large-Scale Structure (LSS) surveys are planning to map a substantial fraction of the visible universe to quantify dark energy through Baryon Acoustic Oscillations (BAO). At increasing redshift, for example that probed by proposed 21-cm intensity mapping surveys, gravitational lensing potentially limits the fidelity (Hui et al., 2007) because it distorts the apparent matter distribution. In this paper we show that these distortions can be reconstructed, and actually used to map the distribution of intervening dark matter. The lensing information for sources at z=1-3 allows accurate reconstruction of the gravitational potential on large scales, l <~ 100, which is well matched for Integrated Sachs-Wolfe (ISW) effect measurements of dark energy and its sound speed, and a strong constraint for modified gravity models of dark energy. We built an optimal quadratic lensing estimator for non-Gaussian sources, which is necessary for LSS. The phenomenon of "information saturation" (Rimes & Hamilton, 2005) saturates reconstruction at mildly non-linear scales, where the linear source power spectrum Delta^2 ~ 0.2-0.5. We find that steeper power spectra, saturate more quickly. We compute the effective number densities of independent lensing sources for LSS lensing, and find that they increase rapidly with redshifts. For LSS/21-cm sources at z ~ 2-4, the lensing reconstruction is limited by cosmic variance at l<~ 100., Comment: submitted to PRD

Long-term precise timing of Galactic millisecond pulsars holds great promise for measuring the long-period (months-to-years) astrophysical gravitational waves. Several gravitational-wave observational programs, called Pulsar Timing Arrays (PTA), are being pursued around the world. Here we develop a Bayesian algorithm for measuring the stochastic gravitational-wave background (GWB) from the PTA data. Our algorithm has several strengths: (1) It analyses the data without any loss of information, (2) It trivially removes systematic errors of known functional form, including quadratic pulsar spin-down, annual modulations and jumps due to a change of equipment, (3) It measures simultaneously both the amplitude and the slope of the GWB spectrum, (4) It can deal with unevenly sampled data and coloured pulsar noise spectra. We sample the likelihood function using Markov Chain Monte Carlo (MCMC) simulations. We extensively test our approach on mock PTA datasets, and find that the algorithm has significant benefits over currently proposed counterparts. We show the importance of characterising all red noise components in pulsar timing noise by demonstrating that the presence of a red component would significantly hinder a detection of the GWB. Lastly, we explore the dependence of the signal-to-noise ratio on the duration of the experiment, number of monitored pulsars, and the magnitude of the pulsar timing noise. These parameter studies will help formulate observing strategies for the PTA experiments., Comment: 10 pages, 12 figures

We study the limits of accuracy for weak lensing maps of dark matter using diffuse 21-cm radiation from the pre-reionization epoch using simulations. We improve on previous "optimal" quadratic lensing estimators by using shear and convergence instead of deflection angles. We find that non-Gaussianity provides a limit to the accuracy of weak lensing reconstruction, even if instrumental noise is reduced to zero. The best reconstruction result is equivalent to Gaussian sources with effectively independent cell of side length 2.0/h Mpc. Using a source full map from z=10-20, this limiting sensitivity allows mapping of dark matter at a Signal-to-Noise ratio (S/N) greater than 1 out to l < 6000, which is better than any other proposed technique for large area weak lensing mapping., Comment: 16 pages, 7 figures, submitted to MNRAS

We present the first optimal power spectrum estimation and three dimensional deprojections for the dark and luminous matter and their cross correlations. The results are obtained using a new optimal fast estimator (Pen 2003) deprojected using minimum variance and SVD techniques. We show the resulting 3-D power spectra for dark matter and galaxies, and their covariance for the VIRMOS-DESCART weak lensing shear and galaxy data. The survey is most sensitive to nonlinear scales k_NL ~ 1 h Mpc^-1. On these scales, our 3-D power spectrum of dark matter is in good agreement with the RCS 3-D power spectrum found by (Hoekstra et al 2002). Our galaxy power is similar to that found by the 2MASS survey, and larger than that of SDSS, APM and RCS, consistent with the expected difference in galaxy population. We find an average bias b=1.24+/-0.18 for the I selected galaxies, and a cross correlation coefficient r=0.75+/-0.23. Together with the power spectra, these results optimally encode the entire two point information about dark matter and galaxies, including galaxy-galaxy lensing. We address some of the implications regarding galaxy halos and mass-to-light ratios. The best fit ``halo'' parameter h=r/b=0.57+/-0.16, suggesting that dynamical masses estimated using galaxies systematically underestimate total mass. Ongoing surveys, such as the Canada-France-Hawaii-Telescope-Legacy survey will significantly improve on the dynamic range, and future photometric redshift catalogs will allow tomography along the same principles., Comment: 17 pages, 19 figures, submitted to mnras