Session Y14: Data Analysis and Modeling Techniques

Image Processing for Galaxy Ellipticity Analysis

Shape analysis of statistically large samples of galaxy images can be used to reveal the imprint of weak gravitational lensing by dark matter distributions. As new, large-scale surveys expand the potential catalog, galaxy shape analysis suffers the (coupled) problems of high noise and uncertainty in the prior morphology. We investigate a new image processing technique to help mitigate these problems, in which repeated auto-correlations and auto-convolutions are employed to push the true shape toward a universal (Gaussian) attractor while relatively suppressing uncorrelated pixel noise. The goal is reliable reconstruction of original image moments, independent of image shape. First test evaluations of the technique on small control samples will be presented, and future applicability discussed.   

A General Prescription for Digging Too Deeply in Imaging Surveys

Dark energy comprises the vast majority (\textgreater 70{\%}) of the energy density of the universe, but its physical nature is still very much a mystery. The most sensitive probes of dark energy in the foreseeable future are likely to weak lensing measurements from large space- and ground-based astronomical imaging surveys. The very sensitivity of these measurements, however, leaves them especially susceptible to small systematic biases. The diagnosis and removal of systematic errors in weak lensing measurements -- especially ``unknown unknowns'' -- is one of the foremost concerns in cosmological measurement today. We present an overview of a new, general method for optimal measurement and systematic error diagnosis in imaging surveys designed to address these issues. Our method perturbs the real images by embedding a known signal -- simulated lensed galaxies, convolved with the instrument's response -- and show how measurements on the modified images can be used to derive unbiased estimators for the signal in question that are optimal (i.e., that saturate the Cramer-Rao bound) for the specific observing conditions and data quality in hand. We also show how embedding a known systematic error can be used to diagnose and remove problems in the lensing measurement. Finally, we present some preliminary results from the application of this formalism to Dark Energy Survey images, and describe the resulting improvements in the statistical and systematic error budget.   

Novel Aspects of the DESI Data Acquisition System

The Dark Energy Spectroscopic Instrument (DESI) will measure the effect of dark energy on the expansion of the universe. It will obtain optical spectra for tens of millions of galaxies and quasars, constructing a 3-dimensional map spanning the nearby universe to 10 billion light years. The survey will be conducted on the Mayall 4-meter telescope at Kitt Peak National Observatory starting in 2018. In order to achieve these scientific goals the DESI collaboration is building a high throughput spectrograph capable of observing thousands of spectra simultaneously. In this presentation we discuss the DESI instrument control and data acquisition system that is currently being developed to operate the 5,000 fiber positioners in the focal plane, the 10 spectrographs each with three CDD cameras and every other aspect of the instrument. Special emphasis will be given to novel aspects of the design including the use of inexpensive Linux-based microcontrollers such as the Raspberry PI to control a number of DESI hardware components.   

Optimising gravitational wave searches for unknown isolated neutron stars

All-sky searches for gravitational waves from isolated neutron stars must be highly sensitive over a large parameter space. This requirement presents a significant computational challenge. Computing power is amassed with the support of the public via the Einstein@Home project. Semi-coherent search methods seek to maximise sensitivity to signals of unknown frequency over the whole sky. Parameter space coverage in each dimension, and coherent integration time at each point in parameter space, impact the search sensitivity and the computing requirements. The optimal search design is a trade-off among these elements, with the best choice depending on the amount of LIGO data and the area in parameter space to be covered. Here I present efforts to address the multi-dimensional challenge of optimising all-sky searches for isolated neutron stars with Einstein@Home.   

Discontinuous Galerkin Methods for Neutrino Radiation Transport

We are developing new computational methods for simulation of neutrino transport in core-collapse supernovae, which is challenging since neutrinos evolve from being diffusive in the proto-neutron star to nearly free streaming in the critical neutrino heating region. To this end, we consider conservative formulations of the Boltzmann equation,\footnote{Cardall, Endeve, \& Mezzacappa 2013, Phys. Rev. D {\bf 88}, 023011} and aim to develop robust, high-order accurate methods. Runge-Kutta discontinuous Galerkin (DG) methods,\footnote{Cockburn \& Shu 2001, J. Sci. Comput. {\bf 16}, 173-261} offer several attractive properties, including (i) high-order accuracy on a compact stencil and (ii) correct asymptotic behavior in the diffusion limit. We have recently developed a new DG method for the advection part for the transport solve,\footnote{Endeve, Hauck, Xing, \& Mezzacappa 2015 (arXiv:1410.7431)} which is high-order accurate and strictly preserves the physical bounds of the distribution function; i.e., $f\in[0,1]$. We summarize the main ingredients of our bound-preserving DG method and discuss ongoing work to include neutrino-matter interactions in the scheme.   

Monte Carlo Neutrino Transport in Post-Merger Disks

The merger of two neutron stars or a neutron star and a black hole are the prime candidate models for short-duration gamma ray bursts and production of r-process elements. Neutrinos can carry away energy and change the ratio of neutrons to protons, in turn affecting the appearance and dynamics of the burst and the types of elements formed from the outflow. We simulate Monte Carlo transport of neutrinos through the accretion disk surrounding the post-merger black hole or hyper-massive neutron star to explore the influence of neutrinos on the disk composition and temperature profile.   

Hot accretion flows onto binary and single black holes

Accreting black holes (BHs) are at the core of relativistic astrophysics as messengers of the strong-field regime of General Relativity and prime targets of several observational campaigns, including imaging the black hole shadow in SagA* and M87 with the Event Horizon Telescope. Binary Black Holes are one of the most promising gravitational wave sources for adLIGO and Pulsar Timing Arrays and -- if accreting -- can provide a strong electromagnetic counterpart. I will present results from global GRMHD simulations of both single and binary BHs embedded in a hot, magnetized disk, highlighting differences in their observational appearance including their gravitational and electromagnetic radiation.   

Examing the Rayleigh-Taylor instability in GRB outflow using wavelets

In the gamma ray burst afterglow model, a relativistic shell forms which is unstable to the Rayleigh-Taylor instability during the formation of a relativistic blastwave. This instability is likely observable, and understanding the resulting turbulence have motivated many previous studies into the Rayleigh-Taylor instability in the context of a relativistic fireball. We study the Rayleigh-Taylor instability in a relativistic gamma ray burst outflow with and without magnetic fields. We also examine the growth of the magnetic fields with different Lorentz factors. The simulations are performed using a wavelet based adaptive mesh refinement method.   

Inferences on Populations of Binary Neutron Stars

The aLIGO network stands to make hundreds of detections over the lifetime of the project. While there is much to be learned from the parameters of single events, the parameter distribution of the population of events is also of great interest for astrophysics, as this kind of parameter inference will help to develop gravitational-wave astronomy. The goal of this project is to develop the tools for estimating such population distributions and accounting for selection bias in such inferences. We will then apply the method to a simulated population of binary systems of neutron stars in order to estimate their mass distribution. We were able to create a technique that takes a set of data from aLIGO and runs it through a parameter estimation pipeline, taking into account selection bias effects. We start with a set of mass measurements, each measurement its own distribution due to noise in the detector. We draw these samples from a 2-dimensional distribution in chirp mass and symmetric mass ratio. We then use a Markov-Chain Monte Carlo method to estimate the parameters of the original distribution, as well as the rate of events.