Search for: All records

Abstract Massively multiplexed spectrographs will soon gather large statistical samples of stellar spectra. The accurate estimation of uncertainties on derived parameters, such as the line-of-sight velocityv_los, especially for spectra with low signal-to-noise ratios (S/Ns), is paramount. We generated an ensemble of simulated optical spectra of stars as if they were observed with low- and medium-resolution fiber-fed instruments on an 8 m class telescope, similar to the Subaru Prime Focus Spectrograph, and determinedv_losby fitting stellar templates to the simulated spectra. We compared the empirical errors of the derived parameters—calculated from an ensemble of simulations—to the asymptotic errors determined from the Fisher matrix, as well as from Monte Carlo sampling of the posterior probability. We confirm that the uncertainty ofv_losscales with the inverse square root of the S/N, but also show how this scaling breaks down at low S/N and analyze the error and bias caused by template mismatch. We outline a computationally optimized algorithm to fit multiexposure data and provide a mathematical model of stellar spectrum fitting that maximizes the so called significance, which allows for calculating the error from the Fisher matrix analytically. We also introduce the effective line count, and provide a scaling relation to estimate the errors ofv_losmeasurements based on stellar type. Our analysis covers a range of stellar types with parameters that are typical of the Galactic outer disk and halo, together with analogs of stars in M31 and in satellite dwarf spheroidal galaxies around the Milky Way. 

Abstract We propose a novel exact method to solve the probabilistic catalog matching problem faster than previously possible. Our new approach uses mixed integer programming and introduces quadratic constraints to shrink the problem by multiple orders of magnitude. We also provide a method to use a feasible solution to dramatically speed up our algorithm. This gain in performance is dependent on how close to optimal the feasible solution is. Also, we are able to provide good solutions by stopping our mixed integer programming solver early. Using simulated catalogs, we empirically show that our new mixed integer program with quadratic constraints is able to be set up and solved much faster than previous large linear formulations. We also demonstrate our new approach on real-world data from the Hubble Source Catalog. This paper is accompanied by publicly available software to demonstrate the proposed method. 

Abstract Building on previous Bayesian approaches, we introduce a novel formulation of probabilistic cross-identification, where detections are directly associated to (hypothesized) astronomical objects in a globally optimal way. We show that this new method scales better for processing multiple catalogs than enumerating all possible candidates, especially in the limit of crowded fields, which is the most challenging observational regime for new-generation astronomy experiments such as the Rubin Observatory Legacy Survey of Space and Time. Here we study simulated catalogs where the ground truth is known and report on the statistical and computational performance of the method. The paper is accompanied by a public software tool to perform globally optimal catalog matching based on directional data. 

The paper introduces DeepForge, a gateway to deep learning for scientific computing. DeepForge provides an easy to use, yet powerful visual/textual interface to facilitate the rapid development of deep learning models by novices as well as experts. Utilizing a cloud-based infrastructure, built-in version control, and multiuser collaboration support, DeepForge promotes reproducibility and ease of access and enables remote execution of machine learning pipelines. The tool currently supports TensorFlow/Keras, but its extensible architecture enables easy integration of additional platforms.