分类： 天文学 >> 天文学 提交时间： 2023-02-19
摘要： The potential of deep learning based image-to-image translations has recently drawn a lot of attention; one intriguing possibility is that of generating cosmological predictions with a drastic reduction in computational cost. Such an effort requires optimization of neural networks with loss functions beyond low-order statistics like pixel-wise mean square error, and validation of results beyond simple visual comparisons and summary statistics. In order to study learning-based cosmological mappings, we choose a tractable analytical prescription - the Zel'dovich approximation - modeled using U-Net, a convolutional image translation framework. A comprehensive list of metrics is proposed, including higher-order correlation functions, conservation laws, topological indicators, dynamical robustness, and statistical independence of density fields. We find that the U-Net approach does well with some metrics but has difficulties with others. In addition to validating AI approaches using rigorous physical benchmarks, this study motivates advancements in domain-specific optimization schemes for scientific machine learning.
分类： 天文学 >> 天文学 提交时间： 2023-02-19
Sandeep Madireddy
Nesar Ramachandra
Nan Li
James Butler
Prasanna Balaprakash
Salman Habib
Katrin Heitmann
The LSST Dark Energy Science Collaboration
摘要： Upcoming large astronomical surveys are expected to capture an unprecedented number of strong gravitational lensing systems. Deep learning is emerging as a promising practical tool for the detection and quantification of these galaxy-scale image distortions. The absence of large quantities of representative data from current astronomical surveys motivates the development of a robust forward-modeling approach using synthetic lensing images. Using a mock sample of strong lenses created upon a state-of-the-art extragalactic catalogs, we train a modular deep learning pipeline for uncertainty-quantified detection and modeling with intermediate image processing components for denoising and deblending the lensing systems. We demonstrate a high degree of interpretability and controlled systematics due to domain-specific task modules trained with different stages of synthetic image generation. For lens detection and modeling, we obtain semantically meaningful latent spaces that separate classes of strong lens images and yield uncertainty estimates that explain the origin of misclassified images and provide probabilistic predictions for the lens parameters. Validation of the inference pipeline has been carried out using images from the Subaru telescope's Hyper Suprime-Cam camera, and LSST DESC simulated DC2 sky survey catalogues.
分类： 天文学 >> 天文学 提交时间： 2023-02-19
Juliana Kwan
Shun Saito
Alexie Leauthaud
Katrin Heitmann
Salman Habib
Nicholas Frontiere
Hong Guo
Song Huang
Adrian Pope
Sergio Rodríguez-Torres
摘要： We construct accurate emulators for the projected and redshift space galaxy correlation functions and excess surface density as measured by galaxy-galaxy lensing, based on Halo Occupation Distribution (HOD) modeling. Using the complete Mira-Titan suite of 111 $N$-body simulations, our emulators vary over eight cosmological parameters and include the effects of neutrino mass and dynamical dark energy. We demonstrate that our emulators are sufficiently accurate for the analysis of the BOSS DR12 CMASS galaxy sample over the range 0.5 < r < 50 Mpc/h. Furthermore, we show that our emulators are capable of recovering unbiased cosmological constraints from realistic mock catalogs over the same range. Our mock catalog tests show the efficacy of combining small scale galaxy-galaxy lensing with redshift space clustering and that we can constrain the growth rate and \sigma_8 to 7% and 4.5% respectively for a CMASS-like sample using only the measurements covered by our emulator. With the inclusion of a CMB prior on H_0, this reduces to a 2% measurement on the growth rate.
点击量
待评议
点击量
下载量
评论
