Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userFast and Accurate Non-Linear Predictions of Universes with Deep Learning
108
0
0.0
(
0
)
Added by
Renan Alves De Oliveira
Publication date
2020
fields
Physics
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Cosmologists aim to model the evolution of initially low amplitude Gaussian density fluctuations into the highly non-linear cosmic web of galaxies and clusters. They aim to compare simulations of this structure formation process with observations of large-scale structure traced by galaxies and infer the properties of the dark energy and dark matter that make up 95% of the universe. These ensembles of simulations of billions of galaxies are computationally demanding, so that more efficient approaches to tracing the non-linear growth of structure are needed. We build a V-Net based model that transforms fast linear predictions into fully nonlinear predictions from numerical simulations. Our NN model learns to emulate the simulations down to small scales and is both faster and more accurate than the current state-of-the-art approximate methods. It also achieves comparable accuracy when tested on universes of significantly different cosmological parameters from the one used in training. This suggests that our model generalizes well beyond our training set.
rate research
Read More
The emergence of a complex, large-scale organisation of cosmic matter into the Cosmic Web is a beautiful exemplification of how complexity can be produced by simple initial conditions and simple physical laws. In the epoch of Big Data in astrophysics, connecting the stunning variety of multi-messenger observations to the complex interplay of fundamental physical processes is an open challenge. In this contribution, I discuss a few relevant applications of Information Theory to the task of objectively measuring the complexity of modern numerical simulations of the Universe. When applied to cosmological simulations, complexity analysis makes it possible to measure the total information necessary to model the cosmic web. It also allow us to monitor which physical processes are mostly responsible for the emergence of complex dynamical behaviour across cosmic epochs and environments, and possibly to improve mesh refinement strategies in the future.
To exploit the power of next-generation large-scale structure surveys, ensembles of numerical simulations are necessary to give accurate theoretical predictions of the statistics of observables. High-fidelity simulations come at a towering computational cost. Therefore, approximate but fast simulations, surrogates, are widely used to gain speed at the price of introducing model error. We propose a general method that exploits the correlation between simulations and surrogates to compute fast, reduced-variance statistics of large-scale structure observables without model error at the cost of only a few simulations. We call this approach Convergence Acceleration by Regression and Pooling (CARPool). In numerical experiments with intentionally minimal tuning, we apply CARPool to a handful of GADGET-III $N$-body simulations paired with surrogates computed using COmoving Lagrangian Acceleration (COLA). We find $sim 100$-fold variance reduction even in the non-linear regime, up to $k_mathrm{max} approx 1.2$ $h {rm Mpc^{-1}}$ for the matter power spectrum. CARPool realises similar improvements for the matter bispectrum. In the nearly linear regime CARPool attains far larger sample variance reductions. By comparing to the 15,000 simulations from the Quijote suite, we verify that the CARPool estimates are unbiased, as guaranteed by construction, even though the surrogate misses the simulation truth by up to $60%$ at high $k$. Furthermore, even with a fully configuration-space statistic like the non-linear matter density probability density function, CARPool achieves unbiased variance reduction factors of up to $sim 10$, without any further tuning. Conversely, CARPool can be used to remove model error from ensembles of fast surrogates by combining them with a few high-accuracy simulations.
The Helioseismic and Magnetic Imager (HMI) onboard NASAs Solar Dynamics Observatory (SDO) produces estimates of the photospheric magnetic field which are a critical input to many space weather modelling and forecasting systems. The magnetogram products produced by HMI and its analysis pipeline are the result of a per-pixel optimization that estimates solar atmospheric parameters and minimizes disagreement between a synthesized and observed Stokes vector. In this paper, we introduce a deep learning-based approach that can emulate the existing HMI pipeline results two orders of magnitude faster than the current pipeline algorithms. Our system is a U-Net trained on input Stokes vectors and their accompanying optimization-based VFISV
As a part of NASAs Heliophysics System Observatory (HSO) fleet of satellites,the Solar Dynamics Observatory (SDO) has continuously monitored the Sun since2010. Ultraviolet (UV) and Extreme UV (EUV) instruments in orbit, such asSDOs Atmospheric Imaging Assembly (AIA) instrument, suffer time-dependent degradation which reduces instrument sensitivity. Accurate calibration for (E)UV instruments currently depends on periodic sounding rockets, which are infrequent and not practical for heliophysics missions in deep space. In the present work, we develop a Convolutional Neural Network (CNN) that auto-calibrates SDO/AIA channels and corrects sensitivity degradation by exploiting spatial patterns in multi-wavelength observations to arrive at a self-calibration of (E)UV imaging instruments. Our results remove a major impediment to developing future HSOmissions of the same scientific caliber as SDO but in deep space, able to observe the Sun from more vantage points than just SDOs current geosynchronous orbit.This approach can be adopted to perform autocalibration of other imaging systems exhibiting similar forms of degradation
Traditional neuroimage analysis pipelines involve computationally intensive, time-consuming optimization steps, and thus, do not scale well to large cohort studies with thousands or tens of thousands of individuals. In this work we propose a fast and accurate deep learning based neuroimaging pipeline for the automated processing of structural human brain MRI scans, replicating FreeSurfers anatomical segmentation including surface reconstruction and cortical parcellation. To this end, we introduce an advanced deep learning architecture capable of whole brain segmentation into 95 classes. The network architecture incorporates local and global competition via competitive dense blocks and competitive skip pathways, as well as multi-slice information aggregation that specifically tailor network performance towards accurate segmentation of both cortical and sub-cortical structures. Further, we perform fast cortical surface reconstruction and thickness analysis by introducing a spectral spherical embedding and by directly mapping the cortical labels from the image to the surface. This approach provides a full FreeSurfer alternative for volumetric analysis (in under 1 minute) and surface-based thickness analysis (within only around 1h runtime). For sustainability of this approach we perform extensive validation: we assert high segmentation accuracy on several unseen datasets, measure generalizability and demonstrate increased test-retest reliability, and high sensitivity to group differences in dementia.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
