Cosmic Velocity Field Reconstruction Using AI
نویسندگان
چکیده
Abstract We develop a deep-learning technique to infer the nonlinear velocity field from dark matter density field. The architecture we use is “U-net” style convolutional neural network, which consists of 15 convolution layers and 2 deconvolution layers. This setup maps three-dimensional 32 3 voxels or momentum fields 20 voxels. Through analysis simulation with resolution h ?1 Mpc, find that network can predict nonlinearity, complexity, vorticity fields, as well power spectra their value, divergence, its prediction accuracy reaches range k ? 1.4 Mpc relative error ranging 1% ?10%. A simple comparison shows networks may have an overwhelming advantage over perturbation theory in reconstruction fields.
منابع مشابه
Three-dimensional velocity field reconstruction.
The problem of inter-slice magnetic resonance (MR) image reconstruction is encountered often in medical imaging applications, in such scenarios, there is a need to approximate information not captured in contiguously acquired MR images due to hardware sampling limitations. In the context of velocity field reconstruction, these data are required for visualization and computational analyses of fl...
متن کاملSample variance of the cosmic velocity field
Since the cosmic peculiar velocity field depends on small wave-number modes strongly, we cannot probe its universal properties unless we observe a sufficiently large region. We calculate the expected deviation (sample variance) of the peculiar velocity dispersion from its universal value in the case observed volume is finite. Using linear theory we show that the sample variance remains as large...
متن کاملReconstruction of source and cosmic magnetic field characteristics from clusters of ultra-high energy cosmic rays
We present a detailed Monte Carlo study coupled to a likelihood analysis of the potential of next generation ultra-high energy cosmic ray experiments to reconstruct properties of the sources and the extra-galactic magnetic field. Such characteristics are encoded in the distributions of arrival time, direction, and energy of clusters of charged cosmic rays above a few 1019 eV. The parameters we ...
متن کاملOrientation of cosmic web filaments with respect to the underlying velocity field
The large-scale structure of the Universe is characterized by a web-like structure made of voids, sheets, filaments and knots. The structure of this so-called cosmic web is dictated by the local velocity shear tensor. In particular, the local direction of a filament should be strongly aligned with ê3, the eigenvector associated with the smallest eigenvalue of the tensor. That conjecture is test...
متن کاملUnmanned aerial vehicle field sampling and antenna pattern reconstruction using Bayesian compressed sensing
Antenna 3D pattern measurement can be a tedious and time consuming task even for antennas with manageable sizes inside anechoic chambers. Performing onsite measurements by scanning the whole 4π [sr] solid angle around the antenna under test (AUT) is more complicated. In this paper, with the aim of minimum duration of flight, a test scenario using unmanned aerial vehicles (UAV) is proposed. A pr...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: The Astrophysical Journal
سال: 2021
ISSN: ['2041-8213', '2041-8205']
DOI: https://doi.org/10.3847/1538-4357/abf3bb