اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدFast reconstruction of single-shot wide-angle diffraction images through deep learning
200
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Single-shot X-ray imaging of short-lived nanostructures such as clusters and nanoparticles near a phase transition or non-crystalizing objects such as large proteins and viruses is currently the most elegant method for characterizing their structure. Using hard X-ray radiation provides scattering images that encode two-dimensional projections, which can be combined to identify the full three-dimensional object structure from multiple identical samples. Wide-angle scattering using XUV or soft X-rays, despite yielding lower resolution, provides three-dimensional structural information in a single shot and has opened routes towards the characterization of non-reproducible objects in the gas phase. The retrieval of the structural information contained in wide-angle scattering images is highly non-trivial, and currently no efficient rigorous algorithm is known. Here we show that deep learning networks, trained with simulated scattering data, allow for fast and accurate reconstruction of shape and orientation of nanoparticles from experimental images. The gain in speed compared to conventional retrieval techniques opens the route for automated structure reconstruction algorithms capable of real-time discrimination and pre-identification of nanostructures in scattering experiments with high repetition rate -- thus representing the enabling technology for fast femtosecond nanocrystallography.
قيم البحث
اقرأ أيضاً
Resonant transmission of light is a surface-wave assisted phenomenon that enables funneling light through subwavelength apertures milled in otherwise opaque metallic screens. In this work, we introduce a deep learning approach to efficiently compute
and design the optical response of a single subwavelength slit perforated in a metallic screen and surrounded by periodic arrangements of indentations. First, we show that a semi-analytical framework based on a coupled-mode theory formalism is a robust and efficient method to generate the large training datasets required in the proposed approach. Second, we discuss how simple, densely connected artificial neural networks can accurately learn the mapping from the geometrical parameters defining the topology of the system to its corresponding transmission spectrum. Finally, we report on a deep learning tandem architecture able to perform inverse design tasks for the considered class of systems. We expect this work to stimulate further work on the application of deep learning to the analysis of light-matter interaction in nanostructured metallic films.
We present single-shot electron velocity-map images of nanoplasmas generated from doped helium nanodroplets and neon clusters by intense near-infrared and mid-infrared laser pulses. We report a large variety of signal types, most crucially depending
on the cluster size. The common feature is a two-component distribution for each single-cluster event: A bright inner part with nearly circular shape corresponding to electron energies up to a few eV, surrounded by an extended background of more energetic electrons. The total counts and energy of the electrons in the inner part are strongly correlated and follow a simple power-law dependence. Deviations from the circular shape of the inner electrons observed for neon clusters and large helium nanodroplets indicate non-spherical shapes of the neutral clusters. The dependence of the measured electron energies on the extraction voltage of the spectrometer indicates that the evolution of the nanoplasma is significantly affected by the presence of an external electric field. This conjecture is confirmed by molecular dynamics simulations, which reproduce the salient features of the experimental electron spectra.
Despite the tremendous advances made by the ab initio theory of electronic structure of atoms and molecules, its applications are still not possible for very large systems. Therefore, semi-empirical model Hamiltonians based on the zero-differential o
verlap (ZDO) approach such as the Pariser-Parr-Pople, CNDO, INDO, etc. provide attractive, and computationally tractable, alternatives to the ab initio treatment of large systems. In this paper we describe a Fortran 90 computer program developed by us, that uses CNDO/2 and INDO methods to solve Hartree-Fock(HF) equation for molecular systems. The INDO method can be used for the molecules containing the first-row atoms, while the CNDO/2 method is applicable to those containing both the first-, and the second-row, atoms. We have paid particular attention to computational efficiency while developing the code, and, therefore, it allows us to perform calculations on large molecules such as C_60 on small computers within a matter of seconds. Besides being able to compute the molecular orbitals and total energies, our code is also able to compute properties such as the electric dipole moment, Mulliken population analysis, and linear optical absorption spectrum of the system. We also demonstrate how the program can be used to compute the total energy per unit cell of a polymer. The applications presented in this paper include small organic and inorganic molecules, fullerene C_60, and model polymeric systems, viz., chains containing alternating boron and nitrogen atoms (BN chain), and carbon atoms (C chain).
An interatomic potential for Al-Tb alloy around the composition of Al90Tb10 was developed using the deep neural network (DNN) learning method. The atomic configurations and the corresponding total potential energies and forces on each atom obtained f
rom ab initio molecular dynamics (AIMD) simulations are collected to train a DNN model to construct the interatomic potential for Al-Tb alloy. We show the obtained DNN model can well reproduce the energies and forces calculated by AIMD. Molecular dynamics (MD) simulations using the DNN interatomic potential also accurately describe the structural properties of Al90Tb10 liquid, such as the partial pair correlation functions (PPCFs) and the bond angle distributions, in comparison with the results from AIMD. Furthermore, the developed DNN interatomic potential predicts the formation energies of crystalline phases of Al-Tb system with the accuracy comparable to ab initio calculations. The structure factor of Al90Tb10 metallic glass obtained by MD simulation using the developed DNN interatomic potential is also in good agreement with the experimental X-ray diffraction data.
The fission of highly charged sodium clusters with fissilities X>1 is studied by {em ab initio} molecular dynamics. Na_{24}^{4+} is found to undergo predominantly sequential Na_{3}^{+} emission on a time scale of 1 ps, while Na_{24}^{Q+} (5 leq Q leq
8) undergoes multifragmentation on a time scale geq 0.1 ps, with Na^{+} increasingly the dominant fragment as Q increases. All singly-charged fragments Na_{n}^{+} up to size n=6 are observed. The observed fragment spectrum is, within statistical error, independent of the temperature T of the parent cluster for T leq 1500 K. These findings are consistent with and explain recent trends observed experimentally.
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
