A universal deep learning-based framework towards fully ab initio simulation of atmospheric aerosol nucleation

Atmospheric aerosol nucleation contributes to around half of cloud condensation nuclei globally. Despite the importance for climate, detailed nucleation mechanisms are still poorly understood. Understanding aerosol nucleation dynamics is hindered by non-reactivity of force fields and high computational costs due to rare event nature of aerosol nucleation. Developing reactive force fields for nucleation systems are even more challenging than covalently bonded materials because of wide size range and high dimensional characteristics of non-covalent hydrogen bonding bridging clusters. Here we proposes a system transferable framework to train an accurate reactive force field (FF) based on deep neural network (DNN) and further bridges the DNN-FF based molecular dynamics (MD) with cluster kinetics model based on Poisson distributions of reactive events to overcome high computational costs from direct MD. We found that previously reported acid-base formation rates tend to be underestimated several times, emphasizing acid-base nucleation observed in multiple environments should be revisited.

JAX-ReaxFF: A Gradient Based Framework for Extremely Fast Optimization of Reactive Force Fields

Molecular dynamics (MD) simulations facilitate the study of physical and chemical processes of interest. Traditional classical MD models lack reactivity to explore several important phenomena; while quantum mechanical (QM) models can be used for this purpose, they come with steep computational costs. The reactive force field (ReaxFF) model bridges the gap between these approaches by incorporating dynamic bonding and polarizability. To achieve realistic simulations using ReaxFF, model parameters must be optimized against high fidelity training data, typically with QM accuracy. Existing parameter optimization methods for ReaxFF consist of black-box techniques using genetic algorithms or Monte-Carlo methods. Due to the stochastic behavior of these methods, the optimization process can require millions of error evaluations for complex parameter fitting tasks, significantly hampering the rapid development of high quality parameter sets. In this work, we present JAX ReaxFF, a novel software tool that leverages modern machine learning infrastructure to enable extremely fast optimization of ReaxFF parameters. By calculating gradients of the loss function using the JAX library, we are able to utilize highly effective local optimization methods, such as the limited Broyden–Fletcher–Goldfarb–Shanno (LBFGS) and Sequential Least Squares Programming (SLSQP) methods. As a result of the performance portability of JAX, JAX-ReaxFF can execute efficiently on multi-core CPUs, GPUs (or even TPUs). By leveraging the gradient information and modern hardware accelerators, we are able to decrease parameter optimization time for ReaxFF from days to mere minutes. JAX-ReaxFF framework can also serve as a sandbox environment for domain scientists to explore customizing the ReaxFF functional form for more accurate modeling.

Development and Application of ReaxFF Methodology for Understanding the Chemical Dynamics of Metal Carbonates in Aqueous Solutions

A new ReaxFF reactive force field has been developed for metal carbonate systems including Na+, Ca2+, and Mg2+ cations and the CO32- anion. This force field is fully transferable with...

Hydraulic calculation of the fire extinguishing system of port facilities

Цель статьи – разработать алгоритм расчёта гидравлических характеристик системы пожаротушения для портовых сооружений. В системе пожаротушения имеется погружной насос, трубопроводы, предназначенные для транспортировки воды от места ее забора, до места возможного возгорания. Использован насос UGP-M-1210-04. Задачи исследования: анализ результатов испытаний; получение эмпирических зависимостей показателей работы насоса от его производительности, исследования работы насоса в сети. Проведены расчеты скорости истекающей струи и полной реактивной силы в рабочей точке насосной установки. Исследовано влияние скорости на реактивную силу. Анализ результатов показывает, что точность расчетов приемлема для инженерных расчетов. Полученный алгоритм расчета может быть использован при проектировании систем тушения пожаров в портовых сооружениях. The purpose of the article is to develop an algorithm for calculating the hydraulic characteristics of a fire extinguishing system for port facilities. The fire extinguishing system has a submersible pump, pipelines intended for transporting water from the place of its intake to the place of possible ignition. The UGP-M-1210-04 pump is used. Research objectives: analysis of test results; obtaining empirical dependencies of pump performance indicators on its performance, research of pump operation in the network. Calculations of the velocity of the expiring jet and the total reactive force at the working point of the pumping unit are carried out. The effect of velocity on the reactive force is investigated. Analysis of the results shows that the accuracy of calculations is acceptable for engineering calculations. The resulting calculation algorithm can be used in the design of fire extinguishing systems in port facilities.

About possibility of reduction weight and dimensions of dock’s rack of ship’s cross bulkhead

Для восприятия значительной части реакции килевой дорожки при доковании корабля в конструкцию поперечных переборок вводится центральная (доковая) стойка, размеры поперечного сечения которой выбираются из условия ее прочности и устойчивости. В традиционных методиках необходимая площадь поперечного сечения доковой стойки определяется без учета работы обшивки переборки и ближайших к диаметральной плоскости вертикальных стоек, что приводит к значительным размерам и весу доковой стойки. Кроме того, сжимающее усилие в стойке в методиках полагается линейно убывающим от максимального значения у днища до нулевого у палубы. В данной работе численно исследуется взаимодействие конструктивных элементов плоской поперечной переборки корабля при его доковании. Целью исследования является выяснение степени участия обшивки переборки и ближайших к доковой стойке вертикальных стоек в восприятии реакции килевой дорожки и уточнение характера распределения этой реакции по высоте доковой и вертикальных стоек. На первом этапе исследования авторами была разработана более простая плоская конечно-элементная модель собственно переборки. На втором этапе использовалась также разработанная авторами пространственная конечно-элементная модель части корпуса корабля, включающая, кроме самой поперечной переборки, прилегающие к ней палубы, борта и днище. Результаты, полученные по обеим моделям, показывают, что сжимающие напряжения в переборке локализуются в нижней центральной ее части, а реактивное усилие от килевой дорожки распределяется между доковой стойкой, ближайшими к ней вертикальными стойками и обшивкой переборки. Причем по высоте доковой стойки реактивное усилие уменьшается от днища к палубе не линейно, а быстрее, что приводит к увеличению критической нагрузки стойки при прочих равных условиях. Установлено, что при обеспечении устойчивости соседних с доковой вертикальных стоек эффективная площадь обшивки переборки и вертикальных стоек, воспринимающая реакцию килевой дорожки вместе с доковой стойкой, может составить более 50% от площади профиля доковой стойки. To perceive a significant part of the reaction of the keel track when docking a ship, a central (dock) rack is introduced into the structure of transverse bulkheads, the cross-sectional dimensions of which are chosen on the basis of its strength and stability. In classic methods, the required cross-sectional area of the dock rack is determined without taking into account the work of the bulkhead skin and the vertical racks closest to the diametrical plane, which leads to significant size and weight of the dock rack. In addition, the compression force in the rack in the methods is assumed to decrease linearly from the maximum value at the bottom to zero at the deck. In this paper, the interaction of structural elements of a ship's flat transverse bulkhead during its docking is numerically investigated. The purpose of the study is to clarify the degree of participation of the bulkhead skin and the vertical racks closest to the dock stand in the perception of the keel track reaction and to clarify the nature of the distribution of this reaction along the height of the dock and vertical racks. At the first stage of the study, the authors developed a simpler flat finite-element model of the bulkhead itself. At the second stage, the space finite-element model of the ship's hull part developed by the authors was also used, including, in addition to the most transverse bulkhead, the decks, sides and bottom adjacent to it. The results obtained from both models show that the compressive stresses in the bulkhead are localized in its lower central part, and the reactive force from the keel track is distributed between the dock rack, the nearest vertical racks and the bulkhead skin. Moreover, by the height of the dock rack closest to it, the reactive force decreases from the bottom to the deck not linearly, but faster, which leads to an increase in the critical load of the rack, all other things being equal. It has been established that with ensuring the stability of the vertical racks adjacent to the dock, the effective area of the bulkhead skin and vertical racks, which perceives the reaction of the keel track together with the dock rack, can contain more than 50% of the required area of the dock rack’s profile.

Nanostructures Failures and Fully Atomistic Molecular Dynamics Simulations

Nowadays, the concern about the limitations of space and natural resources has driven the motivation for the development of increasingly smaller, more efficient, and energy-saving electromechanical devices. Since the revolution of “microchips”, during the second half of the twentieth century, besides the production of microcomputers, it has been possible to develop new technologies in the areas of mechanization, transportation, telecommunications, among others. However, much room for significant improvements in factors as shorter computational processing time, lower energy consumption in the same kind of work, more efficiency in energy storage, more reliable sensors, and better miniaturization of electronic devices. In particular, nanotechnology based on carbon has received continuous attention in the world’s scientific scenario. The riches found in different physical properties of the nanostructures as, carbon nanotubes (CNTs), graphene, and other exotic allotropic forms deriving from carbon. Thus, through classical molecular dynamics (CMD) methods with the use of reactive interatomic potentials reactive force field (ReaxFF), the scientific research conducted through this chapter aims to study the nanostructural, dynamic and elastic properties of nanostructured systems such as graphene single layer and conventional carbon nanotube (CNTs).