Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Tianhe2-JK Time Allocation Application –
Computational study on high-speed disperse flows
The aims of this proposed project are dual and mutually dependent. One aim is to advance the
understanding of the basic physics that control the multiphase interactions in high speed,
non-dilute disperse flows. The other aim is to advance computational methods and computer
implementations of such methods so as to achieve practical direct numerical simulations necessary
for meeting the first aim. We have done already the ground work towards both aims, so our
research plans are well defined and the work involves a minimal amount of risk.
On the first aim, we should mention that high-speed disperse flows are not only one of the
cornerstones of advanced technologies, they are also at the essence of the science of fluid
mechanics since the averaging process of the field equations yields an ill-posed problem, which
despite enormous efforts over the past 50 years remains unresolved. On the second aim, as we
explain below, and as evidenced by the meager results available in the literature so far, the need
for practical simulations of high fidelity, involving no modelling assumptions continues to be great.
The computational demands are great because of the essential threedimensionality of the flow on
the one hand, and the tremendous resolution requirements that high-speed flows (thin boundary
layers) entail. This proposal sets us to operate at the intersection of these two major challenges.
To better appreciate the scope of the first aim, we need to mention recent experiments
(Theofanous et al. [1], our long-term collaborators in the USA) involving the interaction of shock
waves with a curtain of particles made to occupy the full cross-sectional area of the channel of a
shock tube. They found that the curtain expands rapidly in an explosive like manner and at
accelerations that reach up to 10,000 g. Efforts made by these authors to compute this behavior
using an effective field model (Lhuillier et al. [4]) failed by under predicting the accelerated
expansion by about 40%. Another work by Wagner et al. [2], involving particles and curtain
thickness that were an order of magnitude smaller, conformed to the scaling approach of
Theofanous et al. [1], but again the effective field approach failed by the same 40%. Ling et al. [3]
presented results by a point-particle method, and claimed excellent agreement, but it is trivial to
see a major error in their formulation that led to an over prediction of the added mass and transient
viscous force totaling 40%. It is also trivial to see that both of these forces are negligibly small for
the conditions of both tests. Clearly then there are new physics here that may impact even the
whole classical idea of averaging! To better appreciate the scope of the second aim, we note that
current experience with DNS of disperse flows has been limited to a handful of studies. Most of
them are for incompressible flow with Reynolds numbers only up to 180, and resolutions only up
to 13 cells per particle [5, 6]. DNS simulations of compressible disperse flows are limited in 2D
with at most 100 stationary spheres [8, 9]. To our knowledge, there is no 3D DNS for
compressible flow so far. Our present code for compressible 3D flows was demonstrated can
handle a few hundreds of particles, and to produce promising results relative to the experiments
mentioned above (as we explain below). The aims of the present work is to extend the capability
to 2000 particles, Re of up to 10,000, and resolutions of 40 cells per particle for a total of 400
million cells, and using the results to get an appreciation of the forces on individual particles and
how they are affected by the collective effect of nearby particles (the concentration of particles).
Application for computing time
Applicant: Yang Ding
The proposed research is to explore the principles of the mechanics and control of
biological locomotion, and apply these principles to more advanced bionic machineries and
bio-robots. To compute the dynamics of fish swimming, high performance computational
fluid dynamics simulation is an essential tool.
Different animals show diversity in undulatory motion, such as the amplitude of the
body undulation, speed, body shape, and maneuverability. Such diversity can also provide
inspirations for the design of robots. We will seek for the laws in the variation, and the
relation between the body shape, the internal structure, the muscle activity, and the
mechanical environment. The results will contribute to our understanding of each component
of locomotion and how these components coordinate. Specifically, we will try to explain the
diverse muscle activation pattern during fish swimming with realistic 3D simulation of
various fishes.
Computational studies of plasmons and excitons in nanomaterials and their energy
applications
Principle Investigator: Shiwu Gao
Participantes: Fei Gao, Chao Zheng, Yi Yang, Saranya G. Jinling Li
Materials and Energy Division, CSRC
This project focuses on the theoretical and computational study of electronic excitations, plasmons
and excitons in particular, at surfaces and interfaces of nanomaterials. Such excitations play
important roles in many material properties and photochemical phenomena at surfaces including
sensing, photovoltaics, solar cells, photocatalysis, and optoelectronic devices. Theoretical
understanding on the formation and decay of these excitations is essential to gain insights into the
nature and dynamics of such excited states and to the design and optimization of materials and
device applications. Combined with analytical models and semiclassical theories, we will conduct
large-scale numerical computations for nanostructured materials and on well-defined systems
using home-developed and open source codes (octopus of TDDFT, BerkeleyGW package), so that
quantitative results can be achieved and comparisons with spectroscopy and measurement can be
made. Calculations of electronic excitations in nanostructures involve large-scale and massively
parallel computations, which usually requires an order of magnitude more computational efforts
and resources than the ground state DFT calculations. This is in part due to the nature and
complexity of surface plasmons and interface excitons, which involve long-range interactions
between many electron-hole pairs, and in part due to the broking symmetry of molecule-surface
systems, so that a large number of atoms (hundreds to thousands of atoms) have to be treated.
Here in this project, first-principles density functional theory will be combined with a
semiclassical theory of plasmon-electron coupling to describe these processes in order to gain
insight into nanoscale light-matter interaction, plasmon and exciton induced electron and energy
transfer processes.
Tianhe2-JK Time Allocation Application
Ab initio study of thermodynamically consistent physical properties of warm dense
plasma
A computational resources of 2000 CPU core for 1000 hours (2,000,000 core hours) is
applied to study the physical properties of warm dense matter. Matter at extreme conditions (i.e.
ultra-high temperatures, up to several million Kelvins, and density, up to several times normal
solid density) is also referred to as “warm dense matter”.
Understanding the behavior of matter at extreme conditions is a major challenge and of critical
importance for a number of scientific fields ranging from astrophysics to the inertial confinement
fusion. The temperature, pressure, and density of the matter under such extreme conditions often
vary several orders of magnitude. Therefore, experimentally, realizing controllable systems at
extreme conditions and probing these systems are understandably extremely difficult. Accurate
theoretical calculations and predictions also face enormous challenge both in terms of constructing
reliable model and in terms of the required computational resources.
In 2014, Sandia National Laboratories of the US carried out a benchmark experiment on the
opacity of warm dense iron, with Te~2,00,000K and solid density. In the high density region, the
measured wavelength dependent opacity is 30–400 per cent higher than predicted. Since iron
accounts for a quarter of the total opacity of the Sun, such discrepancy will lead to reductions of
30–50 per cent in the inferred amounts of carbon, nitrogen and oxygen in the Sun based on
standard solar models. Therefore, some new theoretical methods which can treat these many-body
interactions more adequately need to be developed.
In this study, density functional theory (DFT) based first-principles electronic structure
methods, which have successfully applied for many-body system under conventional conditions,
are applied to the study of the EOS and opacities of warm dense matter. In this study, we will
concentrate on the iron opacity problem, where the calculations will be carried out using the DFT
based method in the similar conditions of the Sandia’s new experiment. Because such kind of
calculation can take into account the many-body interactions in the warm dense matter, the results
can serve as a benchmark to the average atom model as well as experimental measurements,
which should be crucial to astrophysics and ICF studies.
Investigation of Spin-dependent Energy Conversion Processes in Organic
Photovoltaic and Lighting-emitting Devices
Over the past decade, organic electronics have been attracting a great deal of attentions.
In comparison with their inorganic counterparts, organic devices have added advantages
of low cost, easy fabrication, and mechanical flexibility. While great progress has been
made in OPV solar cells and OLEDs, their efficiencies are still far below their
thermodynamic limits. A deeper fundamental understanding is critically needed to
generate breakthroughs that will lead to practical application of OPV solar cells and
OLED solid-state lighting. One distinct difference between the energy conversion
processes, either from light to electrical energy or from electrical to light energy, in
organic versus inorganic semiconducting materials lies in “spin dependency”. In
inorganic materials, energy conversion is mediated through creation/annhilation of
unbound electrons and/or holes (charge carriers), which is spin independent; while in
organic materials, energy conversion, mediated through creation/annhilation of bound
electron-hole pairs (excitons), is spin dependent. Specially, photogeneration of charges in
OPV solar cells is a complex, only partially understood process that involves multiple
steps, which can all be spin dependent. It starts with the absorption of a photon to form a
mobile excited state, or exciton. The exciton then diffuses to an interface between donor
(D) and acceptor (A) domains where it dissociates into a charge transfer (CT) state,
which may be regarded as a tightly bound polaron pair (PP) across the D-A interface.
Next, the CT state undergoes a multistep process to dissociate into free charges (i.e., the
charge separated (CS) state) that ultimately are collected at the electrodes. This complex
reaction and its dynamics is strongly governed by the spin multiplicity of the excited state
(e.g., singlet vs. triplet, having different life time), film morphology at the junction of D
and A, and the energetics of both the bound CT and CS states. Because of the number of
steps involved, there are many routes that an exciton and subsequent products may take
following its generation that can lead to dissipative reactions that degrade the efficiency
of the OPV cell, or alternatively to reactions that may enhance the cell efficiency.
Similarly, the efficiency of OLED device, governed by the rate of exciton recombination,
depends also on the spin multiplicity of excitons.
The mission of this proposal is to provide a significantly deeper understanding of the
energy conversion processes, in particular their spin dependency, in polymeric and
molecular OPVs and WOLEDs through a concerted research effort using spectroscopy,
morphological and electrical characterization, and advanced theoretical and
computational methods. The proposed research will be at the scientific forefront to
address one of the “grand challenges” identified for basic energy science in “How do we
control material processes at the level of electrons?” and the basic energy research needs
in “Solar Energy Utilization and Solid-State Lighting”. Our three-year scientific research
goal is to understand the dynamics of “spin-dependent” electronic processes at the donor-
acceptor (D-A) interfaces in OPV and use it for enhancing solar power conversion
efficiency; as well as singlet-triplet energy flow in WOLEDs and use it for enhancing
electrical power conversion efficiency.
Tianhe2-JK Time Allocation Application
Advanced Functional Materials and Green Energy
Leader: Prof. Jiang Jing(姜晶,中心教授),Leo Lau
Members: Wen-Jin Yin, Ting-Cha Wei, Yu-Xuan Wu, and Jian Wu
Application:3.5 million core hours, or 400 cores for for 8640 hours
slim (the code can be run without cuda or mic)
It is great desire to reduce both the emission and accumulation of CO2
in the atmosphere as is the main greenhouse and ocean acidification gas.
The major source of CO2 emission is the traditional fossil fuel-fired
plants. Apart from the natural photosynthesis, various strategies have
been proposed to mitigate CO2 emission including carbon capture, energy
conservation, and energy storage through the chemical approaches.
Photo-catalytic reduction of CO2 is an efficient way to convert CO2 into
synthetic fuels or other useful chemicals by harnessing the renewable
solar energy. Titanium dioxide (TiO2) is a prototype photo-catalyst for
water splitting, and degradation of organic contaminants, because it is
highly stable, nontoxic and cheap. A series of experiments for
photo-catalytic reduction or fixation of CO2 into fuels have been
performed on the TiO2-based materials. The early experiment was
proposed by Inoue et al., who reported that photo-catalytic reduction of
CO2 in an aqueous suspension of Titania powder can form formaldehyde
(HCHO), formic acid (HCOOH), methanol (CH3OH), and methane (CH4)
as main products.5 Although CO2 can be successfully converted or fixated
through photo-catalytic reduction, both the efficiency and selectivity of
photo-catalytic system are extremely low and poor. In order to design
more efficient and selective photo-catalyst, many related works have been
carried out on the different phases of TiO2.
Tianhe2-JK Time Allocation Application
Leader: Leo Lau(刘焕明)
Postdocs and students: Da Wang, Zhen-Kun Tang, Xi-Bo Li, Jian Wu and
Wei-Wei Liu
The total resource application:600 cores for 8400 hours (1 year)
Advanced Functional Materials and Green Energy
With the rapid development of electronic market, energy storage for renewable
energy sources and hybrid electric vehicles is a pressing technological challenge.
Lithium ion batteries (LIBs) have been the subject of intense investigations due to
their good cycling performance, high storage capacity and high energy density. In
order to meet the requirements for the high power tools and electric vehicles,
electrode materials with high Li reversible storage capacity and fast Li and electron
transport are needed for lithium ion batteries. After graphene became experimentally
accessible in 2004, the layered materials like graphite have attracted great attention
mainly due to their unique physical properties and capability to fulfill the demands of
future nanoelectronic industry on adaptability, flexibility, and multi-functionality.
Although their intensive investigations, excellent anodes with good electrical
conductivity and high reversible lithium storage are still under development. In our
studies, we intend to systematically investigate the electrochemical properties of
a number of layered materials, such as germanium, silicon, black phosphorus,
transition oxides and transition metal dichalcogenides (TMD), etc., by means of
density functional theory (DFT) calculations. Our works propose to provide valuable
insights into exploring new types of high-capacity layered materials for potential
battery applications.
Proposal for TianHe-‐JK computing time Haiguang Liu et al. Algorithm/Complex System
division
Mapping experimental data to conformational space and recovering free energy landscape
using cryoEM-‐MDSimulation hybrid approach
Protein molecules are dynamic in cellular environment, such that the biological functions are
carried out via a series of conformational changes. Existing structure determination methods
mostly focus on determining static snapshots of the conformation space. Such static structures are
do facto some forms of ensemble averaged structures. Single molecule structure determination
methods, such as cryo electron-microscopy (cryoEM) or X-ray scattering from single particles,
can detect structure information of individual molecules at distinct conformation state. However,
mapping the experimental data to each structure is challenging due to the information deficiency
from each measurement (2D measurements of the projections in unknown orientation of 3D
structures). Molecular Dynamics simulations, on the other hand, take static structures as starting
model and evolve to continuous trajectories, or in silico molecular movies. The generated
structures can compensate the unknown information from each measurement, such that each 2D
projection can be reliably mapped to specific structures that lie in the conformational space.
Furthermore, the free energy landscape can be recovered by exploiting the relation between
probability distribution and free energy as in Boltzmann distribution. Briefly, this proposal is aim
to develop a computational protocol that extract information from single molecule structure
determination method using computationally generated structures as seeds. The basic flow is as
the following: Experimental data can be converted to ensemble averaged structure (static), which
serves as the initial seed for the conformation generating; then the computational modeling, such
as physics based MD simulation methods, is applied to generate hypothetical structures
(dynamics); as the last step, the experimental data are mapped to the most likely conformation
states in the structure pool. The distribution function can be obtained at the end of the process,
which provides free energy information, and possible conformational transition pathways (energy
landscape). Most of the computations are perfectly parallelizable, taking full advantages of the
Tianhe-2 computer clusters. We request a computing time allocation of 0.6 million CPU hours
with some GPU resources, which is justified in the detailed research plan.
Tianhe2-JK Time Allocation Application
Leader: Li-Min Liu
6.72 million cores hours, or 800 cores for 8400 hours
(slim queue, the code can run against with CUDA or MIC)
As the world’s population increases and substantial industrial growth
continues, the energy demands of society increase rapidly. The energy
produced by the sun and carried down to earth by its radiation provides
an alternative and clean source of energy to meet and exceed the world’s
energy consumption demands. Organic-inorganic halide perovskites such
as CH3NH3PbI3 (MAPbI3) exhibit great promise for low cost and high
efficiency thin film solar cells due to the relative ease through solution
casting to achieve high degrees of crystallinity, excellent carrier transport
properties, tunable optical bandgaps, and strong light absorption.
Perovskites based solar cells have been under very fast development over
the past three years with the power conversion efficiency (PCE) rapidly
improved from 3.8% to 19.3%. Hence, perovskites based solar cells were
seen as the third-generation solar cells.
To solve the problem still in the photocatalytic applications of
CH3NH3PbI3 perovskite and make sure the mechanism of charge
transfer, we plan to perform large scale calculations with multiple
methods based on our initial results and achievements have been made
(two SCI publications and one submit).
Tianhe2-JK Time Allocation Application
Project Title: Quantum mechanical simulations of optoelectronic devices
Optoelectronics involves the study and application of electronic devices that source, detect
and control light, usually considered as a sub-field of photonics. In this project, we propose to
study quantum mechanically the electrical-to-optical or optical-toelectrical conversion in different
devices, including solar cells, light-emitting diode and plasmonic devices. In particular, we will
develop methods within our software package to study metal-halide perovskite solar cells,
dye-sensitized solar cells and GaN lightemitting diode. We propose here to employ our newly
developed quantum mechanical method to investigate in details the properties and mechanism of
these different optoelectronic devices. Unlike the existing classical or semi-classical theoretical
methods that focus on electronic structure calculation from which parameters such as effective
mass, bap gap and mobility can be inferred and subsequently substituted into classical or
semi-classical models, the quantum mechanical method that we recently developed can be used to
simulate directly the photo-excitation, electron-hole separation/recombination, charge carrier
migration, and heat dissipation, and evaluate the absorption coefficient, recombination rate,
radiative & non-radiative decay and power conversion efficiency without resorting to any classical
or semi-classical approximations. By simulating the electron-photon interaction process once, all
the relevant optoelectronic properties can be obtained. It was used to simulate quantum
mechanically the photo-induced current through a photovoltaic device made of a semiconductor
PN junction, and evaluate directly its power conversion efficiency, internal quantum efficiency,
open circuit voltage and short circuit current. In the proposed project, we plan to further develop
our method for applications of different optoelectronic devices, this includes light-emitting diode,
metal halide perovskite solar cells, dye-sensitized solar cells and plasmonic devices. The
electron-photon interaction will be included within the framework of non-equilibrium Green’s
function. The electromagnetic field is quantized to account for quantum mechanically the coupling
between the electrons and photons, and the heat dissipation due to the electron-phonon interaction
can be treated similarly. Specifically, absorption of photons, recombination of electron-hole pairs
or excitons, separation of electrons and holes and mobility of charge carriers of these devices will
be examined in depth to understand power conversion efficiency and to explore the possibility to
design more efficient devices.
The computational cost of the proposed project is high, mainly due to numerical integration
of Green’s function in large energy range. In addition, the frequency range of solar spectrum has
to be covered to obtain the power conversion efficiency of the solar cell. Fortunately, most of the
simulations are independent and can be readily parallelized.
Tianhe2-JK Time Allocation Application
QM and MD Simulations-Based Materials Design for Hydrogen Production and
Storage
Sateesh Bandaru, Limin Liu, Wei Cai
1.3 million core hours (150 cores for 8400 hours )
(slim queue, the code can run against with CUDA or MIC)
Hydrogen has emerged as one of the most promising alternative power fuel sources,
both as a transportation fuel and for proton-exchange membrane fuel cells, and as a
replacement for batteries for portable electric source. However, a number of difficulties
have risen in regards to small and large scale storage. Significant efforts are underway
to design materials for efficient chemical hydrogen storage, from both synthetic
and computational perspectives. Overall, tremendous progress has been made in recent
years in the discovery of new materials. However, the fundamental advantage of
hydrogen production from efficiency of photochemical water splitting is that the
hydrogen may be used as a mechanism for energy storage; this is considered to be an
efficient storage medium by many within the scientific community.
This study intends to make use of the quantum chemical theories to assist the catalyst
design and understanding the reaction profile of splitting of water using a metal catalyst.
The prime aim of this research is concerned with investigating mechanisms for
artificial water splitting by transition metal complexes, together with the concept of
insilico design of transition metal complexes which are commercially cheaper and
experimentally (synthetically) viable catalysts for suitable water oxidation catalysis
(WOC). Although the catalytic process of transition metal complexes for water
oxidation has been explored by several experimental and theoretical studies, the subtle
mechanisms of some key reaction processes still remain elusive. So, our focus is mainly
to explore the mechanistic pathway at both lower- and higher-concentration
mechanisms and explore intermediate reaction steps. In particular, assessing their
nature, e.g., whether proton-coupled electron transfer or stepwise processes, and,
finally, ease of O2 release from the pre- catalyst, are sine qua non. One of the main
objectives of this project is to develop theoretical methods that can serve as efficient
screening tools for the reactivity of putative catalyst candidates. Before exploring these
mechanisms with transition metal complexes, we need to validate the various DFT
simulation methods. These kinds of system will need more comprehensive simulation
techniques, such as fully explicit solvent models or the inclusion of counter-ions to
determining more accurately the redox potentials of these highly oxidized species.
Therefore, another objective of this project is to develop the theoretical methods that
can serve as efficient screening tools for the reactivity of putative catalyst candidates.
Until now, we do not have any detailed theoretical insights into WOC reactions. There
is an essential need to know which methodology is most appropriate. Initially, we want
to validate first-principles DFT studies to transition-metal-catalyzed water oxidation
reactions by using well-known catalyst such as single-site polypyridyl ruthenium
complexes [Ru(tpy)(bpm)(OH2)]+2
. Then the reliable methods will be applied to
unknown designed catalysts, and the calculated redox potentials and activation energy
barriers compared with known catalysts. The central aim of this proposal is to design a
novel and commercially cheap and experimentally viable structures for artificial
water-splitting by transition metal complexes.
The other important project in which the molecular dynamics simulation (MD) and ab
initio molecular dynamics (AIMD) will be performed of interfaces between metal
oxides and liquid water, to capture the rich tapestry of chemical and physical adsorption
interactions in all of their physic-chemical complexity, employing state-of-the-art
treatments of dispersion. Partial and full coverage of chemically and physically
adsorbed water molecules will be investigated, as well as full condensed phases of
liquid water in contact with metal oxide surfaces. A particular focus will be on the
dynamical properties of hydrogen bonds between protons in water molecules and the
bridging oxygen atoms at the surface, as well as the variation in bond-stretch and
bond-angle bending modes in the water molecules. The dipolar orientation of the water
molecules vis-à-vis the surface-normal will also be studied, as well as
water-dissociation kinetics and thermodynamics on these surfaces. The importance of
kinks and surface irregularities will also be studied very closely. It is to be expected that
this will contribute to our somewhat lacking understanding of water interactions with
metal oxide surfaces, to enhance efforts in the optimal design of nano-material surfaces
for photo-catalytic water-splitting.
Abstract of proposed research
Yi-bing ShanComplex Systems Divison
Proteins are building blocks of life and they are associated in almost all aspects of cellularactivities. Since a protein’s function is rooted in its distinct three-dimensional structures, the deter-mination of the atomic structures and to establish the structure-function relationships are crucialfor understanding life phenomenon at the molecular level.
The objective of our research is to understand the activation mechanism of plexin, an importantcell surface receptor protein that transduces signals for regulating neuronal axon guidance andimmune responses. Malfunction of plexin signaling has been related to neurological disorders andvarious cancers. It has been proposed as promising drug targets for a number of important diseases.Xuewu Zhang and colleagues form the University of Texas Southwestern Medical Center showedthat the upstream signal leads to dimerization and activation of plexins, and upon the activationplexins bind and activate the GTPase activity of a Rap proteins as a GTPase activation protein(GAP); the Rap proteins in turn interact with effectors and produce a downstream signal. However,in structural terms how plexins are restrained in its inactive state, and how they are activated bythe induced dimerization remain unclear.
The regulation mechanisms remained largely elusive until Xuewu Zhang et al. from the Universityof Texas Southwestern Medical Center published their pioneer experimental works on the structuralbasis for activation of plexin. They found plexin activation requires binding of Rap (a kind ofGTP-binding protein), but the mechanism of this process is still unclear and difficult to study byexperiments. Xuewu Zhang hence hope to cooperate with us on solving this important problemby investigating the functional related conformational dynamics of plexin using all-atom moleculardynamics simulations.
All-atom molecular dynamics simulations (MD) is an increasingly applied method for studies ofprotein conformational dynamics, which reveals atomic details of motions of macromolecules thatare often functionally highly important. However, despite the rapid development of hardware andsoftware, demanding computation cost still poses a major limitation to MD. Tianhe2-JK, a top rankgeneral-purpose supercomputer in China, is ideal for carrying out the demanding computation ofour research in a reasonable time scope.
Furthermore, with the MD simulations of plexin and other proteins as benchmark, we also intentto develop a novel sampling method aiming to accelerate molecular dynamics simulations. Gentlestascending dynamics (GAD) was proposed by Weinan E as a new sampling method that efficientlyidentifies the transition conformations of a protein in conformational changes. This method cangreatly reduce the computing time once it is adopted for protein simulations. Efficient implementa-tion of GAD on the platform of GPGPU technology (CUDA or OpenCL) will produces a softwarepackage that can be incorporated into the mainstream MD packages such as AMBER or GROMACS.
We shall study the functional-related conformational dynamics of plexin by all-atom moleculardynamics. GAD scheme will also be implemented and tested using plexin as the primary testingsystem. Such work requires simulations of plexin systems to reach microsecond timescale. Our testruns of these systems suggest that 30 slim nodes (600 cores) and one GPU node on Tianhe2-JK areneeded to complete such runs in approximately one year.
1
Tianhe2-JK Allocation application:
Simulating life’s crucial protein machinery for genetic and
metabolic control
Gene transcription is the first step in gene expression. The essential enzymes that
direct the processes are RNA polymerases (RNAPs; see Fig 1 left), which copy
information from template DNA to a newly synthesized RNA strand based on
Watson-Crick base pairing [1]. In recent years, high-resolution structures of RNAPs
become abundant, however, functional mechanisms in structural dynamics detail
remain illusive. In this application, we want to implement atomistic molecular
dynamics (MD) simulations to investigate physical mechanism of RNAP function.
The simulations take advantage of high performance computation (HPC), and apply to
all-atom biomolecular systems up to milliseconds with femtosecond time step [2]. Fig
1. Schematic structural views of an RNA polymerase (left) and F1-ATPase (right) In the
incoming year, we would focus on MD simulations of RNAP from bacteriophage T7,
a nice model system to study fundamentals of transcription. We have examined how a
critical amino acid selects right nucleotides over wrong ones in T7 RNAP [3], and
also simulated the product release and constructed the corresponding Markov state
model (MSM) [4]. Further, we will conduct systematical free energy calculations to
quantify the nucleotide selectivity essential for fidelity control. We will also construct
the MSM for the T7 RNAP translocation and design mutant polymerases for desired
functions, such as backtracking and/or proofreading.
In addition, we want to study metabolically essential protein machinery
F1-ATPase (Fig 1 right) using also the MD simulation. F1-ATPase is part of the
synthase and ATPase, which can synthesize or hydrolyze ATP, the energy currency of
the cell. Our focus is on how the sequential hydrolysis pattern arises on the ATPase
ring, and the mechanisms can be universal for many ring-shaped motor proteins (such
as the viral DNA packaging motor [5] we studied). We want to combine atomistic
simulations with stochastic modeling and coarse-grained simulations, so that to
decipher the inter-subunit coupling leading to the sequential pattern at multiple time
scales.
In charge of the research work: Dong-Bo Zhang
Title in CSRC: Faculty
Application for the Tianhe2-JK Time.
Compared to electronic and photonic transports, lattice heat conduction is less well
understood for many materials. Our planed research projects are focusing the exploration of the
lattice thermal conductivities of bulk materials through ab initio molecular dynamics and classical
molecular dynamics. Two systems are considered: Fe-bearing MgSiO3 perovskite and MgO
periclase with defects. Recently, the PI developed a hybrid approach that combines ab initio
molecular dynamics and lattice dynamics to enable the first principle study of lattice anharmoncity
of complex crystal. In this approach, the anharmonic effect induced by intrinsic phonon-phonon
interaction are fully accounted for in terms of phonon quasi-particle. This approach allows not
only the calculation of anharmonic phonon dispersion, but also the calculation of phonon linewith.
The phonon lifetime, group velocity, and heat capacity can thus be obtained straightforward. The
knowledge on phonon properties enable the study of lattice thermal conductivity through phonon
gas model, where Boltzmann transport equation is employed.
MgSiO3 perovskite and MgO periclase are the most abundant minerals in the earth lower
mantle. Their properties determines to a great extent the properties of the lower mantle. Therefore,
the understanding on their thermal conductivity will help gain insight into the heat transfer inside
the deep earth. The present project will focus on the behavior Fe-bearing MgSiO3 perovskite and
MgO periclase, under extreme conditions, i.e., high pressure and high temperature mimicking the
lower mantle condictions. The study on Fe-bearing MgSiO3 perovskite aims to elucidate the role
of Fe ingredient in the thermal conductivity of MgSiO3 perovskite. On the other hand, the study
on MgO periclase will address the impact of defects on the thermal conductivity of MgO periclase.
Both are of crucial importance in geoscience.
First-Principles Studies in the Understanding and Design of Lithium-ion
Battery Materials
Yanning Zhang
Lithium-ion batteries (LIBs) are now ubiquitous in portable electronics because of
their high specific energy, high energy density, low weight and low volume.
Intercalation of LIBs to other advanced battery systems have also been developed for
a wide range of promising applications in, such as electrotraction in hybrid cars,
storage devices of solar power systems, and micro battery systems for integration in
electronic and medical devices. However, as LIBs are widely introduced into
commercial applications, several additional factors such as safety, minimum time to
charge and discharge, and cycle life are becoming increasingly important. Therefore,
researchers continually seek new electrode materials that are significant in structural
stability, safety, specific energy, and low cost, based on a good understanding and
control of the involved processes during charging-discharging (delithiation-lithiation)
cycle of the battery.i,ii
Encouraged by the significant development of pseudopotential plane wave approaches
and tremendous increase in computational capacity, in the last 20 years, the
first-principles electronic structure methods have shown highly reliability in the
prediction of phase stability and a large variety of physical and electronic
properties.iii,iv,v,vi
In the LIB research field, computational methods have successfully
elucidated several key properties of electrodes, such as cell voltage, ionic and
electronic conductivities, phase stability, and thermal stability, and more importantly,
predicted many high energy density electrode materials before experimental
realization, such as the Li(Al,Co)O2 system.vii,viii,ix
Obviously, a complementary
theoretical and experimental approach, combining aspects of materials science,
physics and chemistry, will be a cost-effective way to explore new materials for
battery applications.
i R. Marom, S.F. Amalraj, N. Leifer, D. Jacob, and D. Aurbach, J. Mater. Chem. 21, 9938 (2011).
ii J.R. Szczech and S. Jin, Energy Environ. Sci. 4, 56 (2011).
iii J.N. Reimers, Journal of Power Sources 54, 16 (1995).
iv J.B. Goodenough and Y. Kim, Chemistry of Materials 22, 587 (2010).
v J.S. Tse and J.J. Yang, “Lithium Ion Batteries: Advanced Materials and Technologies”, Chapter
6 “First-Principles Methods in the Modeling of Li-Ion Battery Materials”. Edited by X.X. Yuan,
H.S. Liu, and J.J. Zhang, CRC Press (2012).
vi L.M. Yan, J.M. Su, C. Sun, and B.H. Yue, Adv. Manuf. 2, 358 (2014).
vii V. L. Chevrier, S. P. Ong, R. Armiento, M. K. Y. Chan, and G. Ceder, Phys. Rev. B 82, 075122
(2010).
viii G. Ceder, G. Hautier, A. Jain, and S.P. Ong, MRS Bulletin 36, 185 (2011).
ix G. Ceder, Y.M. Chiang, D.R. Sadoway, M.K. Aydinol, Y.I. Jang, B. Huang, Nature 392, 694
(1998).