New advances in chemistry and materials science with CPMD and parallel computing

doi:10.1016/S0167-8191(00)00014-4

Parallel Computing

Volume 26, Issues 7–8, July 2000, Pages 819-842

https://doi.org/10.1016/S0167-8191(00)00014-4 Get rights and content

Abstract

A short overview is presented of the density functional theory and molecular dynamics (DFT–MD) method and of a code (CPMD) based on a plane wave scheme. Its power is shown through the survey of specific applications to diverse frontier areas of chemistry and materials science that make use of parallel computing.

Introduction

The method of choice for computational studies of the chemistry of extended systems is density functional theory (DFT) [28], [32], [33]. The unification of DFT and molecular dynamics (DFT–MD), as achieved in the seminal paper of Car and Parrinello [12], has greatly enhanced the flexibility of DFT by enlarging the range of possible applications, e.g. from solids to liquids, from statics to dynamics, and from solid-state physics to physical chemistry. By providing the ability to treat larger systems and longer time scales, parallel computing makes the applicability of DFT to real systems progressively more feasible.

In this paper, after briefly recalling the basics of the DFT–MD method and of its implementation in the CPMD parallel code by Jürg Hutter [20], we give a survey of applications that we have made over the past few years in diverse frontier areas of chemistry and materials science.

Section snippets

DFT–MD and the CPMD parallel code

In this section, we shall briefly describe the DFT–MD method and the CPMD code [20] and emphasize some details of its parallelization [35]. We point the reader to the relevant literature as well.

The CPMD code is an implementation of DFT in the Kohn–Sham (KS) formulation [32], [33], [28] and of the Car–Parrinello MD scheme [12]. Plane waves are used as the basis set for the valence electron wave functions and pseudo-potentials to describe the interaction between the valence electrons and the

Applications

With the method explained above, one can handle diverse atomic aggregation phases, i.e. molecules and clusters (“isolated” [7]) as well as solids and liquids (with periodic boundary conditions), one can simulate the dynamics of a given system at either constant energy, temperature [43] or pressure [44], and also impose geometrical constraints.

Table 1 contains a summary of applications of the method discussed in Section 2 that have been made over the past decade on IBM machines of various

Conclusion

Here, a brief discussion has been presented of the DFT–MD method, of the computational features of the CPMD code and its advantages when tuned with parallel machines, as well as of the progress made with it in the study of several problems in chemistry and materials science. This constitues only the beginning of a new, direct approach to real materials issues from first principles. In fact, the largest systems one can currently treat accurately on a machine of common access (32 nodes SP2) are a

References (58)

M. Bühl et al.
Chem. Phys. Lett.
(1997)
E.A. Carter et al.
Chem. Phys. Lett.
(1989)
A. Curioni et al.
Chem. Phys. Lett.
(1998)
P. Liu et al.
Surf. Sci.
(1998)
D.S. Maddison et al.
Synth. Metals
(1988)
J. Wiggs et al.
Comp. Phys. Commun.
(1994)
M.S.A. Abdou et al.
J. Am. Chem. Soc.
(1997)
W. Andreoni
Ann. Rev. Phys. Chem.
(1998)
W. Andreoni et al.
Phys. Rev. Lett.
(1996)
W. Andreoni et al.
Appl. Phys. A
(1998)

W. Andreoni, P. Giannozzi, in: W. Andreoni (Ed.), The Physics of Fullerene-Based and Fullerence-Related Materials,...

W. Andreoni et al.

J. Am. Chem. Soc.

(1996)

R.N. Barnett et al.

Phys. Rev. B

(1993)

A.D. Becke

Phys. Rev. A

(1988)

C.M. Brown et al.

Chem. Mat.

(1996)

G.E. Brown

Chem. Rev.

(1999)

R. Car et al.

Phys. Rev. Lett.

(1985)

P. Carloni, M. Sprik, W. Andreoni, J. Phys. Chem. 104 (2000)...

D.M. Ceperley et al.

Phys. Rev. Lett.

(1980)

J. Perdew et al.

Phys. Rev. B

(1981)

L.J. Clarke et al.

Comp. Phys. Commun.

(1993)

R. Colle et al.

J. Am. Chem. Soc.

(1998)

See e.g. Conjugated Polymers and Related Materials: The Interconnection of Chemical and Electronic Structure, Oxford...

CPMD 3.0 Copyright IBM Corporation (1990–1997) and MPI Festkörperforschung Stuttgart,...

A. Curioni, W. Andreoni, Synth. Metals, in...

A. Curioni et al.

J. Am. Chem. Soc.

(1999)

A. Curioni et al.

J. Am. Chem. Soc.

(1994)

A. Curioni et al.

J. Am. Chem. Soc.

(1997)

A. Curioni et al.

Appl. Phys. Lett.

(1998)

Cited by (145)

Integrating state of the art compute, communication, and autotuning strategies to multiply the performance of ab initio molecular dynamics on massively parallel multi-core supercomputers
2021, Computer Physics Communications
The development in today’s supercomputer hardware is that the compute power of the individual nodes grows much faster than the speed of their interconnects. To benefit from this evolution in computer hardware, the challenge in modernization of simulation software is to increase the computational load on the nodes and to reduce simultaneously the inter-node communication. Here, we demonstrate the implementation of such a strategy for plane-wave based electronic structure methods and ab initio molecular dynamics (AIMD) simulations. Our focus is on ultra-soft pseudopotentials (USPP), since they allow to shift workload from fast Fourier transforms (FFTs) to highly node-efficient matrix–matrix multiplications. For communication intensive routines, as the multiple distributed 3- $d$ FFTs of the electronic states and the distributed matrix–matrix multiplications related to the $β$ -projectors of the pseudopotentials, parallel MPI+OpenMP algorithms are revised to make use of overlapping computation and communication. The necessary partitioning of the workload is optimized by auto-tuning algorithms. In addition, the largest global MPI_Allreduce operation is replaced by highly tuned node-local parallelized operations using MPI shared-memory windows to avoid inter-node communication. A batched algorithm for the multiple 3- $d$ FFTs improves the throughput of the MPI_Alltoall communication and, thus, the scalability of the implementation, both for USPP and for frequently used norm-conserving pseudopotentials. The new algorithms have been implemented in the AIMD program CPMD (www.cpmd.org). The enhanced performance and scalability of the code is demonstrated on simulations of liquid water with up to 2048 molecules. It is shown that 100 ps simulations with many hundred water molecules can be done now routinely within a few days on a moderate number of nodes.
Alternative insight into aluminium-phosphate glass network from ab initio molecular dynamics simulations
2021, Ceramics International
Phosphate glasses are an important group of materials for wide applications. They have received attention as biomaterials, eco-fertilizers, in optoelectronic devices, waste immobilization, etc.
The subject of the studies is binary Al₂O₃–P₂O₅ glass, seen by ab initio molecular dynamics simulations. In the paper, Mayer's Bond Order analysis was employed as a suitable tool to observe the changes in the network due to Al₂O₃ substitutions. As well as, the bonding properties across the atoms in the network, which were characterized by electron localization function. The basic structural properties of the glasses are presented.
The simulations showed that oxygen atoms in [PO₄] tetrahedrons create a weak covalent bond between the atoms. The interaction spreads across the whole phosphate network via bridging oxygens. Whereas, there is no such interaction between the atoms in [AlO_x] polyhedrons. In the case of the Al₂O₃ rich glasses, there are evidence non-network oxygen atoms which influence the distribution of phosphate structural units predicted based on the O/P ratio. The possibility of estimation of the oxygens in an experiment is discussed.
The atoms building the network are going to take specific values of the total bond orders and the glass network is an interplay between atoms’ affinity to saturate bond orders and the glass network neutrality.
A revised mechanism for (p)ppGpp synthesis by Rel proteins: The critical role of the 29-OH of GTP
2020, Journal of Biological Chemistry
Citation Excerpt :
The coordinates of the enzyme–substrate complex, equilibrated by MM simulations, were taken for further QM/MM simulations of the Rel–GTP–ATP, as well as the Rel–dGTP–ATP complex. These hybrid QM/MM simulations were carried out using the CPMD/GROMOS interface (39, 40). The QM subsystem includes (d)GTP4-, ATP4-, a Mg2+ ion, an active site water molecule, and the side chains of Lys52, Lys60, and Asp79 (Fig. S7).
Bacterial Rel proteins synthesize hyperphosphorylated guanosine nucleotides, denoted as (p)ppGpp, which by inhibiting energy requiring molecular pathways help bacteria to overcome the depletion of nutrients in its surroundings. (p)ppGpp synthesis by Rel involves transferring a pyrophosphate from ATP to the oxygen of 3′-OH of GTP/GDP. Initially, a conserved glutamate at the active site was believed to generate the nucleophile necessary to accomplish the reaction. Later this role was alluded to a Mg²⁺ ion. However, no study has unequivocally established a catalytic mechanism for (p)ppGpp synthesis. Here we present a revised mechanism, wherein for the first time we explore a role for 2′-OH of GTP and show how it is important in generating the nucleophile. Through a careful comparison of substrate-bound structures of Rel, we illustrate that the active site does not discriminate GTP from dGTP, for a substrate. Using biochemical studies, we demonstrate that both GTP and dGTP bind to Rel, but only GTP (but not dGTP) can form the product. Reactions performed using GTP analogs substituted with different chemical moieties at the 2′ position suggest a clear role for 2′-OH in catalysis by providing an indispensable hydrogen bond; preliminary computational analysis further supports this view. This study elucidating a catalytic role for 2′-OH of GTP in (p)ppGpp synthesis allows us to propose different mechanistic possibilities by which it generates the nucleophile for the synthesis reaction. This study underscores the selection of ribose nucleotides as second messengers and finds its roots in the old RNA world hypothesis.
Mechanism of Nucleotidyltransfer Reaction and Role of Mg<sup>2+</sup> Ion in Sugar Nucleotidyltransferases
2020, Biophysical Journal
Sugar nucleotidyl transferases (SNTs) catalyze nucleotidyltransfer reactions to form sugar-nucleotides and pyrophosphate in the presence of two Mg²⁺ ions (Mg²⁺_A and Mg²⁺_B). We unveil the mechanism and free energetics of nucleotidyl transfer reaction in an SNT called GlmU through hybrid quantum mechanics-molecular mechanics molecular dynamics simulations and free energy calculations. The study identifies the roles of the active site residues and the Mg²⁺ ions in catalyzing the reaction. Of great significance, we are able to compare the free energy barrier for the reaction with that for the Mg²⁺-assisted release of the product (i.e., pyrophosphate) into the solution, shedding light on the general mechanistic and kinetic aspects of catalysis by SNTs.
Influence of aluminum on structural properties of iron-polyphosphate glasses
2020, Ceramics International
Citation Excerpt :
The obtained final atoms configurations were the starting point for the CPMD simulations. The simulations were conducted using the CPMD package [49]. Norm-conserving Troullier-Martins pseudopotentials were used for P, Fe, Al, and O atoms [50].
Phosphate glasses are materials that can be used in many technological fields. Their possible application may be limited due to the existence of easily hydrated P–O–P bonds. Addition of Al₂O₃, Fe₂O₃ strongly improves the durability of the glasses.
The subject of the studies are iron-polyphosphate glasses in which iron is gradually substituted by aluminum. The glasses have been tested using FT-IR, Raman, and Mössbauer spectroscopies. The obtained experimental results were compared to the theoretical prediction of the glass network structure by ab initio molecular dynamics.
It was shown that the substitution leads to an increase in the glass network polymerization. The glass network is inhomogenous with Al/Fe-rich and poor regions. In the poor regions, Al/Fe cations are glass network modifiers whereas in the rich they can be the glass network formers. Mayer's Bond Order analysis was used to describe bonding mechanisms in the glasses. The existence of the mix-valence iron atoms has been postulated.
Adsorption of formaldehyde on transition metal doped monolayer MoS<inf>2</inf>: A DFT study
2019, Applied Surface Science
Citation Excerpt :
For optimized CH2O, the bond lengths of CO and CH for CH2O molecule are about 1.21 Å and 1.12 Å, respectively, which are the same as those in the experimental values. For calculation of thermal stability, the calculation of ab initio molecular dynamics at 100 K, 300 K and 500 K were performed using the Car-Parrinello method (CPMD) [49–51]. All CPMD calculations were performed for 6 ps with 60,000 steps at different temperatures controlled by the standard Nosé thermostate [52].
Density functional theory (DFT) calculations have been performed to investigate the electronic properties of pristine monolayer MoS₂ and various transition-metal (TM) (Ni, Pt, Ti and Pd) doped MoS₂ as well as the adsorption of formaldehyde on these monolayer 2D structures. It was found that the CH₂O adsorption energy (0.11 eV) on monolayer MoS₂ was very weak, indicating that pristine MoS₂ is insensitive to formaldehyde (CH₂O). However, the adsorption of CH₂O on monolayer TM-MoS₂ shows relatively high affinity. Thus, TM-doping can substantially improve the sensitivity towards CH₂O. In order to investigate the mechanism of adsorption, the electronic properties such as molecular projected density of states and differential charge transfer were calculated. The results showed that the adsorption on Ti-doped monolayer MoS₂ (1.59 eV) is very promising with the most stable adsorption. Both differential charge transfer and ab initio molecular dynamic were used to investigate the adsorption mechanism. The ab initio molecular dynamic at 500 K was performed using the Car-Parrinello method. The most stable configuration shows the highest adsorption energy and CH₂O can be adsorbed even at 500 K. These suggest that Ti-doped monolayer MoS₂ would be a feasible material for CH₂O removal at room temperature.

View all citing articles on Scopus

View full text

New advances in chemistry and materials science with CPMD and parallel computing

Abstract

Introduction

Section snippets

DFT–MD and the CPMD parallel code

Applications

Conclusion

Chem. Phys. Lett.

Chem. Phys. Lett.

Chem. Phys. Lett.

Surf. Sci.

Synth. Metals

Comp. Phys. Commun.

J. Am. Chem. Soc.

Ann. Rev. Phys. Chem.

Phys. Rev. Lett.

Appl. Phys. A

J. Am. Chem. Soc.

Phys. Rev. B

Phys. Rev. A

Chem. Mat.

Chem. Rev.

Phys. Rev. Lett.

Phys. Rev. Lett.

Phys. Rev. B

Comp. Phys. Commun.

J. Am. Chem. Soc.

J. Am. Chem. Soc.

J. Am. Chem. Soc.

J. Am. Chem. Soc.

Appl. Phys. Lett.