QXMD: Non-adiabatic Quantum Molecular Dynamics Simulation Engine
QXMD is a scalable, parallel program for Quantum Molecular Dynamics simulations with various eXtensions. Its simulation engine is based on (time-dependent) density functional theory using pseudopotentials and plane-wave basis, while extensions include nonadiabatic electron-nuclei dynamics and multiscale shock technique. QXMD serves as a community-development platform for new methods and algorithms, a research platform on high-end parallel supercomputers, and an educational platform for hands-on training.
The parallel version of QXMD requires a Fortran compiler, as well as FFT and MPI libraries. There are no other library requirements for the QXMD.
To get started, clone this repository to your computer.
~$ git clone https://github.com/USCCACS/QXMD.git
First, change working directory to QXMD/
~$ cd QXMD
You will see following files and directories when you use the ‘ls’ command:
QXMD $ lsInclude/ Examples/ Lib/ Makefile QXMD_Manual_v3.2/ Sources/ util/
Include/ contains necessary libraries for compilation, Examples contains sample input files for various example simulations, Lib contains pseudopotential files, Makefile will be used to build the program, QXMD_Manual_v3.2 is the manual for this software packgae, Sources/ contains all QXMD source codes,and util/ contains helpful codes for post-processing output data from QXMD.
You have to choose makefile based on the configuration of your machine. For most of the facility supercomputers makefile is already configured. Complete list of the preconfigured makefile can be obtain by typing make help.
$make helpmake dec : DEC Alpha Degital Fortranmake gnu : GNU Fortran (only for classical MD)make ffc : PC LINUX (Fujitsu Fortran&C)make ffc-v64 : PC LINUX (Fujitsu Fortran&C) on VT-64make ifc : PC LINUX (Intel Fortran)make ifc-x86 : PC LINUX (Intel Fortran) on x86-64make ifc-v64 : PC LINUX (Intel Fortran) on VT-64make hpc : PC LINUX (Intel Fortran) on x86-64 at USC-HPCmake hp : HP FORTRAN 90make sp : IBM SPmake origin : SGI ORIGIN 2000make origin26 : SGI ORIGIN 2600make t3e : CRAY T3Emake sr2201 : HITACHI SR2201make sr8000 : HITACHI SR8000make sr11000 : HITACHI SR11000 at ISSPmake ha8000 : HITACHI HA8000-tc/HT210make vpp : Fujitsu VPP5000make sx7 : NEC SX-7make altix : SGI Altix3700/1280 at ISSPmake p5 : IBM eServer model p5make primequest : FUJITSU PRIMEQUEST 580make primergy : FUJITSU PRIMERGY RX200S3make cx400 : FUJITSU PRIMERGY CX400make primehpc : FUJITSU PRIMEHPC FX10make sekirei : Intel Fortran on SGI ICE XA/UV (systemB) at ISSP
If you are not using aforementioned machine, you might need to modify the Makefile according to your computing environment.
Precompiler flags are defined in CPPDEFS as:
CPPDEFS= -DFLAG1 -DFLAG2
Following flags are defined in QXMD
-LIBFFTW = link FFTW library-LIBFFTW3 = link FFTW3 library-DPOINTER64 = used for 64-bit machines-DSSL2VP = used on Fujitsu machine-DVECTOR = special implementation. It may result in faster computation on some machine.
There are several flags are defined for future development. Currently, they are not used. Several flags are specific to compiler. Please check compiler manual for further details.
Makefile defines which compiler you will use to build the QXMD executable.There are several predefined compiler setting which must be set. First, open Makefile available in QXMD directory and define the machine name. For example, theta machine at Argonne National Lab
theta: $(WDIR) $(DDIR) $(SOURCEFILE)sed "s/^#THETA#//" $(SOURCEFILE) > $(WORKFILE)
This process will ensure that make will understand the target. Next, we need to define the macro such as LINKER, LDFLAGS and FFTLIB in the makefile included in Source directory. For example, theta machine definition are as follow
#THETA# LINKER = ftn#THETA# FFLAGS = -c -fast -warn none#THETA# LDFLAGS = -O3 -ipo -no-prec-div#THETA# FFLAGS = -c -O3 -ipo -no-prec-div -warn none#THETA# CPPDEFS = -traditional $(CPPDEF) -DLIBFFTW3 -DIFC -DVECTOR#THETA# MKLPATH = $(MKLROOT)/intel/lib64#THETA# FFTLIB = -L$(MKLPATH) -Wl,--start-group -lmkl_intel_lp64 -lmkl_core -lmkl_sequential -Wl,--end-group
Definition of each macro is as follow:
LINKER= MPI Fortran compilerFFLAGS= Compiler flagsLDFLAGS= linking librariesCPPDEFS= Preprocessing MACROSMKLPATH= Variable defined for MKL librariesFFTLIB= FFT library
It is up to the user to define a new macro for better readability. Once you have defined all the macros you are ready to build code for new architecture
After tailoring the Makefile for your computing environment, type the command below to build the QXMD executable.
QXMD $ make $(MACHINE_NAME)QXMD $ make qxmd
To create a parallel version of QXMD, please use the following steps
QXMD $ make $(MACHINE_NAME)QXMD $ make qxmd_mpi
Check to see if you the QXMD executable has been created, and if so, you are ready to start a simulation.
qxmd $ lsInclude/ Makefile Qxmd/ qxmd_mpi Sources/ util/
Parallel build is activated by default. The program will use all available core on the machine to build the program efficiently.
In order to run QXMD, there are a few mandatory directories and files that must be present and in the correct hierarchy (Fig. 1). The working directory, from which a QXMD job is launched must contain a directory called “data/”, where all output data will be dumped, as well as a directory called “control/”, which must contain the following:
• CONFIG: a configuration file detailing the atomic coordinates of the system to be simulated• filename: a simple text file containing the path to the main input file• IN.PARAM: an input parameter file with various settings for the dynamics simulation• VELOC: an optional initial velocity files containing the initial three component velocities for each atom in the system• NCPP/ or PAW/ or USPP/: a directory containing pseudopotential files for each atomic species in the system
There are many example input files for various types of simulations in the Examples/ directory, including optimization of water, adiabatic QMD of water in the canonical ensemble, non-adiabatic QMD of monolayer MoSe2 in the microcanonical ensemble, and a MSST simulation of Si. The NAQMD and MSST examples are explained in more detail in the QXMD manual. Follow those tutorials to create images like the ones shown below!
To learn more about QXMD, please refer to QXMD manual.
This project is licensed under the GPL 3.0 license - see the LICENSE file for details