|
LAMMPS讲解69-GARFfield安装GARFfield是由加州理工学院Andres Jaramillo-Botero所开发的针对ReaxFF力场的拟合软件，全称是Genetic Algorithm based Reactive Force field optimizer method。它是基于遗传算法，爬山法和共轭梯度最小化法开发的多目标参数优化软件，可以实现Pareto-optimal参数优化。目前GARFfield支持ReaxFF力场，eFF-ECP力场，Morse力场，COMB力场，Tersoff力场和EAM合金力场的开发。其开发流程和商业的SCM软件类似，包括输入文件等。 GARFfield下载地址加微信baolu_yao NOTE: GARFfield is currently available for non-profit academic and research activities. Its use for commercial purposes requires a commercial license that can be requested via email to ajaramil@caltech.edu. If you want to try GARFfield, please read and accept the terms of the LICENSE. 本教程为该软件在Linux下的安装过程。安装需要确保你系统中已安装有gcc，g++和gfortran
首先下载软件，将软件放在合适的目录下，比如就放在了如下目录中 /home/bly/software 输入解压缩命令 tar -xvzf GARFfield.tar.gz 然后下载mpich3.4.3和FFTW 下载链接分别为 https://www.mpich.org/static/downloads/3.4.3/mpich-3.4.3.tar.gz http://www.fftw.org/fftw-3.3.10.tar.gz 下载好后放在和GARFfield一样的目录下，如 /home/bly/software 分别输入解压缩命令解压mpich和fftw的安装包 tar -xvzf mpich-3.4.3.tar.gz tar -xvzf fftw-3.3.10.tar.gz 进入mpich解压缩出的目录，依次执行以下命令 ./configure --prefix=/home/bly/software/mpich --enable-shared --disable-fortran make make install 进入fftw解压缩出的目录，依次执行以下命令 ./configure --prefix=/home/bly/software/fftw --enable-shared make make install mpich和fftw安装完成后打开.bashrc，在末尾添加如下两行 export PATH=/home/bly/software/mpich/bin:$PATH export LD_LIBRARY_PATH=/home/bly/software/mpich/lib:$LD_LIBRARY_PATH 保存文件，然后执行 source ~/.bashrc 进入GARFfield下的这个目录 /home/bly/software/GarFfield/lib/lammps/lib/reax 执行以下命令 make -f Makefile.gfortran 打开如下目录中的Makefile.ion文件 /home/bly/software/GarFfield/lib/lammps/src/MAKE 在打开的Makefile.ion文件，在文件头部位置处修改MPICH和FFTW两个变量的值为 MPICH = /home/bly/software/mpich FFTW = /home/bly/software/fftw 保存文件，进入lammps的src目录 /home/bly/software/GarFfield/lib/lammps/src 执行以下命令 make ion make makeshlib make -f Makefile.shlib ion 进入以下目录 /home/bly/software/GarFfield/lib/optlist 执行 make 进入以下目录 /home/bly/software/GarFfield/lib/pgapack/source 打开该目录下的Makefile文件，在该文件的头部修改MPI_HOME变量的值为 MPI_HOME = /home/bly/software/mpich 保存文件，执行以下命令 make 进入以下目录 /home/bly/software/GarFfield/src 打开该目录下的Makefile文件修改变量MPI_HOME，LAMMPS，FFTW的值为 MPI_HOME = /home/bly/software/mpich LAMMPS = /home/bly/software/GarFfield/lib/lammps FFTW = /home/bly/software/fftw 删掉该Makefile文件中LAMMPS_LIB变量的-ljpeg，保存文件 执行以下命令 make 编译完成后会在以下目录中找到garffield_ion文件 /home/bly/software/GarFfield/bin 打开.bashrc文件在文件末尾添加以下内容 export PATH=/home/bly/software/GarFfield/bin:$PATH export LD_LIBRARY_PATH=/home/bly/software/GarFfield/lib/lammps/src:$LD_LIBRARY_PATH export LD_LIBRARY_PATH=/home/bly/software/GarFfield/lib/optlist:$LD_LIBRARY_PATH export LD_LIBRARY_PATH=/home/bly/software/GarFfield/lib/pgapack/lib/ion:$LD_LIBRARY_PAT 执行 source ~/.bashrc 运行 mpirun -np 2 garffield_ion 输入以下内容则软件安装成功 GARFfield force field parameter optimization framework, version 1.0 Author: Andres Jaramillo-Botero (ajaramil@caltech.edu, Caltech (c) 2010-2014)
Usage: garffield_ion <geo> <ffield> <trainset> <params> [<restraints>] [GA options] [other options] ... or mpirun -np NumProcs garffield_ion <geo> <ffield> <trainset> <params> [<restraints>] [GA options] [other options]
GA options are: -i ['range' or 'percent'] {parameter interval} -g ['GRGA' or 'SSGA'] {replacement strategy: generational or steady-state} -m [float between 0-1] {mutation rate} -o [float between 0-1] {crossover rate} -p [int] {population size} -s ['maxiter' or 'nochange'] {convergence criteria} -t [int] {stop after int iterations} -z [int] {set SSGA population replacement percent} -x {perform mutation and crossover} Other options are: -c [int] {Hillclimb every int iterations} -C [int] {mincap array size in reax/c allocations} -d ['full_debug', 'lammps_debug'] debug mode -e ['MPE' or 'SIWE' or 'RMSE' or 'NRMSE' or 'MSE'] {error function} -f {check forces on unit cells, instead of performing minimization} -F ['reax'] -G [int] {switch to Conjugate Gradient after int iterations} -I [int] {Number of molecular mechanics energy minimization steps} -l <filename> {write log to filename, instead of console} -M ['fire' or 'sd'] {switch structure minimization algorithm - default is 'cg'} -q <filename> {Use pQeq for ReaxFF charges} -r [int] {print report every int iterations} -R [int] {Restart every int interations}
-S [float] {safezone factor in reax/c memory allocation} -u ['real'] {Force units to be real, i.e. lengths in Angstroms, energies in kcal/mol and forces in kcal/mol-Angstrom} -v {Use low-gradient van der Waals for reax/c} -w <filename> {write best force field file at same print report report_frequency} -W {randomize section weights in training file} Abort(1) on node 0 (rank 0 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0
感谢鲍路瑶老师的分享，内容来自于鲍老师分享出来的资料 如有需要添加微信：lmp_zhushou 进入微信群，帮助他人，共建社区 获取完整版lammps讲义可以加微信lmp_zhushou或加入QQ群994359511 |
