INPUT文件

此文件中包含15个文件，逐个分析他们的功能

Cu_1620.data：LAMMPS的初始文件需要搭配in文件一起使用
README.md：解释文件，解释各个文件的功能以及操作流程
VASP/：文件夹
auto.sh：执行最主要的文件
coord.py
fill_with_vacuum.py
in.my：LAMMPS的初始文件需要搭配data文件一起使用
init.almtp
md_al_mtp.sh
pot_save/
pre_train.sh
run_lammps.sh
select.sh
train.sh
train_one.cfg

VASP/文件夹（里面的文件就是满足完成一个POSCAR的AIMD模拟）

 INCAR
 KPOINTS
 POTCAR
 README.md
 run_vasp.sh：提交任务的脚本文件
 sub.sh：提交脚本任务的脚本

auto.sh

#!/bin/bash  # 指定该脚本使用的解释器为 Bash。cp init.almtp curr.almtp 
cp train_one.cfg train.cfg
cp Cu_1620.data dump.trjis_training_finished=0sbatch -J pre_train -o pre_train.out -e pre_train.err  -N 1 --exclusive -p high,MMM ./pre_train.sh --time=3:00:00 
#用于在Slurm作业调度系统中提交一个作业，并运行名为pre_train.sh的脚本。
#这个脚本的功能就是进行一次主动学习势能训练，用上面三个文件while [ $is_training_finished -ne 1 ]
dosleep 5if [ -f is_training_finished.txt ]; thenis_training_finished=1echo $is_training_finishedelseis_training_finished=0echo $is_training_finishedfi
done
sleep 5
rm -f *.txt#判断上面的训练是否完成，并重置for p in $(seq 1621 3000)
doecho "# LAMMPS data file written by OVITO Basic 3.7.11" > Cu_${p}.dataecho "${p} atoms" >> Cu_${p}.dataprev=$((${p}-1))cat Cu_${prev}.data | tail -n +3 | head -n 11 >> Cu_${p}.datacat dump.trj | tail -n ${prev} >> Cu_${p}.datapython coord.py $p
#更新原子坐标信息echo "active learning with ${p} atoms is starting"./md_al_mtp.sh
#执行一系列操作cp dump.trj pot_save/cp train.cfg pot_save/cd pot_savemv dump.trj dump_${p}.trjmv train.cfg train_${p}.cfgcp curr.almtp curr_${p}.almtpcd ../cp pot_save/curr.almtp .
done

md_al_mtp.sh

#!/bin/bashsource ~/.bashrc
n_cores=$1
rm -f preselected.cfg.*
rm -f preselected.cfg
rm -f selected.cfg
rm -f nbh.cfg
rm -f *.txt
while [ 1 -gt 0 ]
do
cp curr.almtp pot_save/
touch preselected.cfg
is_lammps_finished=0
sbatch -J lammps -o lammps.out -e lammps.err  -N 1 -n 16 -p high,MMM ./run_lammps.sh
while [ $is_lammps_finished -ne 1 ]
dosleep 5if [ -f is_lammps_finished.txt ]; thenis_lammps_finished=1echo $is_lammps_finishedelseis_lammps_finished=0echo $is_lammps_finishedfi
done
sleep 5
cat preselected.cfg.* >> preselected.cfg
rm -f preselected.cfg.*
n_preselected=$(grep "BEGIN" preselected.cfg | wc -l)
is_selection_finished=0
if [ $n_preselected -gt 0 ]; thenecho "selection"sbatch -J selection -o selection.out -e selection.err  -N 1 -n 1 -p bigmem ./select.shwhile [ $is_selection_finished -ne 1 ]dosleep 5if [ -f is_selection_finished.txt ]; thenis_selection_finished=1echo $is_selection_finishedelseis_selection_finished=0echo $is_selection_finishedfidonesleep 5if test -f selected.cfg; thenecho " selected.cfg exists"elseecho "selected.cfg not exist"exitfirm -f preselected.cfg./mlp cut_extrapolative_nbh selected.cfg nbh.cfg --cutoff=8 ./mlp convert nbh.cfg POSCAR --output_format=poscar rm -f selected.cfgn_poscar=$(grep "BEGIN" nbh.cfg | wc -l)n_poscar_prev=$((${n_poscar}-1))if [ $n_poscar -eq 1 ]; thenmv POSCAR POSCAR0fifor p in $(seq 0 ${n_poscar_prev})dopython fill_with_vacuum.py $pdonefor ((i=0; i<$n_poscar; i++))docp POSCAR_output"$i" VASP/mkdir -p VASP/"$i"if [ $n_poscar -eq 1 ]; thenmv VASP/POSCAR_output"$i" VASP/0/POSCARelif [ $n_poscar -gt 1 ]; thenmv VASP/POSCAR_output"$i" VASP/"$i"/POSCARficp VASP/POTCAR VASP/"$i"/cp VASP/INCAR VASP/"$i"/cp VASP/KPOINTS VASP/"$i"/cp VASP/run_vasp.sh VASP/"$i"/cp VASP/sub.sh VASP/"$i"/donefor ((i=0; i<$n_poscar; i++))docd VASP/"$i"sbatch ./run_vasp.shcd ../../doneis_vasp_finished=0while [ $is_vasp_finished -ne 1 ]dosleep 5n_vasp_finished=$(grep -rH "1" VASP/*/is_vasp_finished.txt | wc -l)if [ $n_poscar -eq $n_vasp_finished ]; thenis_vasp_finished=1echo $is_vasp_finishedelseis_vasp_finished=0echo $is_vasp_finishedfidonesleep 5for ((i=0; i<$n_poscar; i++))do./mlp convert VASP/"$i"/OUTCAR VASP/"$i"/calculated.cfg --input_format=outcar #--elements_order=29 --absolute_elementscat VASP/"$i"/calculated.cfg >> train.cfgrm -r VASP/"$i"/doneis_training_finished=0sbatch -J train -o train.out -e train.err  -N 1 --exclusive -p high,MMM ./train.sh --time=3:00:00while [ $is_training_finished -ne 1 ]dosleep 5if [ -f is_training_finished.txt ]; thenis_training_finished=1echo $is_training_finishedelseis_training_finished=0echo $is_training_finishedfidonesleep 5rm -f nbh.cfgrm *.txtrm POSCAR*rm -f nbh_36.cfg    
elif [ $n_preselected -eq 0 ]; thenexit
fidone