博文
再谈USPEX
|||
关注：
1)对缺乏实验数据对比的、有争议的体系，有必要采用不同的预测方法预测，对比结果
2）USPEX工作流程
1. 参数准备：
建立需预测结构目录，拷贝USPEX脚步进入该目录，在USPEX文件夹下
(1) 进入Specific/ 设定INCAR1-5
(2) 进入USPEX目录，编辑：INPUT，设定体系组成，包含的分子数(如6 18 for M6H18)，体积(与M6H18对应的体积)，键长(参考汉典元素周期表中的共价键)，化合价(参考汉典元素周期表中的氧化态)等参数，无需特别指明VASP路径，程序会通过locate.m脚步定位可执行的VASP程序
(3) 进入USPEX/seeds/下，输入初始猜测的POSCAR，并命名为POSCARS，cp POSCARS seed
注意：各文件夹下的可执行文件，应chmod +x 变成实际可执行的文件
2. 提交：
matlab <USPEX.m> log &
/opt/matlab2013/bin/matlab <USPEX.m> log &
或为保证程序周期性执行，修改并执行下述脚本，通过该脚本提交， 如，nohup ./run-uspex.sh，其中run-uspex.sh内容如下：
cat uspex_Sc-H
#!/bin/sh
while true;do
date > log
/opt/matlab2013/bin/matlab < USPEX.m >>log
sleep 60 【估计每个优化任务运行时间，可通过延长sleep时间如600s，来延缓队列中提交的任务数量的增加速度】
done
3. 看结果：
进入results文件夹/ vi goodstructures 找到对应的低焓结构 goodstructue_POSCAR
4. 重新提交及杀掉任务
（1） 每次重新提交，最好：
rm -rf results* 、
./clean_tough（清除CalcFold*文件夹） 、
cp Seeds/seed Seeds/POSCARS
（2） 改变VASP优化计算用CPU核的数目，需同时修改INPUT.TXT、submitJob_local.m、及SubmissionsubmitJob_local.m下的CPU数目
（3）如需 杀掉命令，则
ps aux | grep uspex
yexq 23255 0.0 0.0 106092 1212 pts/1 S 04:51 0:00 /bin/sh ./uspex_Sc-H
yexq 31252 0.0 0.0 103244 876 pts/8 S+ 05:15 0:00 grep uspex
kill 23255
对于在队列中已产生的大量任务，可通过脚本清除：
cat del-uspex.sh
#!/bin/sh
for((i=50747;i<50767;i++))
do
qdel $i
done
% i为Job ID.
5. 探索：
1） matlab <USPEX.m> log & 产生结构后，自动生成 提交 结构优化任务的vasp的脚本
vi results1/OUTPUT.txt
......
* There are 5 local relaxation steps for each individual structures *
Step Abinitio Code Execute Command K-resolution
1 VASP vasp 0.160
2 VASP vasp 0.120
3 VASP vasp 0.100
4 VASP vasp 0.080
5 VASP vasp 0.060
The script for job submission is prepared seperately in Submission/*_local.m
5 parallel calculations are performed simutaneously
2）解读 Submission/*_local.m
function doneOr = checkStatus_local(jobID)
%--------------------------------------------------------------------
%This routine is to check if the submitted job is done or not
%One needs to do a little edit based on your own case.
%1 : whichCluster (0: no-job-script, 1: local submission, 2: remote submission)
【在INPUT.TXT文件中已经指定】
%--------------------------------------------------------------------
%Step1: the command to check job by ID.
[a,b] = unix(['qstat ' jobID ''])
%Step2: to find the keywords from screen message to determine if the job is done
%Below is just a sample:
%-------------------------------------------------------------------------------
%Job id Name User Time Use S Queue
%------------------------- ---------------- --------------- -------- - -----
%2455453.nano USPEX qzhu 02:28:42 R cfn_gen04
%-------------------------------------------------------------------------------
%If the job is still running, it will show as above.
%If there is no key words like 'R/Q Cfn_gen04', it indicates the job is done.
if isempty(findstr(b,'R low')) & isempty(findstr(b,'Q low'))
doneOr = 1
unix('rm USPEX*'); % to remove the log file
else
doneOr = 0;
end
function jobNumber = submitJob_local()
%-------------------------------------------------------------
%This routine is to check if the submitted job is done or not
%One needs to do a little edit based on your own case.
1 : whichCluster (default 0, 1: local submission, 2: remote submission)
%-------------------------------------------------------------
%Step 1: to prepare the job script which is required by your supercomputer
fp = fopen('run.sh', 'w');
fprintf(fp, '#!/bin/sh\n');
fprintf(fp, '#PBS -l nodes=1:ppn=6,walltime=2:00:00\n');
fprintf(fp, '#PBS -N USPEX\n');
fprintf(fp, '#PBS -j oe\n');
%fprintf(fp, '#PBS -V \n');
fprintf(fp, '#PBS -q dell\n'); 【自己服务器上的队列名称】
%fprintf(fp, '#PBS -V \n');
fprintf(fp, '#PBS -S /bin/bash\n');
fprintf(fp, 'cd $PBS_O_WORKDIR\n');
fprintf(fp,'mpirun -np 8 vasp > vasp.out\n');
fclose(fp);
%Step 2: to submit the job with the command like qsub, bsub, llsubmit, .etc.
%It will output some message on the screen like '2350873.nano.cfn.bnl.local'
[a,b]=unix(['qsub run.sh'])
%Step 3: to get the jobID from the screen message
end_marker = findstr(b,'.');
jobNumber = b(1:end_marker(1)-1);
3） 解读：USPEX.m文件
cat USPEX.m
try
unix('chmod +x getStuff');
unix('chmod +x getFromTo');
unix('chmod +x FunctionFolder/USPEX/getStuff');
unix('chmod +x FunctionFolder/USPEX/getFromTo');
unix('chmod +x Specific/getStuff');
unix('chmod +x Specific/getFromTo');
unix('chmod +x FunctionFolder/spacegroup_init/random_2d');
unix('chmod +x FunctionFolder/spacegroup_init/random_cell');
unix('chmod +x FunctionFolder/spacegroup_init/random_cell_mol');
unix('chmod +x FunctionFolder/spacegroup_stokes/findsym_new');
catch
end
[nothing, whichCode] = unix(['./getStuff INPUT.txt calculationMethod 1']);
if isempty(whichCode)
whichCode = 'USPEX';
else
whichCode(end) = [];
end
if strcmpi(whichCode, 'USPEX')
ev_alg();
elseif strcmpi(whichCode, 'VCNEB')
vcNEB();
elseif strcmpi(whichCode, 'META')
metasoft();
end
4） 解读./getStuff
#!/usr/bin/perl
$file = shift;
$string = shift;
$column = shift;
unless(defined($column)) {
print "usage : <file> <string> <column>\n";
exit;
}
open(FILE, $file) or die "Can't open $file : $!\n";
while(<FILE>) {
if(/$string/) {
@a = split /\s+/;
print $a[$column-1] . "\n";
}
}
5）解读INPUT.txt
PARAMETERS EVOLUTIONARY ALGORITHM
******************************************
******************************************
* TYPE OF RUN AND SYSTEM *
******************************************
******************************************
USPEX : calculationMethod (USPEX, VCNEB, META)
300 : calculationType (dimension: 0-3; molecule: 0/1; varcomp: 0/1) 【变胞取1，如取301后，接下来如何处理？】
1 : optType (1=enthalpy, 2=volume, 3=hardness, 4=struc_order, 5=aver_dist)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% What symmetry(s) have to be satisfied by the randomly created structures
% symmetries
2-230
% endSymmetries
% numbers of specices (ions/molecules/blocks) of each type
% numSpecices
6 18
% EndNumSpecices
%%%%%%%%%%%%%%%%%%%%
% Here come the atomic numbers of the atoms involved
% atomType
Sc H
% EndAtomType
%%%%%%%%%%%%%%%%%%%%
% valences【是赝势的价电子数目吗？No,指汉典元素周期表中对应的氧化态。如何改变氧化态？同时具有+2,+3价的元素如何处理？】
3 1
% endValences
******************************************
* POPULATION *
******************************************
30 : populationSize (how many individuals per generation)
100 : initialPopSize
30 : numGenerations (how many generations shall be calculated)
25 : stopCrit
******************************************
% FIRST VAR. COMP. GENERATION *
******************************************
20 : firstGeneMax (how many different compositions for first generation)
8 : minAt (minimum amount of atoms/cell for first generation)
16 : maxAt (maximum amount of atoms/cell for first generation) 【可能需根据实际情况指定？】
【USPEX中如何实现变胞？8，16表达了什么含义？】
******************************************
* SURVIVAL OF THE FITTEST AND SELECTION *
******************************************
0 : reoptOld
0.6 : bestFrac
******************************************
******************************************
* VARIATION OPERATORS *
******************************************
******************************************
0.50 : fracGene (fraction of generation produced by heredity)
0.20 : fracRand (fraction of generation produced randomly from space groups)
0.20 : fracAtomsMut (fraction of the generation produced by softmutation)
0.10 : fracTrans (fraction of the generation produced by transmutations)
****************************************
* CONSTRAINTS *
****************************************
1.5 : minVectorLength ( minimal length of any lattice vector)
%%%%%%%%%%%%%%%%
% IonDistances
1.44 088
0.0 0.32
% EndDistances
% 汉典元素周期表中的共价键长：http://www.zdic.net/appendix/f7.htm
或参考
http://www.webelements.com/plutonium/atom_sizes.html
%%%%%%%%%%%%%%%%
*****************************************
* CELL *
*****************************************
% The following is what you know about the lattice. If you know the lattice
% vectors, type them in as 3x3 matrix. If not, type the estimated volume.
% For variable composition - type the estimated atomic volume for each element.
% Latticevalues (this word MUST stay here, type values below)
177
% Endvalues (this word MUST stay here)
%日本NIMS数据库：http://crystdb.nims.go.jp/crystdb/search-materials
找到对应物质的模型结构，下载cif 文件，通过VESTA看体积数据
*****************************************
* DETAILS OF AB INITIO CALCULATIONS *
*****************************************
% supported: 1-vasp, 2-siesta, 3-gulp, 4-LAMMPS, 5-NeuralNetworks
% 6-dmacrys, 7-cp2k, 8-QuantumEspresso, 9-ASE, 10-ATK, 11-CASTEP
abinitioCode (which code from CommandExecutable shall be used for calculation? )
1 1 1 1 1
ENDabinit
%Resolution for KPOINTS - one number per step or just one number in total)
% KresolStart
0.16 0.12 0.10 0.08 0.06
% Kresolend
% commandExecutable
vasp
% mpiexec -n 8 -machinefile .nodelists vasp > out.vasp
mpiexec -np 8 vasp >log
% EndExecutable
5 : numParallelCalcs (how many parallel calculations shall be performed)
1 : whichCluster (0: no-job-script, 1: local submission, 2: remote)
【matlab语法，‘：’的作用】
submission)
0.010 : toleranceFing (tolerance for identical structures)
*****************************************
* RESTART *
*****************************************
0 : pickUpYN (if pickUpYN~=0 , then a previous calculation will be continued )
0 : pickUpGen (at which generation shall the previous calculation be picked up? If = 0 , then a new calculation is started)
0 : pickUpFolder (number of the results folder to be used. If = 0 , then the highest existing number is taken)
手册摘录解读：Specify the following varcomp-only options:
variable firstGeneMax 【变量 第一代产生的different compositions数量】
Meaning: How many different compositions 【怎样确保产生自己想要的组成？！】are sampled in the first generation. If 0, then the number is equal to initialPopSize/4. For binaries, we recommend firstGeneMax=11, for ternaries a higher value is needed, e.g. 30.
Default: 11
Format:
10 : firstGeneMax
variable minAt
Meaning: Minimum number of atoms (for calculationType=301/201/300) or molecules (for calculationType=311) in the unit cell for the first generation.
Default: No default
Format:
10 : minAt
variable maxAt
Meaning: Maximum number of atoms (for calculationType=301/201/300 or in META calculations) or molecules (for calculationType=311) in the unit cell for the first generation.
Default: No default 【没有默认值，需自己指定】
Format:
20 : maxAt
根据自己服务器提交VASP任务脚本修改的submitJob_local.m文件
function jobNumber = submitJob_local()
%-------------------------------------------------------------
%This routine is to check if the submitted job is complete or not
%One needs to do a little edit based on your own situation.
%1 : whichCluster (default 0, 1: local submission, 2: remote submission)
%-------------------------------------------------------------
%Step 1: to prepare the job script that is required by your supercomputer
fp = fopen('myrun', 'w');
fprintf(fp, '#!/bin/sh\n');
fprintf(fp, '#PBS -l nodes=1:ppn=8,walltime=1:30:00 -q cfn_short\n');
fprintf(fp, '#PBS -N USPEX\n');
fprintf(fp, '#PBS -j oe\n');
fprintf(fp, '#PBS -V \n');
fprintf(fp, '#PBS -q dell \n');
fprintf(fp,'#PBS -S /bin/bash \n');
fprintf(fp, 'cd ${PBS_O_WORKDIR}\n');
fprintf(fp, 'mpiexec -np 8 vasp > vasp.out\n');
fclose(fp);
%Step 2: to submit the job with a command like qsub, bsub, llsubmit, etc.
[a,b]=unix(['qsub myrun'])
%Step 3: to get the jobID from the screen message
%It will output some message on the screen like '2350873.nano.cfn.bnl.local'
end_marker = findstr(b,'.');
jobNumber = b(1:end_marker(1)-1);
为保证程序周期性执行，修改并执行下述脚本，nohup ./run-uspex.sh & ：
cat run-uspex.sh
#!/bin/sh
while true
do
date >> log
/opt/matlab2013/bin/matlab < USPEX.m >>log
sleep 300
done
USPEX问答：
最近在使用USPEX遇到以下两个问题，期盼您的解答：
1) 怎样限制USPEX提交的任务数量，即始终保持只有五个任务在队列中排队或运算，而不是像下述情形一样，有上百个任务在队列中：
50375.mu01 yexq dell vasp 12 -- 7200: R 47:25
50443.mu01 yexq dell USPEX 8 -- 01:30 Q --
50444.mu01 yexq dell USPEX 8 -- 01:30 Q --
50445.mu01 yexq dell USPEX 8 -- 01:30 Q --
50446.mu01 yexq dell USPEX 8 -- 01:30 Q --
50447.mu01 yexq dell USPEX 8 -- 01:30 Q --
50448.mu01 yexq dell USPEX 8 -- 01:30 Q --
50449.mu01 yexq dell USPEX 8 -- 01:30 Q --
50450.mu01 yexq dell USPEX 8 -- 01:30 Q --
50451.mu01 yexq dell USPEX 8 -- 01:30 Q --
50452.mu01 yexq dell USPEX 8 -- 01:30 Q --
50453.mu01 yexq dell USPEX 8 -- 01:30 Q --
......
2) IonDistances离子之间距离，感觉应该是共价半径的加和，不知对否？
Answer from Lirong Han
(1)这个我还真不知到在哪里设置，我用的都是已经写好的脚本，您可以查一下手册，上面应该有写。
(2) 一般设置为离子半径的平均值，实际只要在差不多的范围内就没有关系
From Xing Wang
#!/bin/sh
#PBS -l nodes=1:ppn=12
#PBS -N UZr2_phonon
#PBS -j oe
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/software/compiler/mkl/lib/em64t
cd $PBS_O_WORKDIR
NPROCS=`wc -l < $PBS_NODEFILE`
echo This job has allocated $NPROCS nodes
echo The root node is
uname -n
master=`hostname | cut -f 1 -d '.'`
node_list=`cat $PBS_NODEFILE`
##cat $PBS_NODEFILE | awk '{ print $1"-clust" }' > ./hostfile
cat $PBS_NODEFILE | awk '{ print $1" slots=12" }' > ./hostfile
/home/software/mpi/openmpi1.4.3-intel/bin/mpirun -np $NPROCS -hostfile ./hostfile /home/software/vasp/vasp.5.2.2/vasp >&vasp.log
rm hostfile vasp.log EIGENVAL DOSCAR IBZKPT WAVECAR CHG CHGCAR XDATCAR PCDAT
Thanks to QiangZhu and XInyue CHen
1) 怎样限制USPEX提交的任务数量，即始终保持只有五个任务在队列中排队或运算，而不是像下述情形一样，有上百个任务在队列中：
50375.mu01 yexq dell vasp 12 -- 7200: R 47:25
50443.mu01 yexq dell USPEX 8 -- 01:30 Q --
50444.mu01 yexq dell USPEX 8 -- 01:30 Q --
50445.mu01 yexq dell USPEX 8 -- 01:30 Q --
50446.mu01 yexq dell USPEX 8 -- 01:30 Q --
50447.mu01 yexq dell USPEX 8 -- 01:30 Q --
50448.mu01 yexq dell USPEX 8 -- 01:30 Q --
50449.mu01 yexq dell USPEX 8 -- 01:30 Q --
50450.mu01 yexq dell USPEX 8 -- 01:30 Q --
50451.mu01 yexq dell USPEX 8 -- 01:30 Q --
50452.mu01 yexq dell USPEX 8 -- 01:30 Q --
50453.mu01 yexq dell USPEX 8 -- 01:30 Q --......
（程xy博士：这个好像要修改提交的脚本，具体哪个跟每个超级机有关...
你需要修改uspex根目录下的submission里面的相关文件
这个比较复杂，最好问问ZhuQ博士）2) IonDistances离子之间距离，感觉应该是共价半径的加和，不知对否？
（程xy博士：是共价半径的0.8倍，估计值）
USPEX目录下的文件及文件夹：
ls
AntiSeeds Current_POP.mat getStuff multiple_runs results1 Submission USPEX.m
clean ev_alg.m INPUT.txt nohup.out run-uspex.sh submitJob_local.m uspex_MoSi2
clean_tough FunctionFolder log out.uspex Seeds submit.sh
Current_ORG.mat getFromTo metasoft.m out.uspex2 Specific test
运行后产生Cal*等文件夹。
变胞方法： 参考uspex手册
https://m.sciencenet.cn/blog-567091-837003.html
上一篇：bader电荷分析再谈
下一篇：声子谱虚频的另类情况及如何提高声子计算速度
