AUTOSELECT
Introduction :
In order to run my job automatically, I modified the code of EMPFT_SELECT. The modified programs are named as "autoselect". I also parallelized the AUTOSELECT. Therefore, the single node version called autoselet. The parallelized version called autoselect.para. The input of the program is very simple. The first input is the sigma cutoff of of the selection criterion. The selection equation is any particles's CC (corelation coefficient) is higher than CC_average + input_cutoff * CC_sigma will be write out into the output file. After the input sigma cutoff, you just input the filenames that you want to select and a return will be the end of the input. If the name of the original data file is "XXXX.dat_001", the programs of autoselect will save the the selected particle into a file named "XXXXsel.dat_001". The paralized autoselect.para will make all the nodes to select files parallelly which will reduce the seletion time a lot if you have many files. If only inputing one file each time, only one node will work for the selection.
Example scripts for running AUTOSELECT:
- Input all the file at the same time
- Input the file one by one
#! /bin/csh -f
set wdir = "/bio/new-york/rbernal/Cryo"
set binEM = "/home/dart/u19/xiaoc/bin-EM"
set prog_version = ".para"
cd ${wdir}/Data
date >! ${wdir}/Logs/sel${prog_version}_$1.log
${binEM}/autoselect${prog_version} << EOFSEL >> ${wdir}/Logs/sel${prog_version}_$1.log
0.9
${wdir}/Data/6164Bsp.dat_${zeros}$1
${wdir}/Data/6165Bsp.dat_${zeros}$1
${wdir}/Data/6166Bsp.dat_${zeros}$1
${wdir}/Data/6167Bsp.dat_${zeros}$1
${wdir}/Data/6168Bsp.dat_${zeros}$1
${wdir}/Data/6169Bsp.dat_${zeros}$1
${wdir}/Data/6172Bsp.dat_${zeros}$1
EOFSEL
date >> ${wdir}/Logs/sel${prog_version}_$1.log
#! /bin/csh -f
set wdir = "/bio/new-york/rbernal/Cryo"
set binEM = "/home/dart/u19/xiaoc/bin-EM"
set prog_version = ".para"
cd ${wdir}/Data
date >! ${wdir}/Logs/sel${prog_version}_$1.log
foreach datafile ( *Bsp.dat_${zeros}$1 )
${binEM}/autoselect${prog_version} << EOFSEL >> ${wdir}/Logs/sel${prog_version}_$1.log
0.9
${wdir}/Data/${datafile}
EOFSEL
end
date >> ${wdir}/Logs/sel${prog_version}_$1.log
Known problem :
When using the parallized version, all the nodes will print out the selection or seperation information simultaneously which will make the log file unable to read. Therefore, if you want to look at the selection information. You can input file one by one.BUG REPORT :
If you meet any bugs or problems of the programs, please email xc@purdue.edu.