runSim_in_background

#!/bin/bash

#for filename in $raspa_folder/share/raspa/structures/cif/$COFfile.cif 	

		# ------------------------------------------------------------------#
		# runSim
		# THIS IS A SCRIPT TO RUN SCREENING WITH RASPA           	        #
		# ------------------------------------------------------------------#


# 	The script requires these folders
#
#	> start : 		containing simulation files 
#	> parameters:	containing parameters for the screening: adsorbate_list and cof_to_be_run
#	> dependencies:	containing everything needed, but to not be changed, maybe write-protected. 
#	> log:			containing all the log files of the screening
#
# 	If restarts are needed, you place them in start/RestartInitial
#
# 
#	runBatchOfSimulations.sh -p [widom/isotherm/water-filling] -r -o -x -c 12
#
#	-p flag	<arg>	REQUIRED	(protocol) 			to pick the right protocol 
# 	-r flag			OPTIONAL 	(RestartInitial) 	use the restart file. DEFAULT is NO
#	-o flag			OPTIONAL	(overwrite) 		overwrite the folder / simulation if it already exists. DEFAULT is NO
#	-x flag			OPTIONAL 	(exectute) 			if selected the run file is executed and the simulation starts.  DEFAULT is NO
#	-h flag	     	OPTIONAL	(help)				help
#	-c flag	<arg>	OPTIONAL	(CPUs)				number of CPUs (on w7991)
#	-t flag	<arg>	OPTIONAL	(runtime)			in hours (on Eddie)




		# ------------------------------------------------------------------#
		# FOLDERS									       			        #
		# ------------------------------------------------------------------#

# Retrieve environmental variables
source /home/s2216474/toSiSeMatDatastore/local/PATH # on Eddie
source /localdisk/PATH # on w7991


#raspa_folder=~/RASPA/simulations/  								# Just for testing, comment otherwise. ##############################################

folder=$(pwd) 													# Local folder
cif_databank=$raspa_folder/share/raspa/structures/cif/  		# Might vary if you compiled RASPA in a fancy way.	

if [ -e ./log ]; then 
	echo "Log found."			# Check if dependencies are there
	else
	printf "\nERROR: The log folder not found. Are you in the right folder?\n"
	pwd
	break
	fi
	

log_folder=$folder/log							# Log folder
dependencies_folder=$repository_folder/dependencies

		# ------------------------------------------------------------------#
		# FILENAMES									       			        #
		# ------------------------------------------------------------------#
		
cof_list=./parameters/cof_to_be_run.txt					#
adsorbate_list=./parameters/adsorbate_list.txt			#
input_file=./parameters/input.sh						#
log_file=$log_folder/$(date +%F_%T).log					# Log file name with the date and time of execution



		# ------------------------------------------------------------------#
		# FUNCTIONS, CHECKING EXISTANCE AND LOADING INPUT FILES 	        #
		# ------------------------------------------------------------------#

# Loading some essential functions.
#
#	log											print with printf and save it in the log file
#	
#	checkCPUs $CORES_TO_BE USED   $SECONDS		check if there are CPUs avaiable

clear															# Clean the screen	


if [ -e $dependencies_folder/functions_box.sh ]; then  			# Check if dependencies are there
	source $dependencies_folder/functions_box.sh		
	else
	printf "\nERROR: Dependencies not found. Please check $dependencies_folder\n"
	exit
	fi


log $(date +%c)														# Date and time at start for the log.						
ls $cif_databank | awk -F'.cif' '{print $1}' | sort  > database.txt # This is the database of COFs in RASPA.

if [ -e $input_file ]; then  											# Check if the list of COFs is there
	log $line_break
	################################### Force to be UNIX format
	source $input_file 
	log "\n- Files input.sh found and loaded."										
	
	else
	
	log $line_break
	log "\nERROR File input.sh. You can create it. \n"
	exit
	fi

if [ -e $adsorbate_list ]; then  	# Check if the list of COFs is there
	log $line_break
	log "\n- Reading the adsorbate list:\n\n"				
	################################### Force to be UNIX format
	#more $adsorbate_list; 
	#more $adsorbate_list 					>> $log_file  
	
	# Lines in common between the database and the input: THE .cif EXISTS
	else	
	log "\nERROR File adsorbate_list not found.\n"
	exit
	fi

if [ -e $cof_list ]; then  	# Check if the list of COFs is there
	log $line_break
	log "\n- Reading the input file (cof_to_be_run.txt)."						# sort and remove duplicates
	################################### Force to be UNIX format
	sort -o $cof_list $cof_list 	
	sort -o $cof_list $cof_list 			>> $log_file
	log $line_break
	log "\n- Files NOT found in the database:\n\n"
	# Lines only in the input: THE .cif DOES NOT EXIST
	comm -13 database.txt $cof_list 
	comm -13 database.txt $cof_list 		>> $log_file
	log $line_break
	log "\n- Files found in the database:\n\n"
	# Lines in common between the database and the input: THE .cif EXISTS
	comm -12 database.txt $cof_list
	comm -12 database.txt $cof_list 		>> $log_file
	log $line_break
	else
	log "\n- File with COF not found. You can create it doing:"
	log "\n  ls <CIF-FOLDER>/*.cif > cof_to_be_run.txt"
	exit
	fi
	
if	[ -e 'start' ]; then 			# Check if start folder is there, otherwise you create it
	log "\n- Folder start exist. This folder supposedly contains force fields files.\n"
	ls -l start
	ls -l start 							>> $log_file
	else
	log "\nPlease move in the folder named 'start' input and forcefield files."
	mkdir start
	exit
	fi

if	[ -e './start/RestartInitial' ]; then # Check if RestartInitial folder is there, otherwise you create it
	log $line_break
	log "\n- Folder RestartInitial exist."
	else
	mkdir ./start/RestartInitial
	log "\nPlace the Restart Files in here." 
	log $line_break
	fi
	

		# ------------------------------------------------------------------#
		# FLAGS DECLARATION							       			        #
		# ------------------------------------------------------------------#
help=no

while getopts p:roxhc:t: flag 
	do
		case "${flag}" in
			p) protocol=${OPTARG};;			# Pick the protocol
			r) RestartInitial='yes';;		# Use the restart initial			We need a restart!!!!!!
			o) overwrite='yes';;			# Overwrite the folder
			x) execute='yes';;				# Execute
			h) help='yes';;
			c) NumberOfCPUs=${OPTARG};; 	# Number of CPUs
			t) maxtime=${OPTARG};;
		esac
	done


		# ------------------------------------------------------------------#
		# CASE TO SET-UP VARIABLES ACCORDING TO THE PROTOCOL SELECTED       #
		# ------------------------------------------------------------------#
echo $help

if [ $help = 'yes' ] ; then 
clear
more $dependencies_folder/help.dat
exit; fi




case $protocol in

  'widom')
    conditions_list=$folder/parameters/widom.protocol ;
    ;;

  'isotherm')
    conditions_list=$folder/parameters/isotherm_concentrations.protocol ; 
    ;;

  'hydration')
    conditions_list=$folder/parameters/hydration_pressure.protocol ;
    ;;

  'BET')
    conditions_list=$folder/parameters/BET_pressure.protocol; Temperature=77;
    ;;

  'water-filling')
    conditions_list=$folder/parameters/water-filling.protocol ;
    ;;
	
  'charges')
    conditions_list=$folder/parameters/charges.protocol ;
    ;;
	
  *)
    log "\nERROR No protocol selected.\n"				# If none of the procols are recognized, the script interrupts.
	exit 
    ;;
esac

log $line_break

#log "\n - Condition file selected looks like this."
#log "\n$conditions_list\n"

more $conditions_list

		# --------------------------------------------------#
		# PRINT THE MAIN PARAMETER TO AVOID SURPRISES       #
		# --------------------------------------------------#
		
log $line_break
log "\nPROTOCOL:                        $protocol"
log "\nNumber Of Adsorbate Molecules:   $NumberOfAdsorbateMolecules" # Initial adsorbate mol
log "\nSwap Probability of Adsorbate:   $SwapProbabilityofAdsorbate"
log "\nNumber Of Cycles                 $NumberOfCycles"
log "\nNumber Of Initialization Cycle:  $NumberOfInitializationCycles"
log "\nNumber of Cores:                 $NumberOfCPUs"
log "\nTemperature:                     $Temperature"
log "\nPressure:                        $Pressure"
log "\n"
if [ $overwrite = 'yes' ] ; then 
log "\nWARNING: Careful, you might overwrite what is already there."; fi

if [ $RestartInitial = 'yes' ] ; then 
log "\nWARNING: Please check, you might not have all the Restart Files in ./start. "; fi

if [ $execute = 'yes' ] ; then 
log "\nWARNING: In this way, you are running the simulations. "; fi

log "\n"
log "\nYou run simulations for $(wc -l < $cof_list) structures, $(wc -l < $adsorbate_list) adsorbates and $(wc -l < $conditions_list) conditions each."
log $line_break
log "\n"

#read -p "Press enter to continue. CTRL+C to break. If nothing happens I will wait 50s and then I will proceed." YESorNO			# Give the chance to let the user to check and wait before moving forward.
#wait
sleep 2
log "Starting...\n"
sleep 2

		# -----------------------------------------------------------#
		# NESTED LOOPS OVER CONDITIONS, ADSORBATES, CIF STRUCTURES   #
		# -----------------------------------------------------------# 

#	Nested directories are created as follow:	
#	
#	$filename > $adsorbate > $variable > Output,Movies,Restart (run,simulation.input,force_field) > System_0 > (files)		
#
#	$filename 	is COFname
#	$adsorbate 	is the elements of $adsorbate_list (e.g. urea_KBFF, charge)
#	$variable 	is the composition in the protocol named 'isotherm' or the name of the protocol in all the others
		
cd $folder
		
		
while read COFname; 			# Read the names of the COF to be simulated from the file 'cof_to_be_run.txt'
do

filename=$cif_databank$COFname.cif

if [ -e $filename ] ; then  #### The .cif file existOther conditions.

	#checkCharges $filename   # It check if charges are ok. 
	#checkComposition $filename
	#### ANY CHECK ON THE STRUCTURE NEED TO BE DONE HERE #####################
	#
	#	> CHECK CHARGES, PRINT HISTOGRAM, CHECK CHARGE NEUTRALITY, CHECK IF THE CHARGE IS ZERO
	#	> zeo++
	#	> COMPOSITION
	#	> XRD
	#
	##########################################################################

    mkdir -p $COFname  								# Create a folder with .cif file name.
	cd $folder/$COFname										# Move into it.
	

####################################################################################################################

while read adsorbate; 
do

	mkdir -p $adsorbate
	cd $adsorbate
		
	# --------------------------------------------------#
	# LOOP OVER CONDITION OF SIMULATIONS                #
	# --------------------------------------------------#
	
####################################################################################################################
	
	while read variable; 
	do
		
		############## CHECK IF THERE IS ANY CPU THAT IS FREE. #########################################################
		
		checkCPUs $NumberOfCPUs $CheckEvery # Hold on if CPUs are busy.
		
		folder_name=$( echo $COFname.$Temperature.$adsorbate.$variable ) 	# Folder name in ./COFname/ Delimiter here is a PITA
		
		if [ -d $folder_name ] && [ $overwrite = no ] ; then 	# Prevent overwriting 
		log  '\n> %-40s exists.\tI skip this file.' $folder_name 	# Print if a folder is already there.
		

		
		else
		log  '\n> %-40s Folder created. ' $folder_name 	# Print if a folder is already there.

		# --------------------------------------------------#
		# COMPUTE HOW MANY UNIT CELLS ARE REQUIRED          #
		# --------------------------------------------------#    

		# Getting angles of unit cells
		alpha=$(grep _cell_angle_alpha  $filename  | awk '{print $2}')
		beta=$(grep _cell_angle_beta    $filename  | awk '{print $2}')
		gamma=$(grep _cell_angle_gamma  $filename  | awk '{print $2}')
		# Getting lenght of the cell
		a=$(grep _cell_length_a        $filename    | awk  '{print $2}')
		b=$(grep _cell_length_b        $filename    | awk  '{print $2}')
		c=$(grep _cell_length_c        $filename    | awk  '{print $2}')

		# How big the unit cell needs to be
		UnitCellA=$(echo " $cutoff*2/$a/s($beta*3.1415926535/180)/s($gamma*3.1415926535/180)+1" | tr -d $'\r' | bc -l )
		UnitCellB=$(echo " $cutoff*2/$b/s($alpha*3.1415926535/180)/s($gamma*3.1415926535/180)+1"  | tr -d $'\r' | bc -l )
		UnitCellC=$(echo " $cutoff*2/$c/s($alpha*3.1415926535/180)/s($beta*3.1415926535/180)+1" | tr -d $'\r' | bc -l )

		## Copy the CIF with partial charges in the main folder.
		#cp ./charge_equilibration/$COFname/Movies/System_0/Framework_0_initial_1_1_1_P1.cif   

		mkdir -p $folder_name
		cp -r $folder/start/* $folder_name 
		cd $folder_name

		
		# --------------------------------------------------#
		# WRITE THE INPUT FILE FOR RASPA                    #
		# --------------------------------------------------#  
    
		#log "Writing simulation input file!"
    

    echo "SimulationType              MonteCarlo
NumberOfCycles                  $NumberOfCycles
NumberOfInitializationCycles    $NumberOfInitializationCycles
PrintEvery                      1000
RestartFile					    $RestartInitial

Forcefield                      Local
UseChargesFromCIFFile           yes

Framework                       0
FrameworkName                   $COFname
ShiftUnitCells                  0.0 0.0 0.0
UnitCells                       ${UnitCellA%.*} ${UnitCellB%.*} ${UnitCellC%.*}" > ./simulation.input  # It overwrite any simulation file that was there. That's wanted.


if [ $protocol = 'widom' ]; then 			# Generate the simulation.input for Widom protocol.

	echo "ExternalTemperature             $Temperature" >> ./simulation.input
	echo "component 0         MoleculeName                     $adsorbate
						MoleculeDefinition               Local
						FugacityCoefficient	             1
						WidomProbability                 1
						CreateNumberOfMolecules          $NumberOfAdsorbateMolecules" >> ./simulation.input

fi

if [ $protocol = 'isotherm' ]; then 		# Generate the simulation.input for isotherm protocol.

	MolFraction=$variable
	echo "ExternalTemperature             $Temperature" >> ./simulation.input
	echo "ExternalPressure		$Pressure
						
						component 0         MoleculeName                     $adsorbate
						MoleculeDefinition               Local
						FugacityCoefficient	             1
						MolFraction                      $MolFraction
						TranslationProbability           1
						RotationProbability              1
						ReinsertionProbability           0.5
						SwapProbability                  $SwapProbabilityofAdsorbate
						CreateNumberOfMolecules          $NumberOfAdsorbateMolecules" >> ./simulation.input
		
		
	echo "component 1 		MoleculeName                     SPC-E
						MoleculeDefinition               Local
						FugacityCoefficient              1
						MolFraction                      $(echo "1-$MolFraction" | bc -l)
						TranslationProbability           1
						RotationProbability              1
						ReinsertionProbability           0.5
						SwapProbability                  1.0
						CreateNumberOfMolecules          $NumberOfWaterMolecules" >> ./simulation.input
						
	fi

if [ $protocol = 'hydration' ]; then 		# Generate the simulation.input for isotherm protocol.

	Pressure=$variable
	echo "ExternalTemperature             $Temperature" >> ./simulation.input
	echo "ExternalPressure		$Pressure
						component 0         MoleculeName                  SPC-E
						MoleculeDefinition               Local
						TranslationProbability           1
						RotationProbability              1
						ReinsertionProbability           0.5
						SwapProbability                  0.5
						FugacityCoefficient              1
						CreateNumberOfMolecules          $NumberOfAdsorbateMolecules" >> ./simulation.input

# To update the MolFraction
fi

if [ $protocol = 'BET' ]; then 		# Generate the simulation.input for isotherm protocol.

	Pressure=$variable
	echo "ExternalTemperature             77" >> ./simulation.input
	echo "ExternalPressure		$Pressure
						
						component 0         MoleculeName                  N2_TRAPPE
						MoleculeDefinition               ExampleDefinitions
						TranslationProbability           1
						RotationProbability              1
						ReinsertionProbability           0.5
						SwapProbability                  0.5
						CreateNumberOfMolecules          $NumberOfAdsorbateMolecules" >> ./simulation.input

# To update the MolFraction
fi
	
if [ $protocol = 'water-filling' ]; then 	# Generate the simulation.input for water filling protocol.

echo "ExternalTemperature             77" >> ./simulation.input

echo "component 0         MoleculeName                     $adsorbate
            		MoleculeDefinition               Local
                    FugacityCoefficient	             1
                    TranslationProbability           1
                    RotationProbability              1
                    ReinsertionProbability           0.5
                    SwapProbability                  0
                    CreateNumberOfMolecules          $NumberOfAdsorbateMolecules" >> ./simulation.input
	
	
echo "component 1 		MoleculeName                     SPC-E
            		MoleculeDefinition               Local
                    FugacityCoefficient              1
                    MolFraction                      1
                    TranslationProbability           1
                    RotationProbability              1
                    ReinsertionProbability           0.5
                    SwapProbability                  1.0
            		CreateNumberOfMolecules          $NumberOfWaterMolecules" >> ./simulation.input

fi

if [ $protocol = 'charges' ]; then 			# Generate the simulation.input for charge assignment with Randall-Snurr method.

	echo "ExternalTemperature             77" >> ./simulation.input

    echo "SimulationType              MonteCarlo
NumberOfCycles                  0
NumberOfInitializationCycles    0
PrintEvery                      100
RestartFile					    no

Forcefield                      Local
ChargeFromChargeEquilibration   yes
ChargeEquilibrationPeriodic     yes
ChargeEquilibrationEwald        yes
SymmetrizeFrameworkCharges      no

Framework                       0
FrameworkName                   $COFname
ShiftUnitCells                  0.0 0.0 0.0
UnitCells                       ${UnitCellA%.*} ${UnitCellB%.*} ${UnitCellC%.*}" > ./simulation.input


fi



		if [ $machine = 'Eddie' ]; then # They need to stay out in case you don't submit the job straight away.
						
			# Tailor our run file
			sed -i "s/jobname/COF_$folder_name/g" run_Eddie  #Make the job on the cluster have a decent name
			sed -i "s/#$ -l h_rt=100:00:00/#$ -l h_rt=$(echo $maxtime):00:00/g"  run_Eddie  #Make the job on the cluster have a decent name
									
		
			if [ $execute = 'yes' ]; then 
				#submit job
				qsub run_Eddie 				## Run RASPA in background only if the -x flag was used.
				fi
		 fi
		 
		 
		 if	 [ $machine = 'w7991' ] && [ $execute = 'yes' ] ; then
			
			./run > RASPA_output_file 2>&1 &   	## Run RASPA in background only if the -x flag was used.
			log " Simulation launched."
			
			chmod -R 777 $folder
			
			fi

		
		cd $folder/$COFname/$adsorbate		# Out of the condition folder
		fi								    # End of if condition of overwriting and existance of the .cif file. It need to be here, because we want to check that the condition is not overwritten, not the whole folder.
		

		
		done < $conditions_list					# Repeat again for a different concentration.
	
		cd $folder/$COFname								# Out of the adsorbate folder

done < $folder/$adsorbate_list	# Loop over the adsorbate		

cd $folder									# Go Out of the COF folder


else						# If the .cif file does not exists / other conditional on the file
log "\n> %-40s .cif file seems to not exists. ##################" $COFname
           
fi

done < $folder/$cof_list	# Loop over COFs, the file to be read is used as a list of names.

sleep 2s

log $line_break
log "\nDone.\n"

date +%c

rm database.txt