STARTING INITIAL PATH JOBS WITH PATHSAMPLE

From Docswiki
Jump to navigation Jump to search

This file describes how to perform initial connection jobs with PATHSAMPLE. There are several options depending on the potential that is being used. These instructions assume that you do not already have a path.info file or min.data.info files from OPTIM. If you do have these files, then you can use STARTFROMPATH for a path.info file, or ADDMIN if you have min.data.info.

A. Potentials that DO NOT use start files

Step 1. First generate min.data.info.1 for endpoint 1:

~/svn/OPTIM/builds/ifort_64_2019_5_281/OPTIM 1

Note: the extension 1 refers to the file odata.1 that must be present in the directory.

Step 2. Then, generate similarly for endpoint 2:

~/svn/OPTIM/builds/ifort_64_2019_5_281/OPTIM 2

Note: the extension 2 refers to the file odata.2 that must be present in the directory.

The odata.1 and odata.2 files must contain OPTIM minimisation commands + DUMPDATA and ENDHESS commands to generate the min.data.info.* files. For example, an odata.1 file (that uses the xtb potential) could be of the form:

odata.1:

XTB C24xtb '--acc 0.1 --gfn2 -P 8'
UPDATES 100 100
BFGSCONV 1.0D-5
BFGSMIN 1.0D-5
CONVERGE 1.0D-5
BFGSSTEPS 60000
DEBUG

DUMPDATA
ENDHESS

MAXERISE 1.0D-4 1.0D0

DUMPALLPATHS
STEPS 400

POINTS
C -2.04053148527331 1.05306390371634 -0.253877546750764E-11
C -1.05306390370855 0.172720191240289E-11 -2.04053148528364
C -0.546121488999475E-12 2.04053148529656 -1.05306390369997
C -0.146298180312409E-11 2.04053148530069 1.05306390369198
C -0.914160192067804E-12 -2.04053148529553 -1.05306390370031
C -0.287393036865366E-11 -2.04053148530158 1.05306390369083
C -2.04053148527466 -1.05306390371639 -0.367973225662900E-11
C 2.04053148526781 -1.05306390372175 -0.875778282666703E-12
C -1.05306390372231 -0.979254867490875E-12 2.04053148526605
C 2.04053148527107 1.05306390372396 -0.182101538055068E-11
C 1.05306390370932 -0.340215830916567E-12 2.04053148528387
C 1.05306390370268 0.258658422685307E-13 -2.04053148528689
C 0.710799972739704E-12 1.05306390368809 -2.04053148527393
C 1.05306390371174 -2.04053148526187 0.955967560634191E-12
C -2.04053148525347 0.881144607755502E-13 1.05306390371966
C 0.344943596544608E-11 1.05306390367851 2.04053148528088
C 2.04053148526428 0.762070953548885E-12 1.05306390369870
C 2.04053148526827 0.865526145990094E-13 -1.05306390369420
C -2.04053148526448 0.713448152766709E-12 -1.05306390370007
C -1.05306390370763 2.04053148526678 0.446469677757191E-11
C 1.05306390371344 2.04053148526008 0.938221027950337E-13
C 0.419813438210074E-11 -1.05306390367943 2.04053148528005
C -1.05306390370769 -2.04053148526729 0.524101153866833E-11
C 0.938323209745303E-12 -1.05306390368924 -2.04053148527485

Step 3. Once the two min.data.info.1 and min.data.info.2 files are generated, they are copied to the pathsample directory where they are then combined into one file

cat min.data.info.1 min.data.info.2 > min.data.info

Step 4. The combined file is read with the READMIN command in pathdata and then PATHSAMPLE generates the database files min.data and points.min from which a DIJINITSTART/CONNECTREGION/DIJINITCONT job can be launched. For the above purpose, a pathdata file (that uses the xtb potential) could be of the form:

pathdata:

GCORES '-c8'
SYSTEM X
EXEC ~/svn/OPTIM/builds/ifort_64_2019_5_281/OPTIM
NATOMS 24
SEED 1
DIRECTION AB
CONNECTIONS 1
TEMPERATURE 0.592
PLANCK 9.536D-14

READMIN min.data.info


Note: Before the next step, we should manually create min.A and min.B files to designate the indices of the endpoints.

Step 5. Finally, to launch the initial connection attempt with PATHSAMPLE the following pathdata and odata.connect files can be used (xtb potential):

pathdata:

GCORES '-c8'
SYSTEM X
EXEC ~/svn/OPTIM/builds/ifort_64_2019_5_281/OPTIM
NATOMS 24
SEED 1
DIRECTION AB
CONNECTIONS 1
TEMPERATURE 0.592
PLANCK 9.536D-14

! PERMDIST

EDIFFTOL 1.0D-4
GEOMDIFFTOL 0.3D0

SLURM
! CPUS 10

! READMIN min.data.info

PAIRLIST 1
DIJINITCONT exp
INITIALDISTANCE -1.0
CYCLES 10

! CONNECTREGION 13 11 5.0D0
! CYCLES 10

COPYFILES mass
COPYOPTIM


odata.connect:

XTB C24xtb '--acc 0.1 --gfn2 -P 8'
! DEBUG
NEWCONNECT 1 1 2.0 25.0 10 1.0 0.001
NEWNEB 30 500 0.0001
NEBK 100.0

DIJKSTRA EXP
! PERMDIST

EDIFFTOL 1.0D-4
GEOMDIFFTOL 0.3D0

BFGSMIN 1.0D-6
UPDATES 100 100 50 4
PUSHOPT 0.1 0.001 100
REOPTIMISEENDPOINTS
BFGSSTEPS 2000
MAXBFGS 0.3 0.3
STEPS 1000
RADIUS 999999.9

! comment Perform a hybrid EF/LBFGS transition state search
BFGSTS 2000 20 100 0.0001
! CHECKNEGATIVE
! ENDHESS
NOHESS
MAXTSENERGY 10.0D0
MAXSTEP 0.1
MAXMAX 0.3
TRAD 0.2
USEDIAG 2
PATH 100 0.0001

MAXERISE 1.0D-3 2.0D0
DUMPALLPATHS
DUMPSTRUCTURES
POINTS


B. Potentials that use start files

Step 1. A file called start.1 should contain the xyz coordinates of min1. Another file start.1.1 should be a copy of start.1

Step 2. A file called start.2 should contain the xyz coordinates of min2. Another file start.2.2 should be a copy of start.2

Note: the atomic symbols/masses should also be passed to PATHSAMPLE/OPTIM and this is done differently for each potential. Refer to potential-specific documentation.

For example:

In Tapered ReaxFF, the elements are passed from fort.3 file that must be present. The masses are encoded internaly per-element in the code, so no need to pass them.

In DFTBP, the elements are passed from the dftb_in.hsd file that must be present. The masses are encoded internaly per-element in the code, so no need to pass them.

In xtb, the elements and masses are passed from mass file that must be present.

Step 3. odata.start and odata.finish should contain OPTIM minimisation commands for min1 and min2, respectively. For example, an odata.start file (that uses the DFTBP potential) could be of the form:

odata.start:

UPDATES 100 100
BFGSCONV 1.0D-6
BFGSMIN 1.0D-6
CONVERGE 1.0D-6
BFGSSTEPS 60000
DEBUG
DUMPDATA
ENDHESS
MAXERISE 1.0D-4 1.0D0
DUMPALLPATHS
STEPS 400
DFTBP start

Step 4. odata.connect should contain the OPTIM connection commands. For example, an odata.connect file (that uses the DFTBP potential) could be of the form:

odata.connect

! DEBUG
NEWCONNECT 1 1 2.0 25.0 10 1.0 0.001
NEWNEB 30 500 0.0001
NEBK 100.0

DIJKSTRA EXP
! PERMDIST

EDIFFTOL 1.0D-4
GEOMDIFFTOL 0.3D0

BFGSMIN 1.0D-6
UPDATES 100 100 50 4
PUSHOPT 0.1 0.001 100
REOPTIMISEENDPOINTS
BFGSSTEPS 2000
MAXBFGS 0.3 0.3
STEPS 1000
RADIUS 999999.9

! comment Perform a hybrid EF/LBFGS transition state search
BFGSTS 2000 20 100 0.0001
! CHECKNEGATIVE
! ENDHESS
NOHESS
MAXTSENERGY 10.0D0
MAXSTEP 0.1
MAXMAX 0.3
TRAD 0.2
USEDIAG 2
PATH 100 0.0001

MAXERISE 1.0D-3 2.0D0
DUMPALLPATHS
DUMPSTRUCTURES
DFTBP start

Step 5. pathdata file should contain the DIJINITSTART keyword to start an initial connection job and produce the database files (min.data, ts.data, points.min and points.ts).

For example, a pathdata file (that uses the DFTBP potential) could be of the form:

pathdata:

EXEC ~/svn/OPTIM/builds/ifort_64_2019_5_281/DFTBPOPTIM
NATOMS 32
SEED 1
DIRECTION AB
CONNECTIONS 1
TEMPERATURE 0.592
PLANCK 9.536D-14

! PERMDIST
! LPERMDIST 3 0.2 6.0 0.001
! RANROT 1600

EDIFFTOL 1.0D-4
GEOMDIFFTOL 0.3D0

SLURM
! CPUS 10

DIJINITSTART exp
INITIALDISTANCE -1.0
CYCLES 1

! PAIRLIST 1
! DIJINITCONT exp
! INITIALDISTANCE -1.0
! CYCLES 1

! ADDMIN min.data.info

! CONNECTREGION 13 11 5.0D0
! CYCLES 10

DFTBP
COPYFILES *.skf dftb_in.hsd
COPYOPTIM

Note: after a few connections have been made already (and with the databse present), it is possible to stop the job, and replace DIJINISTART with DIJINITCONT, which will continue the connection attempts and add new stationary points to the database.

Retrieved from "http://wikis.ch.cam.ac.uk/ro-walesdocs/wiki/index.php?title=STARTING_INITIAL_PATH_JOBS_WITH_PATHSAMPLE&oldid=1727"