Dear All
I am running a FSSH MD for a total time of ~1ps but cannot request enough time on cluster to complete the job, so I want to continue from where it ended. On the manual there is a control option fssh_continue that allows restart from previous run using the 396.0 file.
I copied the 396.0 from scratch directory and used $molecule read 396.0 $end to read in the file but output file shows Could not find $molecule section in MoleculeInput::read. Any help would be appreciated, thank you
Can you post your whole input file? That will help in assisting you with this issue.
If you have:
$molecule read $end
in your input then it reads the geometry from the scratch folder. So you need to make sure that the molecule file is located in the scratch folder as well as 396.0
Thank you for your reply. The first paste is the initial .inp file used (to start the calculation) and the second is the .inp used to continue job. The first part runs without problem.
$molecule
!the geometry of my molecule
$end
$rem
JOBTYPE freq
METHOD hf
basis 6-31g*
$end
@@@
$molecule
read
$end
$rem
jobtype aimd
exchange cam-b3lyp
max_scf_cycles 150
max_cis_cycles 100
scf_algorithm diis_gdm
max_diis_cycles 100
basis 6-31g*
cis_n_roots 2
symmetry off
sym_ignore true
cis_singlets true
cis_triplets false
proj_transrot true
fssh_lowestsurface 0
fssh_nsurfaces 2
fssh_initialsurface 1
afssh 0
cis_der_couple true
aimd_steps 10331
time_step 40
aimd_short_time_step 2
aimd_time_step_conversion 1
aimd_print 1
aimd_init_veloc zpe
aimd_integration vverlet
$end
####################################################
$molecule
read ./396.0
$end
$rem
jobtype aimd
fssh_continue 1
exchange cam-b3lyp
max_scf_cycles 150
max_cis_cycles 100
scf_algorithm diis_gdm
max_diis_cycles 100
basis 6-31g*
cis_n_roots 2
symmetry off
sym_ignore true
cis_singlets true
cis_triplets false
proj_transrot true
fssh_lowestsurface 0
fssh_nsurfaces 2
fssh_initialsurface 1
afssh 0
cis_der_couple true
aimd_steps 10331
time_step 40
aimd_short_time_step 2
aimd_time_step_conversion 1
aimd_print 1
aimd_init_veloc zpe
aimd_integration vverlet
$end
#The 396.0 file is in both the scratch and current job submission directory
Perhaps i need to copy other files into my current directory?
Thank you again for any help, I really appreciate it
Remove the 396.0 line from the $molecule section. The 396.0 is a scratch file it isn’t what is read in the $molecule section. Just have the file in the scratch folder that you are using for the job. The read in the $molecule is a specific input while read ./396.0 is a unknown input for this section.
Hi
So i removed ./396.0 but the same error shows. I am using slurm queue and realized that the $SCRATCH directory is assigned to a different node each time. Could that be the reason for the error? I am still new to computing on clusters so I apologize in advance for any confusion
Thank you all again for your help
Can you put an example of your command line that you are using to run the Q-Chem job?
qchem infile outfile savename should be the format you use for this type of job. The savename should be the path to your scratch folder. So for example, the way this series of jobs would look like is first you need to save the scratch folder from the first input qchem -save input1 output1 scratch1. Then you will run the second job using the first scratch folder, qchem -save input2 output scratch1. The -save isn’t required at this step, but it will override everything in the scratch folder then clean it up by removing some scratch files. A modification to the second step could be to create a copy of scratch1 cp -rf scratch1 scratch2 , this will automatically put all the saved scratch files into the secondary scratch folder and keep the first one for future uses. Then the command for the second job would now look like qchem -save input2 output2 scratch2 if you create a copy. If you do it this way you can also speed up your overall calculation by using SCF_GUESS READ so that way you will save on SCF cycles since you have already converged the energy in the first job.
In addition, I was looking over your input again, and noticed the
section of your input. Is this correct with the “#” symbols? If that is the case to run batch jobs you need to separate each input with “@@@”. Or are you just using that to describe the second input which is the FSSH_CONTINUE job which you want to restart the calculation from?