Dear users,
I was running a simple lj fluid simulation in NVT ensemble. Everything was going fine but suddenly I am unable to generate output of any kind. It is not just that the output file is coming out empty.
The log.lammps file that is generated during simulation is not generated at all. I am not sure where to look.
I am running it on my laptop using wsl.
If anyone has any idea, please let me know.
Chaitanya.
Hi @chaitu,
It is very hard to help you without more info on the commands you use but as a rule of thumb, in this kind of situation it is good to first make sure that the disk space given to your Linux partition is not filled. MD simulations are easily greedy in disk space and this is a common caveat.
If this is not the case then you should:
My guess is that LAMMPS was invoked with a command line similar to this:
lmp -in input.in -l output.log
Otherwise, the command log output.log is present at the beginning of your input script.
Thanks Germain and Otello,
The command I am using to run lammps is 
~/path/to/lmp_mpi < in.lmp > output.lmp
The other details are as follows:
I am not sure what is the configuration of wsl or how to get that information. Can you please tell me?
The WSL version I am using is 0.2.1
The following is the input file, in.lmp, that was working last. But it is not working now.
## created by chaitanya
## Simulation details
units lj
boundary p p p
atom_style full
## Defining and creating the system
variable side equal 8 ## Based on N=256
region simbox block 0.0 ${side} 0.0 ${side} 0.0 ${side}
create_box 1 simbox
lattice fcc 0.9 ## Reduced density
create_atoms 1 box
## Specifying mass (can be changed to change identity of molecule)
mass 1 1.0
variable TK equal 0.75 ## reduced temperature
velocity all create ${TK} 12345
## using lj interactions with tail addition
pair_style lj/cut 2.5 ## cut off is reduced units, i.e. 2.5*sigma/sigma
pair_modify tail yes
pair_coeff 1 1 1.0 1.0 2.5 ## cut off specification is optional
#minimize 1.0e-4 1.0e-6 100 1000 ## Yes required since starting from a random configuration
#reset_timestep 0
neighbor 0.3 bin ## proportional to box size
timestep 0.0001 ## The default value set for lj units is 0.005
fix TVSTAT all nvt temp ${TK} ${TK} 0.1
#fix 1 all nve
compute T all temp
fix avgT all ave/time 250 2000 500000 c_T file Tavg mode scalar
thermo 1000
thermo_style custom step cpu etotal ke pe evdwl elong temp press vol density atoms
run 500000
thermo 1000
thermo_style custom step cpu etotal ke pe evdwl temp press vol density atoms
unfix avgT
reset_timestep 0
variable P equal press
fix avgP all ave/time 500 2000 1000000 v_P file Pavg mode scalar
run 1000000
Chaitanya
Thanks for the details. I cannot reproduce the issue with your input command and input file. So I suspect the problem comes from your Linux installation.
I am not sure what is the configuration of wsl or how to get that information. Can you please tell me? The WSL version I am using is 0.2.1
I was more interested in the Ubuntu version installed on the wsl partition and disk usage.
Couple more questions:
output.lmp file contain anything? Is this the log file you are talking about. With your command there should be both output.lmp and log.lammps outputs at the end of a run.df -h?By the way, if your lammps executable is compiled with mpi and used for parallel simulation, I would recommend taking the habit of using the format as @hothello did, that is with command line flags -i and -l (or -in and -log) since they ensure correct execution with mpirun or mpiexec.
Hey Germain
Many thanks for the suggestions.
Yes, according to my command I should have both the output.lmp file as well as log.lammps file. But the output.lmp file comes out empty. And no log.lammps file is generated.
The wsl has a ubuntu version 20.04 LTS.
I did try the command suggested by @hothello but it is all the same.
Chaitanya
The commands format comments are side comments.
Let me ask you again: what is the output of the command df -h? This would clarify the situation.
Hi Germain!
Extremely sorry for forgetting to reply to that part. Thanks for being patient.
Here is the output I get with df -h :
Filesystem Size Used Avail Use% Mounted on
/dev/sdc 251G 6.5G 232G 3% /
none 3.9G 24M 3.9G 1% /mnt/wslg
none 3.9G 4.0K 3.9G 1% /mnt/wsl
tools 476G 209G 267G 44% /init
none 3.9G 0 3.9G 0% /dev
none 3.9G 4.0K 3.9G 1% /run
none 3.9G 0 3.9G 0% /run/lock
none 3.9G 0 3.9G 0% /run/shm
none 3.9G 0 3.9G 0% /run/user
tmpfs 3.9G 0 3.9G 0% /sys/fs/cgroup
none 3.9G 76K 3.9G 1% /mnt/wslg/versions.txt
none 3.9G 76K 3.9G 1% /mnt/wslg/doc
drivers 476G 209G 267G 44% /usr/lib/wsl/drivers
lib 476G 209G 267G 44% /usr/lib/wsl/lib
drvfs 476G 209G 267G 44% /mnt/c
Chaitanya
I think this rules out disk space problems since all your partition have some free space.
Last advice I can give are that I would not redirect output and ensure that your executable is correctly referenced in your command (both compiled and where the command looks for it). This would allow you to see if there is an error message from execution that you would miss.
Also building and testing the execution of a serial executable would make sense to see if this can come from MPI compilation/execution. I’ve at the moment no more insights to provide further than this. I’ve no idea where the issue can come from.
Thank you very much for the directions.
As of now I am running lammps using the windows executable. But I will check up on your suggestions and let you know if I find anything.
Many thanks,
Chaitanya
How did you build or obtain your lmp_mpi executable?
Hi @srtee
I downloaded the tar file from lammps website and after extraction, did make yes to the additional packages and then did make mpi.
Chaitanya