temper command

Dear lammps developers and users,

I am using the temper command to run a replica exchange test. A test case input file is as attached. The case is performed on two platforms (both are MPI-enabled linux-based HPC). When one (PBS) handles the run successfully, the case in the other system(LSF) stopped after the temper and seems to halt in queue. I need to manually kill the job for a forced exit. I am puzzled for the different outcomes generated by the identical input…

The last part of console output is as followed:

Pair time () = 1478.06 (59.2356) Bond time () = 0.565722 (0.0226723)
Kspce time () = 467.286 (18.7273) Neigh time () = 34.9575 (1.40098)
Comm time () = 227.232 (9.10673) Outpt time () = 4.67559 (0.187382)
Other time (%) = 282.442 (11.3193)

the usual output for the FFT/Nlocal/and etc., are not seen.

So, I am wondering if I should add any command after the temper,in order to let lammps notice the end of simulation? Or it is actually calculating something or even falling into a infinite loop?

Any input/comment is greatly appreciated. Thank you for the time and efforts.

LC Liu

in.tip4p (1.73 KB)

300.restart1 (285 KB)

Dear lammps developers and users,

I am using the temper command to run a replica exchange test. A test case
input file is as attached. The case is performed on two platforms (both are
MPI-enabled linux-based HPC). When one (PBS) handles the run successfully,
the case in the other system(LSF) stopped after the temper and seems to halt
in queue. I need to manually kill the job for a forced exit. I am puzzled
for the different outcomes generated by the identical input....

The last part of console output is as followed:

Pair time (\) = 1478\.06 \(59\.2356\) Bond time \() = 0.565722 (0.0226723)
Kspce time (\) = 467\.286 \(18\.7273\) Neigh time \() = 34.9575 (1.40098)
Comm time (\) = 227\.232 \(9\.10673\) Outpt time \() = 4.67559 (0.187382)
Other time (%) = 282.442 (11.3193)

the usual output for the FFT/Nlocal/and etc., are not seen.

So, I am wondering if I should add any command after the temper,in order to
let lammps notice the end of simulation? Or it is actually calculating
something or even falling into a infinite loop?

are you using the latest version of LAMMPS on both machines?

axel.

Hi, Axel,

Thank you so much for the comments.

I was using lammps-4Jul12 (the ok one) and lammps-30Aug12 (the NG one) on the machines. My mistake for not synchronizing the versions.

Hence, I compiled the latest version lammps-17Oct12 and change the tip4p pair style to:

pair_style lj/cut/tip4p/long 1 2 1 1 0.1546 14.0

according to the recent modification : http://lammps.sandia.gov/doc/pair_lj.html

However, lammps does not recognize this pair style. (ERROR: Invalid pair style (force.cpp:168))

According to the 10/7 bug fix log: (http://lammps.sandia.gov/bug.html)

“BACKWARD COMPATIBILITY issues: The pair style lj/cut/coul/long/tip4p was changed to lj/cut/tip4p/long to be shorter and more consistent with other pair style names.”

Is there a change of the tip4p pair style format?

Regards,

LC Liu

Hi, Axel,

Thank you so much for the comments.

I was using lammps-4Jul12 (the ok one) and lammps-30Aug12 (the NG one) on
the machines. My mistake for not synchronizing the versions.

Hence, I compiled the latest version lammps-17Oct12 and change the tip4p
pair style to:

pair_style lj/cut/tip4p/long 1 2 1 1 0.1546 14.0

according to the recent modification :
http://lammps.sandia.gov/doc/pair_lj.html

However, lammps does not recognize this pair style. (ERROR: Invalid pair
style (force.cpp:168))

According to the 10/7 bug fix log: (http://lammps.sandia.gov/bug.html)

"BACKWARD COMPATIBILITY issues: The pair style lj/cut/coul/long/tip4p was
changed to lj/cut/tip4p/long to be shorter and more consistent with other
pair style names."

Is there a change of the tip4p pair style format?

please check with 'lmp_linux -h'
whether any tip4p style is included
and what they are named.

cheers,
    axel.

Hi, Axel,

Thanks for the information. tip4p is not included in my executable…is it something to do with the Makefile?

Regards,

LC Liu

Hi, Axel,

Thanks for the information. tip4p is not included in my executable....is it
something to do with the Makefile?

no, but you probably didn't install the molecule and/or the kspace package.

axel.

p.s.: and you may want to check out the opt package, too. :wink:

Hi, Axel,

Thank you so much for walking me through, yet I am still in trouble. I have lj/cut/tip4p/long installed, but my lammps still does not recognize this pair style. My installed packages are:

Installed YES: package KSPACE
Installed YES: package MANYBODY
Installed YES: package MOLECULE
Installed YES: package OPT
Installed YES: package REPLICA

LC Liu

Hi, Axel,

My 17Oct12 lammps still does not recognize my keywords in input file as:

pair_style lj/cut/tip4p/long 1 2 1 1 0.1546 14.0

so I took a second look to the installed pair style of my compiled lammps-17Oct12 and found these styles containing tip4p keywords:

lj/cut/tip4p/long
lj/cut/tip4p/long/opt
lj/long/tip4p/long

just a thought…is it possible that the naming convention gets mixed up somewhere inside the code?

Thanks a million for the help.

LC Liu

These commands:

pair_style lj/cut/tip4p/long 1 2 1 1 0.1500 6.0 6.0
pair_coeff 1 1 0.1550 3.1536
pair_coeff * 2 0.0000 0.0000
kspace_style pppm/tip4p 1.0e-4

work fine in the current version of LAMMPS.

Steve