Stuck before fix npt start

LAMMPS LAMMPS Mailing List Mirror
1

Dear all

I come across a problem when i do a simulation using tip4p.
The input script is:

fix fxLAN all langevin 300.0 300.0 5000 48279
fix fxNVE all nve/limit 0.1 #(<–needed by fix langevin)
dump 4a all custom 100 obs.dump id mol type xu yu zu
run 10000

unfix fxLAN
unfix fxNVE

fix fxshake all shake 0.0001 20 0 b 9 a 17

fix fxnpt all npt temp 300.0 300.0 100.0 iso 1.0 1.0 1000.0 drag 1.0

run 10000

When it finish nve/limit stage, it stuck at
image.png
and the usage the cpu is full. There is no error raise after a long time wait.
This is the first time i do all-atom simulation and use long-range coul force, so i have no experience about it(question may be really stupid). Please give me some tips about how to fix this problem! The input files are attached.

Thanks in advance!

Roy Kid
University of Science and Technology of China

solvent.in (994 Bytes)

solvent.in.charges (305 Bytes)

solvent.in.init (287 Bytes)

solvent.in.settings (2.16 KB)

system.data (219 KB)

2

the problem is that you have a very high energy initial structure and by the time your system gets to execute fix shake, the geometry of most water atoms is so bad, that it get confused.

you can do the following:

  • increase the force constant by a factor of about 100 (or more if the system is very far from equilibrium) for the water bond (type 9) and the water angle (type 17).
  • start with a minimization (you cannot use fix shake with minimization, but with large force constants on the water bond and angle you prevent large deformations)
  • reduce the time constant for fix langevin, so that it actually can do some thermalization
  • after defining fix shake and fix npt, insert a “run 0” statement. that will reduce the chance that fix shake gets confused during initialization.

axel.

image.png

3

Every tip you mentioned is work, and this question exposes my understanding of molecular dynamics simulation is terrible. This email is valuable and gives me new guidelines to solve problems!

Thank you so much!

Roy Kid

[

李吉辰

邮箱：lijichen365@…223…

](https://maas.mail.163.com/dashi-web-extend/html/proSignature.html?ftlId=1&name=李吉辰&uid=lijichen365%40126.com&iconUrl=https%3A%2F%2Fmail-online.nosdn.127.net%2Fqiyelogo%2FdefaultAvatar.png&items=["邮箱：lijichen365%40126.com"])

签名由 网易邮箱大师 定制