problem with Berendsen barostat

Dear LAMMPS users,

The attachment in my previous email is to large, it is blocked by the server. Now I make it smaller.

My problem is that:

I want to simulate the SPC/E water molecules in NPT ensemble.

Both Nose-Hoover barostat and Berendsen barostat are tested.

The Nose-Hoover one can run smoothly, but the Berendsen barostat always gives the error “Non-numeric box dimensions - simulation unstable (…/pppm.cpp:1917)”.

I checked the source code pppm.cpp line 1917, it is said that my simulation box blows up.

However, I test a simple LJ system, both Nose-Hoover barostat and Berendsen barostat work well.

I am using the LAMMPS (11 May 2018).

Best wishes,

Wei (670 Bytes) (323 KB)

H2O.txt (550 Bytes) (292 Bytes) (45.9 KB) (109 KB)

One issue is that you are running a minimization which ignores fix shake. Therefore your water molecules are not rigid during minimization and hence the distance between your water and hydrogen atoms is not preserved. From a trajectory it seems to make only a small difference but that might be (part of) the problem. It is probably helpful to write a dump file and visualize what is going on. Also a smaller system will be faster to debug.

Dear Stefan,

Thank you for your reply.

I use fix shake in minimization,and then

Yes, but as I said, fix shake is not applied during minimization, only during runs.

The fact that one of the barostats gives issues for a big system but not for the others seems to hint at a potential problem with your input that should be resolved first. What exactly happens at the point of the error? Does your box blow up or shrink too much? Does your run produce dangerous builds? etc.

Dear Stefan,

I change the water model, now it is SPC/Fw model of water(The Journal of Chemical Physics 124, 024503 (2006); ). The bond and angle are not fixed anymore. So we don’t use the “fix shake” command this time.

The same problem!!

I checked the source code pppm.cpp line 1917, it is said that my simulation box blows up.

Now we set Pdamp in Berendsen barostat as 50000, it finally works. But I still don’t understand why?

Best wishes,


Please reply to the list, not to me.

How would I know why it works now? Maybe your damping time was too fast for stable dynamics, or maybe something else was going wrong. As long as you do not virualize your system or check the log files to see what the macroscopic properties are doing there is really no way of telling what is going wrong. Since you never answered any previous questions and instead decide to do seemingly random stuff there is no way in which I (or anyone else for that matter) can help you.