Dear all

I am currently trying to use fix press/berendsen command to reach certain stress value of the system. I used the command:

```
fix 1 all press/berendsen x 0 1000 1000 y 0 1000 1000 z 0 1000 1000
fix 2 all nve/bpm/sphere
```

As suggested, the pdamp is selected as 1000. for my case, p_end is 1000 pa. But I always identified a wired things:

(1) firstly, the boundary is moved with no stress, no contacts, it is normal, I think.

(2) but when particles contact with each and then the stress just increases dramatically to very large value, (from 0 to 10, then 100 then 1000 and then say about 8000000 kPa), then the system and simulation fail.

(3) I restarted the simulation from a early point (where the stress is just 1pa, something like this). However, for the restarted tests, the first line is always very large, say about 3000 kPa.

I am just wondering why i cannot use fix press/berendsen to reach the target stress? why I could see the dramatic increase of stress, much larger than my target.

if anyone have any idea for this, thanks for your help

hope you have a nice day

You are presenting your case as if you are discussing with your adviser or somebody similar that is extremely familiar with your research and the details of your simulation. However, the opposite is true and thus it is next to impossible to provide meaningful advice.

What fix press berendsen does is very straightforward and rather minimal. (Incorrectly) assuming a linear relationship between your objective (pressure) and your control parameter (volume), it computes by what fraction it needs to rescale the control parameter in every step to reach the target by the time t, i.e. the relaxation time constant. For that it makes a second assumption, and that is assuming the compressibility of your material. Both assumptions work well, if you have a large bulk system with mostly water molecules (because the compressibility of water is used unless explicitly set via the command line) and the relaxation time constant is large enough so that any non-linear behavior does not create too much instability.

There is not enough context here to make sense of points 1 to 3.

Hi akohlmey

i see, thanks for your reponse and help,