GCMC can not fully fill a box with particles for high MC movement frequency

Dear Lammps developers,

Here is a empty box, and I want to fill this box with a mono-disperse particle by gcmc operation. The “fix gcmc” command is as follows:

fix ID group-ID gcmc N X M type seed T mu displace keyword values 

where M = average number of MC moves to attempt every N steps.

I remain other parameters constant, and only change the value of M from 10^0 to 10^9.

The particles can normally fully fill the box (about 1000 particles) when M is less than 10^4; but when M is larger than 10^5, the box can not be fully filled: many large empty space can be found by eyes, but the particle number as well as particle positions in the box basically remain the same (about 300 particles) even after a very long running time.

I have to used the final configuration (300 particles) to restart a gcmc simulation with a lower value of M (M=100), and the box is quickly fully filled.

Does M have a restriction range in gcmc process?

Thank you and best regards,
Han

I am not clear if you saying that using a small value of M, the box ends
up more filled than using a large M. Or that they both eventually stop
inserting at the same level of overall fill. If the former,
then perhaps Aidan (CCd) has an idea.

Steve

Hi Steve and Aidan,

Thank you for your reply.

It is the former case: when M >10^5, the number of particles saturates with respect to the number of GCMC steps (even in log scale). This plateau occurs for a number of particles that is significantly lower than for lower values of M. It appears as though all the insertions of new particles get rejected, although a visual inspection shows that there is still plenty of empty space available (with pores larger than the particle diameter).

I look forward to hearing from you.

Thank you again and best wishes.

Sincerely,
Han

Dear Steve,

Thank you for your consideration.

Do you find any solution to solve this problem?

Hope to hearing from you.

Thank you again and best regards,
Han

Hi Steve and Aidan,

Sorry to bother you.

The problem remains unsolved after careful checking.

I appreciate your help and look forward to hearing from you.

Thank you and best regards,
Han

The GCMC behavior should be insensitive to the choice of M, especially for large M. The fact that it fails for M > 10^5 suggests some kind of problem in the code. Can you demonstrate this behavior with the LJ example in the LAMMPS distro? If so, please send an example to me.

I did my own test using in.gcmc.lj and so no dependence on M up to 100000. Are you using the region keyword?

Aidan

Hi Aidan,

Thank you for checking the problem.

Enclosed is the example using LJ potential and M=10^9, where mono90.dat is the final configuration.
The problem remains: It appears as though all the insertions of new particles get rejected when MC steps >= 70, although a visual inspection shows that there is still plenty of empty space available (with pores larger than the particle diameter).
If there is any other need, be free to contact me.

I appreciate your help and look forward to hearing from you.

Best regards,
Han

error report.pdf (453 KB)

input.gcmc (1.15 KB)

mono90.dat (39.5 KB)

dump.gc.lammpstrj (579 KB)

Hi Aidan,

Is there any solution for the problem after our last email last year? If there is anything I can help, please let me know. Thank you again and look forward to hearing from you.

Best regards,
Han