gcmc with atomic style and full style

Dear all,

I tried to use gcmc command with monatomic gases. I tried to use atomic style and full style including molecule command. These seem to produce different results.

The reason why I try also with full style and molecule command is for the case I run several molecules together. I need your advice.

I attached two different input files, two data files, one molecule file and one log file for atomic style and the other is the record of full style simulation running, which is gcmccheck.txt file.

30Jul2016 was used.

Best regards,

Baek

4.4nm_atomic.txt (138 KB)

4.4nm_full.txt (317 KB)

gcmccheck.txt (12.1 KB)

log.atomic (14.9 KB)

test_atomic.in (2.63 KB)

test_full.in (2.75 KB)

CH4.txt (207 Bytes)

Dear all,

I tried to use gcmc command with monatomic gases. I tried to use atomic style and full style including molecule command. These seem to produce different results.

The reason why I try also with full style and molecule command is for the case I run several molecules together. I need your advice.

I attached two different input files, two data files, one molecule file and one log file for atomic style and the other is the record of full style simulation running, which is gcmccheck.txt file.

a quick side by side comparison reveals that:

1) your two data files have a different number of atoms.

2) your two input files have a different value for the chemical potential.

why should those not produce different results?

axel.

Assuming you address Axel’s valid criticisms, there are still good reasons why the two simulations will not agree. Using the energy_full and molecule keywords changes the way in which the pseudo-random number sequence is used to sample the probability distributions for the Monte Carlo moves. Therefore the detailed sequence of moves will be different for each case. The only way to compare the different cases is to generate fully equilibrated states, sample them for long enough to get stable statistical averages and compare these average properties. These differences should be similar in magnitude to the estimated uncertainties in the averages.