Ave/correlate/long output

LAMMPS
,
1

Hello Everyone,

Firstly, I use the latest stable version of LAMMPS (29 Aug 2024) on Ubuntu 24.04.2.

I want to calculate the stress autocorrelation function and use fix ave/correlate/long:

fix g all ave/correlate/long 1 5000 v_pxx v_pyy v_pzz v_pxy v_pxz v_pyz v_nxy v_nxz v_nyz type auto nlen 16 ncount 2 ncorr 30 file G${a}.txt

This is the output:

# Timestep: 0
0 455.129 452.42 455.602 0.00470603 7.74652e-05 0.00511631 0.00404431 0.000122775 0.0055764
# Timestep: 5000
0 459.581 459.5 459.473 0.00432385 0.00418313 0.00394885 0.0175516 0.0172921 0.0167453
0.02 459.577 459.496 459.469 0.00264374 0.00247464 0.00235646 0.0109809 0.0106915 0.0102103
...

the first column is the time lag, and the #timestep is when the calculation if performed. So we have different blocks for each timestep.

My question is the following:
Is it is the final block we are interested in, or we need to average out between different blocks? I think it is the final block?

Thank you,
Alex

PS. I have read the documentation, but did not find a response. Maybe the way the multiple-tau algorithm works would help figuring this out.

2

You should really know that. If you don’t, it means you don’t know what quantity you’re interested in! You should think about it and take the time to read the theory. We can’t know for you, because we don’t know what you’re doing. All quantities printed here could be either relevant or irrelevant, depending on which aspect of the problem you’re studying.

Simon

3

Maybe my question was not understood.

I was not asking what each column is, nor what each quantity being printed is. My question was simple, and so was the answer.

I had forgotten the keyword “overwrite”, and thus the autocorrelations were being printed at fixed timesteps. There is no “theory”.

Lastly, asking if the mulptiple-tau is responsible for the structure of the file is a reasonable question, of somebody trying to understand the “theory”. Responding that no, this is not the case, is quite simple and helpful.

Thank you,
Alex