[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Position folding issue
From: |
Jean-Noël Grad |
Subject: |
Re: Position folding issue |
Date: |
Wed, 19 Jun 2024 20:34:26 +0200 |
User-agent: |
Mozilla Thunderbird |
Hi again,
Could you please check if applying the attached patch with the command
`git apply patch.diff` in the top-level ESPResSo folder, and rebuilding
ESPResSo, fixes your trajectories? You should get the correct image
numbers in the hdf5 output files, and also in the values returned by
p.image_box. Both features were actually broken.
Best,
JN
On 6/18/24 17:54, Jean-Noël Grad wrote:
Hi,
Thank you for bringing our attention to this issue. This is indeed a
bug, and it affects many releases of ESPResSo. We have opened a ticket
(https://github.com/espressomd/espresso/issues/4940) and will look into
this. This is a challenging bug, because it depends on the Verlet lists
bookkeeping, and the solution will most likely be different for each
ESPResSo version.
Best,
JN
On 6/4/24 21:23, Yuan, Xin wrote:
Hello,
I have noticed an issue with position folding under periodic boundary
conditions. When a particle moves from the original simulation box to
the next mirror image, the image number does not increment until the
particle is at a distance (~0.01) away from the boundary. This leads
to an issue where the particle's unfolded position would be incorrect
if calculated from folded position and image number when the particle
is very close to the boundary. This appears to be especially relevant
when recording trajectory in H5MD format, as only the folded positions
and image numbers are recorded in an H5MD file, and there is no way to
extract correct unfolded position from an H5MD file for such
particles. I have attached a Jupyter notebook file to illustrate this
issue, which was tested on a WSL Ubuntu environment set up according
to the instructions in the online documentation as well as other conda
environments. Is this a software bug or something I was not doing
correctly? Thanks for any help!
Regards,
Xin Yuan
patch.diff
Description: Text Data