Unreported bugs in 23 June 2022 stable release

LAMMPS LAMMPS Development
1

Dear LAMMPS users and developers,

We are currently preparing a third update to the 23 June 2022 stable release.
If you know of any bugs a recent LAMMPS version (stable or develop branch) that are not yet reported or not fixed, please let us now immediately.

Since the develop branch has by now diverged quite a bit from the stable release, the effort of backporting bugfixes is growing exponentially and thus this will very likely be the last chance for an update of the stable version and it will likely take us until Summer until we have a new stable version ready.

Thanks in advance.

2 Likes
2

Hi Axel,

I have noticed that the compute inertia/chunk produces bogus results for finite-size particles. It should be easy to fix for ellipsoids, but I haven’t had the time to patch the file (mainly because I’m rubbish with C++).

3

That is not so much a bug but should rather be called a missing feature.

As far as I can see, compute inertia/chunk treats all particles as point particles, which is in line with other similar computes. To properly support extended particles, especially ellipsoids, but also sphere, lines, and triangles quite a bit of extra computation is needed. That kind of computation could probably be lifted from the fix rigid code (rigid bodies are not very different from chunks).

The best that can be done for the stable release is to add a warning.

Could you produce some simple input(s) with known results? I would like to add this to the unit test library and then try to fix the issue in the development branch for real.

4

Here an example with a single ellipsoid:

# create a benzene-like disk (uniaxial ellipsoid)
units        real
atom_style   ellipsoid
region	     box block 0 10 0 10 0 10
create_box   1 box
create_atoms 1 single 0 0 0
set          type 1 mass 78.11
set          type 1 shape 6.4 6.4 3.2
pair_style   gayberne 1 1 2 14.
pair_coeff   1 1 2.0 3.8 0.22 0.22 1 0.22 0.22 1 14.

# Compute the moment of inertia.
compute cc1     all chunk/atom type
compute inertia all inertia/chunk cc1
fix     AV1     all ave/time 1 1 1 c_inertia[*] file inertia_cg_02.out mode vector
run     0

The computed inertia tensor is nil, while it should be diagonal with components:

\begin{equation} I=\left[\begin{array}{ccc} \frac{1}{5}M(b^2+c^2) & 0 & 0\\ 0 & \frac{1}{5}M(a^2+c^2) & 0\\ 0 & 0 & \frac{1}{5}M(a^2+b^2) \end{array}\right]=\left[\begin{array}{ccc} 199.96 & 0 & 0\\ 0 &199.96 & 0\\ 0 & 0 & 319.94 \end{array}\right], \end{equation}

where a, b, c are the semi-axis lengths. Mathworld has a nice table with the moments of inertia of some common solids.

5

There was a delay due to some unexpected problems with the stable release update, but we are back on track and it will likely be published next week.

So this is the last, last, last chance to report bugs we don’t know about.

Thanks, Axel.