please *always* reply to the mailing list, not only individual people. thanks.
It is strange. I'm using the last git version.
did you also do the required package updates?
axel.
It is strange. I'm using the last git version.
did you also do the required package updates?
axel.
No I only downloaded the source with
git clone git://git.lammps.org/lammps-ro.git
and compiled including some packages already present.
Domenico
It is strange. I'm using the last git version.
did you also do the required package updates?
axel.
No I only downloaded the source with
git clone git://git.lammps.org/lammps-ro.git
and compiled including some packages already present.
the question is, when exactly you did the "git clone". there are
continuous updates to the sources and then you have to update your
local git repository and then - and only then - you may need to update
the installed packages as described on the LAMMPS download webpage.
you may alternately try:
git clone git://git.lammps.org/lammps-icms.git
this has a few additional bugfixes and some additional features and
usually includes all changes from the upstream version. i used it for
all precompiled binaries, so i can usually detect and correct
compilation issues faster.
axel.
ahhhh, found the problem.
your mail problem has added a .txt extension to the template file and
thus when i ran the test input, it used your old template file
instead.
sorry for the confusion,
axel.
the question is, when exactly you did the "git clone"
I did "git clone" this morning.
I found that if I use fix nve instead of fix nvt, the inserted molecules have the correct initial velocity
but the simulation crashes immediately because I have
WARNING : Shake determinant < 0.0
ERROR on proc 0: Shake atoms 119 123 missing on proc 0 at step 1956
Domenico
the question is, when exactly you did the "git clone"
I did "git clone" this morning.
as i pointed out before, this will currently not have the molecule
bugfix, since steve is not able to update the repository for a few
days.
I found that if I use fix nve instead of fix nvt, the inserted molecules
have the correct initial velocity
but the simulation crashes immediately because I have
WARNING : Shake determinant < 0.0
ERROR on proc 0: Shake atoms 119 123 missing on proc 0 at step 1956
you are using a *very* new feature and in a way that hasn't been used
before. as you have already seen, this will result in some glitches,
since there are far to many permutations to how LAMMPS can be used
(the price we have to pay for a flexible software). so it is great
that you exercise these new features and let use know, but it is not
always easy to immediately know what is not working as expected and
why.
thanks for your understanding,
axel.
so far, i have found one small bug in LAMMPS:
diff --git a/src/molecule.cpp b/src/molecule.cpp
index 0943486..dfac8cc 100644
--- a/src/molecule.cpp
+++ b/src/molecule.cpp
@@ -126,7 +126,7 @@ void Molecule::compute_mass()
masstotal = 0.0;
for (int i = 0; i < natoms; i++) {
if (rmassflag) masstotal += rmass[i];
- else masstotal += atom->type[type[i]];
+ else masstotal += atom->mass[type[i]];
}
}
but there are two issues with your template file:
- first of all your template file has a masses section, which is
inconsistent with using per type masses. LAMMPS prints a warning about
that, so that section has to be deleted.
- but secondly, you seem to be missing the various Special Bonds
sections. again, there is a warning about this. hard to say what
LAMMPS does with such inconsistent data...
axel.
so far, i have found one small bug in LAMMPS:
diff --git a/src/molecule.cpp b/src/molecule.cpp
index 0943486..dfac8cc 100644
--- a/src/molecule.cpp
+++ b/src/molecule.cpp
@@ -126,7 +126,7 @@ void Molecule::compute_mass()
masstotal = 0.0;
for (int i = 0; i < natoms; i++) {
if (rmassflag) masstotal += rmass[i];
- else masstotal += atom->type[type[i]];
+ else masstotal += atom->mass[type[i]];
}
}but there are two issues with your template file:
- first of all your template file has a masses section, which is
inconsistent with using per type masses. LAMMPS prints a warning about
that, so that section has to be deleted.
- but secondly, you seem to be missing the various Special Bonds
sections. again, there is a warning about this. hard to say what
LAMMPS does with such inconsistent data...axel.
I removed the Masses section and added the Special Bonds
sections and now seems to work very well with fix nvt.
Now i want to try to add Impropers.
Thanks
Domenico
pentacene_template (16.2 KB)
pentacene_final_due.data (23.8 KB)
pentacene_final.in (1.1 KB)
got this change - thanks
Steve