"units lj" causes normalised therm output (energy/num_atom)

LAMMPS LAMMPS Mailing List Mirror
1

Hello.
   I noticed an interesting quirk that I thought I'd ask about.
When you use "units lj", the energies in the therm output are reported
as per-atom energies (energy / num_atoms).

(I attached a simple test file of two molecules initially separated by
a distance of sigma*2^(1/6), and epsilon=1.0. See attached image.
The pair energy (at t=0) should be -3, but when "units lj" is used,
the energy is reported as -0.5. Switching back to "units real" (or
using "thermo_modify norm no") causes the energy to be reported as -3
again.)

If this is not a bug (I suspect it isn't), can somebody add a comment
the doc page for the "units" command?
(It would be nice to warn about this behaviour.)
http://lammps.sandia.gov/doc/units.html

  --- This is a very low priority nitpick of mine. ---
  --- Please look into Adrian's Respa post before worrying about this post: ---
http://sourceforge.net/mailarchive/message.php?msg_id=29656832

Cheers!
Andrew

test_norm.jpg

test_norm.in (859 Bytes)

test_norm.data (472 Bytes)

2

This is documented behavior. Should be explained somewhere on the thermo or thermo_modify pages. Axel.

3

Thanks. You're right, it's in the thermo_modify docs. (Not the first
place I would look though.) If anyone wants to add a mention of it on
the units page, that would help too. (Either way, this won't stop
people who don't notice their energies are off by a factor of N.)
Cheers

Andrew

4

If anyone wants to add a mention of it on
the units page, that would help too

Done.

Steve

5

Cool!