[lammps-users] Possible typo in hertzian potential

LAMMPS LAMMPS Mailing List Mirror
1

Hello users,

I have a question concerning the pair style gran/hertz/history. By comparing the normal force defined in LAMMPS with the cited paper by Zhang,Makse, I get difference of sqrt(2) coming essentially from the polydisperse factor. I believe it should be given by sqrt(2R_iR_j/(R_i+R_j)) instead of sqrt(R_iR_j/(R_i+R_j)) as written in the LAMMPS doc. This of course is just an overall factor which can be easily accounted for by adjusting the values of all the parameters. But then, looking at page 477, it says that for monodisperse systems, the parameters should be multiplied by 2. I believe they should rather be multiplied by sqrt(2). Is that right?

François Fillion-Gourdeau, PhD

Scientifique de la défense | Defence Scientist

Section Protection et Effets d'armes | Weapons Effects and Protection Section

Recherche et développement pour la Défense Canada - Valcartier | Defence Research and Development Canada - Valcartier

2459, Boul. Pie-XI Nord | 2459, Pie-XI Blvd. North

Québec, Quebec Canada G3J 1X5

[email protected]...

Téléphone | Telephone (418) 844-4000 x4395

Télécopieur | Facsimile (418) 844-4876

Gouvernement du Canada | Government of Canada

www.valcartier.drdc-rddc.gc.ca <http://www.valcartier.drdc-rddc.gc.ca/>

2

The comments in the pair granular doc page about a factor of
2 refer to an older version of the code. The current version
should be consistent with the eqs listed on that page.
They may not match the Zhang paper exactly (my recollection
is there was some ambiguity about that paper), but you can,
as you say, account for a pre-factor by setting the coeffs accordingly.

If you search the mail list archives for "Zhang" you will find
some discussion of this issue.

If you think the code does not do what the eqs on the pair granular
doc page say, then that is an issue we should address.

Steve