FENE bond and Shifted-Lennard-Jone potential parameter Δ

[Jin Zeng]

Hi,

Hope everything is well!

I have trouble understanding the documentation of the quantity Δ in FENE bond and Shifted Lennard-Jone (sij) potential. In documentation, it said Δ = (di+dj)/2 + 1. 

I feel baffled about this is that what is the unit for this "1". Is this an type and should be Δ = (di+dj)/2 so that delta represents the hard-core distance between two bead?

Sincerely

[Jens Glaser]

Hi Jin,

You’re baffled for a good reason. The unit “1” is ad-hoc, and this represents a poor implementation choice, which goes way back to when SLJ was first implemented in HOOMD. 1 is consistently subtracted from the sum of radii to determine the neighbor list cut-off, and I believe the reason for this choice was that the implementor of the SLJ potential assumed that no-one would set `sigma` to a value different than unity. For the same reason, all diameters are 1 by default.

That said, you can still work with arbitrary units, simply by subtracting one from the diameters, and shifting the cut-off and/or sigma in the FENE and SLJ potentials accordingly.

I agree this is un-intuitive, and a good thing would be to create an issue on GitHub so this eventually gets fixed, or propose a patch.

Jens

--
You received this message because you are subscribed to the Google Groups "hoomd-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hoomd-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hoomd-users/92e8d5e7-c897-4249-8a5b-3d77ac025f09o%40googlegroups.com.

[Jin Zeng]

Hi Jens,

Please forgave my horrible spelling and grammar. Thank you so much for your explanation.

SIncerely

On Wednesday, June 17, 2020 at 9:27:56 PM UTC-4, Jens wrote:

Hi Jin,

You’re baffled for a good reason. The unit “1” is ad-hoc, and this represents a poor implementation choice, which goes way back to when SLJ was first implemented in HOOMD. 1 is consistently subtracted from the sum of radii to determine the neighbor list cut-off, and I believe the reason for this choice was that the implementor of the SLJ potential assumed that no-one would set `sigma` to a value different than unity. For the same reason, all diameters are 1 by default.

That said, you can still work with arbitrary units, simply by subtracting one from the diameters, and shifting the cut-off and/or sigma in the FENE and SLJ potentials accordingly.

I agree this is un-intuitive, and a good thing would be to create an issue on GitHub so this eventually gets fixed, or propose a patch.

Jens

On Jun 17, 2020, at 8:38 PM, Jin Zeng <jz3...@columbia.edu> wrote:

Hi,

Hope everything is well!

I have trouble understanding the documentation of the quantity Δ in FENE bond and Shifted Lennard-Jone (sij) potential. In documentation, it said Δ = (di+dj)/2 + 1. 

I feel baffled about this is that what is the unit for this "1". Is this an type and should be Δ = (di+dj)/2 so that delta represents the hard-core distance between two bead?

Sincerely

--
You received this message because you are subscribed to the Google Groups "hoomd-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to hoomd...@googlegroups.com.

[Michael Howard]

Hi Jin,

If you don't need to shift the potential using the particle diameter (i.e., a truly polydisperse system) and can get away with setting a Delta parameter per type-pair, you could use an alternative implementation of the slj potential that I wrote here:

https://github.com/mphowardlab/azplugins

The potential is azplugins.pair.slj.

Regards,

Mike

[Joshua Anderson]

All,

I hope to fix this issue in v3.0 by offering FENE and SLJ potentials that accept delta values directly.

The current implementation of both SLJ and FENE was written for this paper https://doi.org/10.1039/b911140a which uses polydisperse particle sizes. Does anyone still have a need for this behavior? Or are per-type diameters sufficient for all use-cases going forward?

------
Joshua A. Anderson, Ph.D.
Research Area Specialist, Chemical Engineering, University of Michigan
http://www-personal.umich.edu/~joaander/

To unsubscribe from this group and stop receiving emails from it, send an email to hoomd-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hoomd-users/2d802616-eddf-42f7-8c85-9189716f54f5o%40googlegroups.com.

[Eric Harper (CIV)]

I was curious if v3.0 of HOOMD-Blue would be including this fix for SLJ, or if SLJ still uses Δ=(di+dj)/2−1

[Joshua Anderson]

Eric,

The fix is not yet present in the v3.0 beta, but I do plan to add it. Would you be interested in contributing the fix?

------
Joshua A. Anderson, Ph.D.
Research Area Specialist, Chemical Engineering, University of Michigan

> To view this discussion on the web visit https://groups.google.com/d/msgid/hoomd-users/a20bfd9d-a50b-48f4-a57c-64d182cb83d9n%40googlegroups.com.

[Eric Harper (CIV)]

Josh,

I'll talk with my team and see if they still need the fix, and if so, if they'd like to contributing.