non-continuous domain interval for non-bonded interaction in force-matching

28 views
Skip to first unread message

Cecília Álvares

unread,
May 31, 2023, 5:54:18 AM5/31/23
to votca
Hello everyone,

I was wondering if there is a way to  define a "non-continuous" domain interval when setting up the settings.xml file to do force-matching of non-bonded potentials (for example) within VOTCA. I mean, in my case, since I am deriving potentials for a xtalline solid, I have peaks in the RDF that go up and down back to 0 (e.g. below), which would give rise to non-sampled intervals for the potential if I was to use a continuous interval for rij going from the first peak to the last.

rdf.png

Cecília Álvares

unread,
May 31, 2023, 6:07:35 AM5/31/23
to votca
(I mean, in this case, I could get "chops" of how the potential would look like in different intervals (corresponding to g(r) != 0) and I later can connected the different portions by doing some sort of interpolation myself.)

Cecília Álvares

unread,
May 31, 2023, 6:27:12 AM5/31/23
to votca
Hmm.. maybe this is actually a very stupid question, because, given that the potential is built by using splines that hold within a small subinterval defined in <step>, I suppose I could get the same result by splitting the ranges of non-zero RDF in different interactions. (it should give the same results, right?)

Christoph Junghans

unread,
Jun 3, 2023, 10:20:38 AM6/3/23
to vo...@googlegroups.com
On Wed, May 31, 2023 at 4:27 AM Cecília Álvares
<cecilia...@gmail.com> wrote:
>
> Hmm.. maybe this is actually a very stupid question, because, given that the potential is built by using splines that hold within a small subinterval defined in <step>, I suppose I could get the same result by splitting the ranges of non-zero RDF in different interactions. (it should give the same results, right?)

The reason why we haven't implemented support for non-continuous
domains is mainly that it is complicated to get all the cases right
and so we thought one could just do a piecewise ibi and then combine
it.

Christoph

>
> Em quarta-feira, 31 de maio de 2023 às 12:07:35 UTC+2, Cecília Álvares escreveu:
>>
>> (I mean, in this case, I could get "chops" of how the potential would look like in different intervals (corresponding to g(r) != 0) and I later can connected the different portions by doing some sort of interpolation myself.)
>>
>> Em quarta-feira, 31 de maio de 2023 às 11:54:18 UTC+2, Cecília Álvares escreveu:
>>>
>>> Hello everyone,
>>>
>>> I was wondering if there is a way to define a "non-continuous" domain interval when setting up the settings.xml file to do force-matching of non-bonded potentials (for example) within VOTCA. I mean, in my case, since I am deriving potentials for a xtalline solid, I have peaks in the RDF that go up and down back to 0 (e.g. below), which would give rise to non-sampled intervals for the potential if I was to use a continuous interval for rij going from the first peak to the last.
>>>
>>>
> --
> Join us on Slack: https://join.slack.com/t/votca/signup
> ---
> You received this message because you are subscribed to the Google Groups "votca" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to votca+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/votca/c201cbeb-641d-4a7c-b043-d55101cfcb78n%40googlegroups.com.



--
Christoph Junghans
Web: http://www.compphys.de

Cecília Álvares

unread,
Jun 5, 2023, 5:04:05 AM6/5/23
to votca
Hey Christoph

You are right, from looking in the literature it does seem complicated.. and my idea of doing some interpolation is not very rightful because by doing that I would be introducing "forces" at given pairwise distances which do not come from predictions from the force matching method. I mean, if the whole purpose is to find a potential that reproduces the forces, I think I would jeopardize the whole thing to one extent or another if my superatoms end up being, even if for a tiny moment, in the artificial repulsive regions I was planning to introduce.

I saw some people who do FM at high temperature in order to get the separate peaks to merge. Seems to have worked for me for the peaks sitting at > 10 angs. And since my first and second peaks in the RDF I showed comes from 1-2 and 1-3 neighbors, I could assign them a separate bonded potential instead. I still have some other problems in the potential I am obtaining but I will figure out :))

Thanks!
Best,
Cecilia
Reply all
Reply to author
Forward
0 new messages