Dear developers
I have a question about hyper command and fix hyper/local and hyper/global
It is written, if I have well understood, that the boost apply to one or more bonds.
But in the commands here is only the cutoff length and the group-ID to which the fix is applied.
If I use group-ID all does it mean that all bonds are treated.
And if I use a group-ID concerning only one molecule for example, does it mean that only all bonds in the group are treated.
As example, I study hydrocarbon molecule (group-ID HC) in O2 (group-ID Oxy)
if I use fix 1 all hyper/local … does it mean all existing bonds (which are not explicit if I use reaxFF pair_style) are affected by the bias potential ?
if I use fix 1 HC hyper/local … does it mean that only C-C and C-H bonds will be considered, if cutbond includes both ?
Or is only the cutbond parameter which defines the bonds that will be boosted ?
I hope I am enough clear ?
Thanks a lot in advance
Pascal
You will have to wait for @sjplimp to answer this (or contact him directly if he doesn’t). I don’t think any of the others hanging out here on a daily basis have any experience with that feature. I certainly don’t have.
Thanks a lot Axel
I will wait for.
Do you mean we can have “private” discussion with a Lammps forum member ? And how to contact him ?
Best
You can try, but it is at the discretion of the individual.
I consider a private communication that is about LAMMPS and not about personal matters, like a personal consulting session and thus will expect some adequate form of compensation for it.
I am ready to donate my time only when the matter is made public and I can see a benefit for the entire LAMMPS community to learn from the discussion.
As you might imagine, I receive quit a significant number of emails from people to my personal email account for giving them personal advice and solving their problems. Often people don’t even properly introduce themselves and “hide” behind some alias. I have stopped responding to all but a few well written and polite requests from people that very junior (eg. high school students etc.) and then also just tell them to post in MatSci.
For some extreme cases, I am wondering, that if they had invested as much effort in investigating their problem than what they have in contacting me, they might have already found the solution for their (most often trivial and self-explanatory) issue. Oh well.
He has an email address and a home page.
Hi Axel
It is not my intent to have lammps discussion outside the forum.
I understand and agree with all your remarks.
Regarding the hyperdynamics topics, in any case, I will give here an account of the answer I will get.
Kind regards
Pascal
See https://pubs.aip.org/aip/jcp/article-abstract/153/5/054116/1065698/Parallel-algorithms-for-hyperdynamics-and-local for the algorithms used (I assume this paper should be linked in the documentation, but may not have been published yet when the code was updated).
Thanks a lot for the reference. I will read it.
Best.
Pascal
Hi all
As I promised, I come back to the answer of my question. I summarize hoxw to handle.
May be it can be useful for other people interested in.
Regarding hyperdynamics, if you use the fix 1 all hyper/local command, then all bonds between atoms located at a distance less than or equal to the cutbond will be affected by the defined potential bias. This means that all bonds present in the system, whether H-H, H-C, C-C, O-H, O-C, or O-O, will be processed as long as they meet the distance criterion imposed by cutbond.
On the other hand, if you use fix 1 HC hyper/local, only bonds where at least one of the two atoms belongs to the HC group will be accelerated. This means that H-H, H-C, C-C, O-H, and O-C bonds will be affected, provided that the atoms are located at a distance less than or equal to the cutbond. The O-O bond, on the other hand, will not be taken into account since oxygen atoms are not part of the HC group. It’s important to note that with this fix, two atoms will be considered bonded as long as their interatomic distance is less than or equal to cutbond. Therefore, there’s no need to explicitly define bonds upstream. With ReaxFF, bonds will be processed as they form.
Regarding hypertime retrieval, the hyper/local and hyper/global fixes each calculate a global vector of values. The cumulative hypertime is stored at index 23 for hyper/local, and at index 10 for hyper/global.
If you use fix 1 HC hyper/local, you can retrieve the hypertime using the command: thermo_style f_1[23] .
Thanks to @Jehiel for help too.
Best regards
Pascal