Re: Creating bonding and giving valency (no. of possible bonds) to active particle systems. #Hoomd-Beginner

[Antonia Statt]

Hi Ojus

I'm not entirely sure what you are trying to do. Hoomd-blue allows you to create bonds between particles on the python level, but it will not change the bonds between the pair you defined dynamically during the simulation. In other words, the bond-architecture is fixed during hoomd.run(2e6) to whatever you set it beforehand. So, you would need to implement the bond creation yourself as a plugin (https://hoomd-blue.readthedocs.io/en/stable/developer.html#plugins-and-components).

I'm not a 100% sure on this but if you use hoomd.init.create_lattice(), you define a system without any bonds. Try setting up your system manually and define the positions and bonds yourself. Then you should at least get a simulation containing bonds between fixed pairs of particles.

Best,

Antonia

On Tuesday, June 11, 2019 at 11:35:21 AM UTC-4, Ojus Bagal wrote:

Sorry, i added a different script.

questions are the same.

thanks.

[Ojus Bagal]

Thanks Antonia.

Let me try to explain my problem once more.

The code I am working on creates 2 types of particles (say type 1 and 2). There exists a Lennard-jones pair potential between different particle types but not among same particle type (there exists pure repulsion forces among same particle types.). The bonds that are created due to this pair potential between them are dynamic in nature.

What i want is to provide these particles with a way to limit the number of dynamic bonds to a set value, so that one particle of a type(type1) can only interact with a limited number of particles of the other particle type(type2).

I believe one way to do this is to add a counter to each particle or using lists (i don't know for sure as i have no background of programming).

can you please suggest a way to do this?

thank you.

[Arun Srikanth]

Hi Antonia

are you using hard sphere potential for same particle types?

What criterion are you going to use to limit the number of bonds between type 1 and type 2, distance criterion?

Arun 

--
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/7a24f827-6152-49a2-9db5-3091e1b32b2a%40googlegroups.com.

[Arun Srikanth]

Sorry I meant Ojus Bagal

[Ojus Bagal]

Hello arun,

As I have stated i am using LJ potential throughout the system, but the attractive part has been switched off for the same particles types.

the criteria to select the valency for each particle will be decided using a different program.

i, however would like to know how to assign these valencies to each particle.

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

[Michael Howard]

fwiw, time-driven MD is probably a poor computational method for the model you describe. If the number of bonds per particle is constrained, you will have discontinuous events as the particles begin/stop feeling forces from close by particles. You will have accordingly poor energy & momentum conservation, and it is questionable whether your results will be sufficiently accurate. MC is probably a more appropriate choice of method, since it relies only on the energy function--but I doubt that HOOMD's HPMC+jit can impose the constraints that you describe.

If you still want to use this method anyway, I believe Eric Jankowski's group has a plugin implementing dynamic bonding in HOOMD. Perhaps it can be modified for your purposes:

https://bitbucket.org/cmelab/hoomd_blue/src/dynamic_bonding/

Regards,

Mike

[Ojus Bagal]

Thank you so much for your insight, Micheal. I'll try to come up with a better model.

Will it be better if i consider these particles as patchy particles and and create constrains on the number of patches and types of interaction among them to overcome the problem you said.?

[Eric Jankowski]

Hi Ojus,

It's tough to say what's "better" here!

For dynamic bonding, our plugin is (of course, I think) a good starting place, but if you can represent the relevant physics with sticky patches that may be simpler to implement. Good advice here depends on what questions you want to answer, how long you have to answer them, the resources you have access to, and your professional aspirations after the project, among other things. I'd recommend working backward from your goals and deadlines with your advisor to calibrate your expectations for what needs to happen (e.g., something like: Writing a paper after analyzing simulations after running them after debugging them after building a model after building a bunch of bad models after learning how to program in a few languages after reading literature in a bunch of different areas). Maybe you find "Alright! Time to dive in!" Maybe it's "Whooo boy... we should collaborate." Or maybe it's something else.

Good luck!

Eric

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/ae667d0e-b902-43b6-aed2-a1791535fb00%40googlegroups.com.

--

Eric Jankowski PhD | Assistant Professor
Micron School of Materials Science & Engineering
Boise State University | 1910 University Dr. | Boise, ID 83725
ericja...@boisestate.edu
(208) 426-5681

[Ojus Bagal]

Thanks a lot Eric, This is going to be very helpful.

To view this discussion on the web visit https://groups.google.com/d/msgid/hoomd-users/ae667d0e-b902-43b6-aed2-a1791535fb00%40googlegroups.com.