Other KMC Codes

[Benjamin Lowe]

Hiya, thought I might point out this KMC Code:

Zacros: http://zacros.org/ (http://www.homepages.ucl.ac.uk/~ucecmst/software.html)

I'm currently looking for KMC code that can deal with long range lateral interactions efficiently, please keep me posted if anyone is aware of such a code.

[Max Hoffmann]

Thanks, I made note of it. When you say long range interaction: could you be a bit more specific? How many sites influence the rate constant? How many different species per site?

[Benjamin Lowe]

Hi, sorry yes. By long range I mean a pair-wise additive interaction potential which requires interaction up to approximately the 10th nearest neighbour. Each site binary (1 or 0) spin i.e. two species.

[Max Hoffmann]

That sounds 3-4 times more long range than what surface chemistry is usually aiming for :-). I cannot think of a ready-to-use general kMC code for this on the top of my head. With that many processes you will have to evaluate rate constants on the fly. Doing this efficiently typically involves some trickery (i.e. symbolic math) depending on the specific form of your interaction.

[Benjamin Lowe]

I see, but I could imagine it would be possible to create a lookup table of precalculated additive terms for the net contribution from each nearest neighbour 'shell'. And then on the fly add on the appropriate amount based on the current configuration. This might be what you mean when you mention trickery!

Zacros caught my eye as it appears to be able to handle 8th nearest neighbour interactions according to the docs (http://zacros.org/index.php/tutorials/9-tutorial-3-cluster-expansion-for-oxygen-on-pt-111?showall=&start=1). I'm interested in doing simulating the long range charge-charge interactions involved in simulating proton binding to surfaces (http://dx.doi.org/10.1021/la9621325).

[Max Hoffmann]

I would claim the number of sites (occupancies) contributing to any one reaction constant is actually a better measure than sheer distance for both size of the problem both in terms of what the kMC code has to conquer and what the complexity of the physics underlying the model is.

A long range charge-charge (i.e. classical monopoles?) interaction certainly sounds like a pretty challenging scenario here. Can you rule out certain to distances (i.e. very small ones) at least for the temperature window you are looking at? And maybe can you make upper and lower bounds for the expected charge concentration? Maybe that could help to reduce the number of processes that actually have to be considered.

The best kmos can do right know in terms lateral interaction is the lat_int backend (compile with "kmos export -blat_int "). But I would be glad to learn more about it if zacros performs better for that many different lateral interactions.

[Benjamin Lowe]

I see what your saying, indeed thermal limits impose the maximum distance
that is viable, and full coverage is unlikely which lowers the upper bound
of the charge concentration (and therefore number of sites contributing).

Is there any documentation for the lat_int backend? What does it do?

Thanks!

On 17 February 2015 at 05:07, Max Hoffmann [e-mail] wrote:

I would claim the number of sites (occupancies) contributing to any one
reaction constant is actually a better measure than sheer distance for both
size of the problem both in terms of what the kMC code has to conquer and
what the complexity of the physics underlying the model is.

A long range charge-charge (i.e. classical monopoles?) interaction
certainly sounds like a pretty challenging scenario here. Can you rule out
certain to distances (i.e. very small ones) at least for the temperature
window you are looking at? And maybe can you make upper and lower bounds
for the expected charge concentration? Maybe that could help to reduce the
number of processes that actually have to be considered.

The best kmos can do right know in terms lateral interaction is the
lat_int backend (compile with "kmos export -blat_int "). But I would be
glad to learn more about it if zacros performs better for that many
different lateral interactions.

—
Reply to this email directly or view it on GitHub
#17 (comment).