I want to know ucf format.

[PumSuk Park]

Hi all,

Thanks to Richard for making such nice open-source program.
This program is very nice!

I have some questions about the format of ucf.

 

At the second section, the comments are written as

" Atoms num, id cx cy cz mat lc hc".

What is cx, cy, cz, mat, lc, hc?

At the third section, the comment is written as 

"Interactions n exctype, id i j dx dy dz Jij".

What is dx, dy,dz? 

Thank you!
P.S. PARK

[richard....@gmail.com]

Hi,

cx, cy and cz are the positions of atoms in the unit cell, given as a fraction of the total unit cell size in x, y and z.

mat is the material ID of the atom, lc is lattice category and hc is height category, where magnetization statistics can be collected for groups of heights.

dx, dy and dz relate to the interactions with neighboring unit cells, eg for

i = 0, j=0, dx=1, dy=0,dz = 0

gives an interaction between atom 0 in the local unit cell and atom 0 in the next unit cell along the x-direction. The interaction list scheme is done this way to allow it to be very general while also fast in terms of setting up the interactions in the system.

All the best,

Richard

[PumSuk Park]

Hi, Richard

Thank you for your quick response. 

I want to calculate "MnB". This space group is PNMA.

By using ucf file, I thought that my plan will be going but, I can't apply this case. 

Can you show me the way of constructing ucf file? 

2015년 3월 6일 금요일 오후 9시 17분 32초 UTC+9, richard....@gmail.com 님의 말:

[richard....@gmail.com]

Hi

My experience with space groups is patchy at best, but in principle this is sort of possible with the ucf format. The interactions between neighboring unit cells are explicit, and vampire just replicates that unit cell in space in x, y and z. So, if the unit cell is not cubic, then the positions of the atoms given by vampire is wrong, but the interactions are correct. The downside is the 'shape' of the crystal is limited to the same shape as the unit cell, and obviously cutting spheres etc from the crystal won't work. I do want to add support for non-orthogonal unit cells, but its something I didn't get round to yet, as there are a couple of tricky bits to sort out in terms of generating different sample shapes and possibly parallel periodic boundary conditions (which relies on a cubic discretization of the structure).

So, we did some work with Mn doped ZnO and in that case we constructed the actual crystal in a separate code (with the correct crystal symmetry and positions), and calculated the interactions between unit cells in the normal way, by using a cut-off radius. We then simply loaded that structure into vampire (which replicated the crystal as a cubic structure) and performed the simulations that way. It only works since we wanted periodic boundary conditions, no particles and no demagnetizing field), but if that is a limitation that would be OK then it should be possible to simulate that structure. 

The alternative is to look at modifying the code to deal with the crystal symmetry. At the moment I am a bit time constrained, but if you are interested in doing that I can certainly give some help with understanding any unclear parts of the code?

All the best,

Richard