Closed jarvist closed 6 years ago
This package is more experimental than the others, hence not registered. I therefore didn’t prioritise it. But it should work in principle. I’ll do some tests.
Are you actually interested in using this (potentially) or are you just trying to figure out what is available? Just to be clear: I would under no circumstances use this package for an actual materials modelling publication. I use it to develop tweaks and modifications to TB models that I can't do (easily) in production software. Maybe if somebody more knowledgeable can invest some time it could actually become more reliable from the physics/chemistry perspective.
Potentially both! I'm getting back into Tight Binding method development; particularly Polarisable Ion TB with the Hairy Probe open-boundary method - applied to computational corrosion. The production codes (https://questaal.org/docs/code/tbe/ ; https://en.wikipedia.org/wiki/PLATO_(computational_chemistry) ) are a little inflexible for model development, so having a Julia implementation to develop tweaks and play with fitting methods would be pretty useful.
needless to say, any contributions are extremely welcome! Even just letting me know where there are shortcomings in your view would already be very helpful!
I'll try to fix this asap.
sorry for the delay - I got bogged down preparing a new talk. I'll get on this now.
Everything should be fine now. (It was actually just a matter of merging the right branch, I just never got around to making sure everything is still working with the latest updates.)
I'm assuming you installed JuLIP
just via Pkg.add
, but if you ever checkout out the master branches on any of my other packages, then please
Pkg.free("JuLIP")
Pkg.free("NeighbourLists")
and then
Pkg.checkout("TightBinding")
Pkg.test("TightBinding")
If this works fine, then please close issue, otherwise just let me know. Anything else not working as expected just let me know or just make a PR. (I try to react quickly)
Just FYI - the main thing that needs to be (triple-)checked in this version of TightBinding
is the implementation of the spd Slater-Koster matrices. I'm somewhat confident that the sp Slater-Koster implementation is ok.
That's great Christoph! All tests pass.
spd
is the part of the periodic table I now inhabit, so I'll check it against the tbe
codes once I get our model Hamiltonian implemented.
Many thanks! Reading the code, there's some really nice features you've implemented.
Currently the package is a little broken - I assume due to your ASE and JuLIP packages changing underneath it.
In fixing, I got as far as:
But still crunch to a halt with:
If you dig down, you can see that it is breaking on this, from matrixassembly.jl:180
At this point I'm pretty lost! I guess the ASE NeighbourList object is arriving not as the expected MultiLevel type, and therefore doesn't have the necessary fields...