Generating BQB formatted cube files through CP2K

[Matthew Graneri]

Hi there,

I'm following the IR spec tutorial in order to generate an IR spectrum of my system. As far as I am aware, the BQB compression system has been added to CP2K, but I can't figure out how to generate my electron density into this compressed style of file.

The tutorial mentions says that, to print the electron densities, you need to include:

&PRINT

   &E_DENSITY_CUBE

       STRIDE 1 1 1

etc..

In the documentation for the E_DENSITY_BQB, it doesn't list any 'STRIDE' option (or equivalent), so it doesn't sound like it replaces the E_DENSITY_CUBE command... Is it possible to generate BQB electron density files in CP2K? If so, how can this be done?

Also, the tutorial mentions that we are using the Voronoi integration, and there is a VORONOI block in the DFT/PRINT block, but the tutorial never mentions it... Is that related, at all?

Thanks!

Matthew~

[Sam Broderick]

Hi Matthew

So if you use the TRAVIS program, you can skip Voronoi in cp2k. TRAVIS will do this for you (it wasn't clear to me either reading the tutorial).

The stride is not to be confused with &REFTRAJ stride for &E_DENSITY_CUBE. This is a sort of down-sampling to reduce file size. Since bqb is is design to compress, there's no need for a stride. This should be all you need.

&E_DENSITY_BQB

    FILENAME =result-E0.bqb

&END E_DENSITY_BQB

TRAVIS reads the bqb natively.

Hope this helps

Sam

[Matthew Graneri]

Hi Sam,

I'm not quite sure if I understand...

I understood the stride to be more like the 'resolution' of the electron density. That being said, I'm totally unfamiliar with AIMD, so I'm sure you know better than me.

I suppose what I'm asking is if it's possible to generate a .bqb file without having to generate a full .cube file (because the .cube file is prohibitively large)... or, is it possible for each frame of the .cube file to be compressed into a .bqb file and then get discarded? Again, my main concern is the memory requirements...

Thanks for your help!

Matthew

[Sam Broderick]

So there are two strides that I have encountered in CP2K, which is confusing.

Stride in &REFTRAJ: STRIDE 8 indicates to take TIME STEPS 1, 9, 17, etc.

Stride in &E_DENSITY_CUBE: STRIDE 1 1 1 indicates to skip no MESH locations, but 2 2 2 would take every 2nd in x, y, z.

There is no need for the .cube if you output .bqb, but you will need either TRAVIS or bqbtool to read it.