ABINIT Discussion Forums

Dear All:

I am calculating electron-phonon properties. The unit cell contains 5 atoms, with k-mesh 8x8x8, and q mesh 2x2x2. There are 8 in-equivalent q points. This leads to a 5x3x8=120 perturbations in MRGGKK input, and 8 DDB files in MRGDDB input.

the mrgddb and mrggkk seems both OK. However, when running anaddb, it says:

read_gkk.F90:302:ERROR
found too many qpoints in GKK file wrt anaddb input
nqpt_anaddb nqpt_gkk = 6 7


I have no single clue that what might be going on, that where the number "6" and "7" come from. Any suggestions? Many thanks!!!

Best,


ps: Due to the too many problem I've met, I am serious thinking of co-authorship if we could conquer these difficulties together. Looking for help sincerely.

Hello,

Please read the nettiquette - it gets really hard to guess what is going on.

For 8 inequivalent q from a 2x2x2 grid that means you have no symmetries at all?

The values read_gkk is complaining about are
1) the number it has calculated from your ngqpt and shifts
2) the number it reads from your main GKK file

the two should agree of course, and it looks like it has found some symmetries, but without any log or output or input or anything it's a bit hard to tell.

Also check that your GKK files are merged grouped together by qpt.

Finally you might try the latest abinit/anaddb version. From 7.6.4 there is a new procedure to calculate the matrix elements separately, on a k-point grid of your choosing (check the doc and tutorials). You may have to be careful to specify the Fermi level explicitly with elph_fermie if you use this non self-consistent scheme.

cheers

Matthieu