The ABINIT Users Mailing List ( CLOSED )

[mjv500@york.ac.uk]

Hello,

it sounds like you are only calculating the q=0 phonons. I don't know where 
you are extracting the gkk from, but you normally need many more phonon 
qpoints. The anaddb code will interpolate for q which are not in your 
default set. If you have too few qpoints you can get bad interpolation and 
the phonon linewidth can go negative artificially. Check if any of your 
phonons have gone soft (w<0) - this could also mess up the gkk and 
linewidths.

On a more general note, the gkk matrix elements don't absolutely need to 
have a particular sign; the observable quantity is the linewidth, which is 
related to the gkk squared. There are, however, hand waving arguments for 
the gkk being positive in stable systems.

Concerning the wires, most monoatomic straight wires are unstable wrt 
deformation into a zigag (see Phys. Rev. B 74 , 153408 (2006) and the 
references to Batra and others therein), so this could be the unstable mode 
you are seeing in a 2 atom unit cell.

Matthieu


On Jun 29 2008, [GB2312] ÕŜs wrote:

> Dear All:
>
>    I'm calculating the elph matrix with ABINIT code. I find that the 
> calculated elph matrix elements change their sign via k-point, and I feel 
> this quite strange. For example, I'm calculating a testing system which 
> is a carbon atom chain. The system has two carbon atoms in an unit cell, 
> 8 bands, and the fermi level is located between iband = 4 and iband =5. I 
> expect the elph coupling between these two bands near fermi k-point(k = 
> -0.5 , k=0.5) is important. So with phonon's q =0, I calculated the elph 
> matrix g(k,k+q,ib1,ib2). The results seems strange because the elph 
> matrix element change their sign via k-point, see below:
>
> q =0, branch =6 (the LO mode, which should be the strongest el-ph 
> coupling mode in C2 chain) ib1=4,ib2=5, k=-0.475, gkk(pert=atom1, 
> z-direction)=-2.432264e-002 +i*8.642045e-001, 
> gkk(pert=atom2,z-direction)=2.432269e-002 +i*-8.642045e-001 k=-0.425, 
> gkk(pert=atom1, z-direction)=2.397082e-002 +i*-8.564656e-001, 
> gkk(pert=atom2,z-direction)=-2.397087e-002 +i*8.564656e-001 k=-0.375, 
> gkk(pert=atom1, z-direction)=-2.323420e-002 +i*8.402542e-001, 
> gkk(pert=atom2,z-direction)=2.323425e-002 +i*-8.402542e-001 k=-0.325, 
> gkk(pert=atom1, z-direction)=-2.204158e-002 +i*8.139450e-001, 
> gkk(pert=atom2,z-direction)=2.204164e-002 +i*-8.139450e-001 k=-0.275, 
> gkk(pert=atom1, z-direction)=2.027373e-002 +i*-7.746554e-001, 
> gkk(pert=atom2,z-direction)=-2.027378e-002 +i*7.746554e-001 We can see 
> that from k= -0.475 --> k= -0.425, the gkk change sign. Also I can not 
> find any rule in the sign change. Why these gkk element change sign via 
> k-points? And is this result correct? Thanks! My input file is attached.
>
>
>
> Regards
>
>
>                                                        Zhang Ting
>                                                        Peking Univ.
>                                                        Jun, 29th. 2008