The ABINIT Users Mailing List ( CLOSED )

["Erik Ylvisaker" <eylvisaker@physics.ucdavis.edu>]

hmm from looking at your input files, I think you want to either generate 
your q-point grid with no shift, or use a 12x12x12 k-point mesh.  Abinit 
needs to calculate matrix elements between wavefunctions at k and k+q. 
Shifting the q-grid will put some of the k+q points in between the 6x6x6 
grid points used for your k-points, so it won't find those in the WFK file.

----- Original Message ----- 
From: "zhangting" <zhangting@pku.edu.cn>
To: <forum@abinit.org>
Sent: Wednesday, May 10, 2006 9:43 AM
Subject: Re: Re: [abinit-forum] Question about the k-point sampling in 
phonon calculation


Dear Erik Ylvisaker:

    Thanks very much for your replying my question. When I use "shiftk 0.5 
0.5 0.5", the mismatch, of which the number of

K-points between the datasets with & without time-reversal symmetry, do 
disappear. But there immediately appears another

trouble in the calculation, the code escapes with such messeges:
-------------------------------------

inwffil: ERROR
The file phonono_DS1_WFK cannot be used to start the
present calculation. It was asked that the wavefunctions be accurate, but
at least one of the k points could not be generated from a symmetrical one.
dksqmax=    3.331709E-04
Action: check your wf file and k point input variables.

leave_new : decision taken to exit ...

--------------------------------------

     This error occurs at the q= 0.8333 0.8333 0.8333, or the 4th dataset, 
while the phonon at Gamma is correctly

completed. What does this message mean? Should I increase the q-point mesh 
to 8x8x8 or should I do something else?

The q-point is generated with scheme in the trf2_2.in, and the k-point 
density is 6x6x6. shiftk here is set to 0.5 0.5

0.5. The .in files are attached.


======================================================================================
think this happens because you are using shiftk 0 0 0.  Time reversal 
symmetry allows half the Brillouin zone to not be calculated.  But without a 
shift, you end up picking points like Gamma, which has no time reversed 
point, and points at the zone surface, whose time reversed point is 
separated by a reciprocal lattice vector, so it can't be eliminated by time 
reversal symmetry because the choice of grid didn't include it.  I think if 
you set shiftk to 0.5 0.5 0.5 you will see the results you expect.




　　　　　　　　zhangting
　　　　　　　　zhangting@pku.edu.cn
　　　　　　　　　　2006-05-11