The ABINIT Users Mailing List ( CLOSED )

["张�s" <zhangting1980323@gmail.com>]

Dear abinit users:

 

     Thanks to Matthieu and Xavier's help, I've successfully performed the phonon calculation at arbitrary q-vector in CNT system. Now another question comes: when I want to get the phonon eigenvector with anaddb code, obvious error appears. Because with an arbitrary q-point in BZ,  there is not any uniform q-mesh and only one q-point's DDB is calculated. So I have to use just this q-point in anaddb code to generate the D-matrix. As a result, with some q-points(for example larger than ( 0.0  0.0  0.30)),  the phonon freqs calculated are completely wrong, even with some negative freqs. I'm sure this is a physical meaningless result, because the expected Kohn anomaly point, (0.0  0.0  0.36xx) is still quite far away, and with uniform 1x1x16 q-mesh, the phonon calculated near these q-points are perfect(q= 0.3125 and 0.3750 in 1x1x16 mesh calculation, the results are quite good).  I think this is because the asr fails under this condition. Only one q-point in anaddb is not enough to impose asr, but how could I do to get correct freq and eigenvectors with anaddb? or I have to do this job with other method? Thanks!

 

----------------------------------------------------anaddb.in which will make wrong results

!Input file for the ifc code. Analysis of the q=(0.0 0.0 0.3)'s DDB                      

!Flags
 ifcflag   1             ! Interatomic force constant flag

!Wavevector grid number 1 (coarse grid, from DDB)
  brav    1              ! Bravais Lattice : 1-S.C., 2-F.C., 3-B.C., 4-Hex.)
  ngqpt   1  1  1     ! Monkhorst-Pack indices
  nqshft  1             ! number of q-points in repeated basic q-cell
  q1shft  0.0 0.0 0.3

!Effective charges
     asr   1     ! Acoustic Sum Rule. 1 => imposed asymetrically
  chneut   1     ! Charge neutrality requirement for effective charges.

!Interatomic force constant info
  dipdip  1      ! Dipole-dipole interaction treatment

!Phonon band structure output for band2eps 
   eivec  1

!Wavevector list number 1 (Reduced coordinates and normalization factor)        
  nph1l    1      ! number of phonons in list 1                            

  qph1l   0.0000  0.0000  0.3000 1.0   

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

 Regards

                                                      Zhang Ting

                                                      Peking Univ.