[Pw_forum] Phonon out of q=0 for a magnetic insulator

[Merlin Meheut]

Dear pwscf users and developpers,

Best wishes for the New Year!

I have trouble with phonon calculations on magnetic insulators: the
calculations works well at gamma (k=0) point, and I am able to compute
dielectric tensor and effective charges, but as soon as I try out of the
zone center, the calculation never works, with errors such as (in the ph
output):

------------------------------------------------------------
---------------------------
    Representation #  1 mode #   1

     Self-consistent Calculation
     kpoint   1 ibnd  75 solve_linter: root not converged  1.071E-01
     kpoint   2 ibnd  75 solve_linter: root not converged  1.924E-01
     kpoint   3 ibnd  75 solve_linter: root not converged  1.745E-02
     kpoint   4 ibnd  75 solve_linter: root not converged  1.722E-01
------------------------------------------------------------
-------------------------------

On the other hand, when I do the very same calculation, but just changing
the insulator state to a metallic state (in the preceeding scf step, that
is), then the calculation goes perfectly well. Is there a particular reason
why that shouldn't work? The electronic bands show that this system is
indeed an insulator with a significant gap.

Thank you very much for your help,

Best Regards,

Merlin Méheut

1/ scf input for insulator state:
&system
    ibrav =0,
    celldm(1) =  12.459185,
    nat =20, ntyp = 4,
    ecutwfc =80.0, ecutrho = 320.0,
    nspin=2, starting_magnetization(1)=0.5,
    tot_magnetization=4,
/&end

2/ for metallic state
&system
    ibrav =0,
    celldm(1) =  12.459185,
    nat =20, ntyp = 4,
    ecutwfc =80.0, ecutrho = 320.0,
    nspin=2, starting_magnetization(1)=0.5,
    occupations='smearing', smearing='mp', degauss=0.04,
/&end

3/ phonon input:
&inputph
   amass(1)=57.9353,
   amass(2)=39.9626,
   amass(3)=27.9769,
   amass(4)=15.9949,
   alpha_mix(1) = 0.5,
   ! ldisp=.true., nq1=2, nq2=3, nq3=2,
   tr2_ph =  1.0D-17,
   prefix='NiCPX',
   fildyn='mat.$PREFIX',
   lraman=.false.,
   epsil =.false.,
   trans =.true.,
   zue = .false.,
   zeu= .false.,
   outdir='/tmpdir/$LOGNAME/'
/&end
  -0.0006214  -0.0004751  -0.5041257


-- 
Merlin Méheut
adresse labo:
GET - OMP   - Université Paul Sabatier
14 avenue Edouard Belin
31400 Toulouse
FRANCE
tel: (+33) 5 61 33 26 17 <+33%205%2061%2033%2026%2017>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20170102/d51600ae/attachment.html>