ABINIT Discussion Forums

hello abinit's
I wanted to calculate the phonon Dispersion, calculation works well at gamma point but for other points that do not converge. I do not know where is the problem
my input file is in attachement


Perturbation wavevector (in red.coord.) 0.333333 0.000000 0.000000
Perturbation : displacement of atom 1 along direction 1
Found 2 symmetries that leave the perturbation invariant.
symkpt : the number of k-points, thanks to the symmetries,
is reduced to 48 .

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

-inwffil : will read wavefunctions from disk file FeBo_DS1_WFK
-P-0000 - newkpt: read input wf with ikpt,npw= 1 915, make ikpt,npw= 1 1829
-P-0000 - newkpt: read input wf with ikpt,npw= 2 1836, make ikpt,npw= 2 1836
-P-0000 - newkpt: read input wf with ikpt,npw= 2 1836, make ikpt,npw= 3 1836
-P-0000 - newkpt: read input wf with ikpt,npw= 3 1817, make ikpt,npw= 4 1817

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

-inwffil : will read wavefunctions from disk file FeBo_DS1_WFK
-P-0000 - newkpt: read input wf with ikpt,npw= 2 1836, make ikpt,npw= 1 1836
-P-0000 - newkpt: read input wf with ikpt,npw= 2 1836, make ikpt,npw= 2 1836
-P-0000 - newkpt: read input wf with ikpt,npw= 1 915, make ikpt,npw= 3 1829
-P-0000 - newkpt: read input wf with ikpt,npw= 4 1822, make ikpt,npw= 4 1822

loper3 : total number of electrons, from k and k+q
fully or partially occupied states are 4.400000E+01 and 4.400000E+01.
Initialisation of the first-order wave-functions :
ireadwf= 0
P newkpt: treating 180 bands with npw= 1836 for ikpt= 1 by node 0
P newkpt: treating 180 bands with npw= 1836 for ikpt= 2 by node 0
P newkpt: treating 180 bands with npw= 1829 for ikpt= 3 by node 0
P newkpt: treating 180 bands with npw= 1822 for ikpt= 4 by node 0
P newkpt: treating 180 bands with npw= 1822 for ikpt= 5 by node 1
P newkpt: treating 180 bands with npw= 1817 for ikpt= 6 by node 1
P newkpt: treating 180 bands with npw= 1814 for ikpt= 7 by node 1
P newkpt: treating 180 bands with npw= 1814 for ikpt= 8 by node 1
P newkpt: treating 180 bands with npw= 1821 for ikpt= 9 by node 2
P newkpt: treating 180 bands with npw= 1812 for ikpt= 10 by node 2
P newkpt: treating 180 bands with npw= 1812 for ikpt= 11 by node 2
P newkpt: treating 180 bands with npw= 1808 for ikpt= 12 by node 2
P newkpt: treating 180 bands with npw= 1828 for ikpt= 13 by node 3
P newkpt: treating 180 bands with npw= 1828 for ikpt= 14 by node 3
P newkpt: treating 180 bands with npw= 1825 for ikpt= 15 by node 3
P newkpt: treating 180 bands with npw= 1817 for ikpt= 16 by node 3
P newkpt: treating 180 bands with npw= 1817 for ikpt= 17 by node 4
P newkpt: treating 180 bands with npw= 1822 for ikpt= 18 by node 4
P newkpt: treating 180 bands with npw= 1813 for ikpt= 19 by node 4
P newkpt: treating 180 bands with npw= 1813 for ikpt= 20 by node 4
P newkpt: treating 180 bands with npw= 1809 for ikpt= 21 by node 5
P newkpt: treating 180 bands with npw= 1820 for ikpt= 22 by node 5
P newkpt: treating 180 bands with npw= 1820 for ikpt= 23 by node 5
P newkpt: treating 180 bands with npw= 1804 for ikpt= 24 by node 5
P newkpt: treating 180 bands with npw= 1806 for ikpt= 25 by node 6
P newkpt: treating 180 bands with npw= 1806 for ikpt= 26 by node 6
P newkpt: treating 180 bands with npw= 1792 for ikpt= 27 by node 6
P newkpt: treating 180 bands with npw= 1820 for ikpt= 28 by node 6
P newkpt: treating 180 bands with npw= 1820 for ikpt= 29 by node 7
P newkpt: treating 180 bands with npw= 1810 for ikpt= 30 by node 7
P newkpt: treating 180 bands with npw= 1802 for ikpt= 31 by node 7
P newkpt: treating 180 bands with npw= 1802 for ikpt= 32 by node 7
P newkpt: treating 180 bands with npw= 1806 for ikpt= 33 by node 8
P newkpt: treating 180 bands with npw= 1816 for ikpt= 34 by node 8
P newkpt: treating 180 bands with npw= 1816 for ikpt= 35 by node 8
P newkpt: treating 180 bands with npw= 1812 for ikpt= 36 by node 8
P newkpt: treating 180 bands with npw= 1828 for ikpt= 37 by node 9
P newkpt: treating 180 bands with npw= 1828 for ikpt= 38 by node 9
P newkpt: treating 180 bands with npw= 1825 for ikpt= 39 by node 9
P newkpt: treating 180 bands with npw= 1817 for ikpt= 40 by node 9
P newkpt: treating 180 bands with npw= 1817 for ikpt= 41 by node 10
P newkpt: treating 180 bands with npw= 1822 for ikpt= 42 by node 10
P newkpt: treating 180 bands with npw= 1813 for ikpt= 43 by node 10
P newkpt: treating 180 bands with npw= 1813 for ikpt= 44 by node 10
P newkpt: treating 180 bands with npw= 1809 for ikpt= 45 by node 11
P newkpt: treating 180 bands with npw= 1820 for ikpt= 46 by node 11
P newkpt: treating 180 bands with npw= 1820 for ikpt= 47 by node 11
P newkpt: treating 180 bands with npw= 1804 for ikpt= 48 by node 11


iter 2DEtotal(Ha) deltaE(Ha) residm vres2
-ETOT 1 2582696.5824959 2.582E+06 2.504E+01 5.386E+10
ETOT 2 2102426164.6901 2.100E+09 4.155E+03 4.332E+13
ETOT 3 2858490.3786928 -2.100E+09 3.568E+03 5.406E+10
ETOT 4 1162310.1268631 -1.696E+06 8.758E+02 2.015E+10
ETOT 5 1190021.1841785 2.771E+04 9.691E+01 2.532E+10
ETOT 6 29130.937598120 -1.161E+06 5.567E+01 4.214E+08
ETOT 7 10067.434177251 -1.906E+04 2.762E+01 1.906E+08
ETOT 8 72.904035136211 -9.995E+03 1.275E+01 1.034E+06
ETOT 9 16.189033707846 -5.672E+01 7.646E+00 8.165E+04
ETOT 10 22.813691574573 6.625E+00 3.311E+00 2.077E+05
ETOT 11 18.555502227962 -4.258E+00 2.105E+00 1.286E+05
ETOT 12 19.970003390068 1.415E+00 8.908E-01 6.616E+04
ETOT 13 16.984162679842 -2.986E+00 9.264E-01 5.913E+04
ETOT 14 63.133630047078 4.615E+01 7.548E-01 4.991E+05
ETOT 15 18.276276763790 -4.486E+01 1.442E+00 2.035E+04
ETOT 16 116.53773025635 9.826E+01 1.184E+00 1.693E+06
ETOT 17 12.912215286574 -1.036E+02 6.808E-02 2.900E+03
ETOT 18 14.085012171117 1.173E+00 1.390E+00 2.672E+04
ETOT 19 60.153558295754 4.607E+01 2.055E-01 8.703E+05
ETOT 20 13.202376142594 -4.695E+01 1.419E-01 1.430E+04
ETOT 21 19.070935888370 5.869E+00 1.334E-01 1.215E+05
ETOT 22 20.104316099486 1.033E+00 1.699E-01 8.134E+04
ETOT 23 22.415302073775 2.311E+00 2.250E-01 1.187E+05
ETOT 24 16.408112222822 -6.007E+00 2.391E-01 2.257E+04
ETOT 25 24.143277351434 7.735E+00 3.528E-01 1.934E+05
ETOT 26 52.316563532230 2.817E+01 3.618E-01 2.873E+05
ETOT 27 73.062845515788 2.075E+01 5.541E-01 7.620E+05
ETOT 28 31.194009265978 -4.187E+01 5.687E-01 3.501E+05
ETOT 29 528.70033285783 4.975E+02 8.596E-01 1.019E+07
ETOT 30 177.00581830748 -3.517E+02 8.734E-01 2.258E+06
ETOT 31 253.58339008075 7.658E+01 1.350E+00 2.140E+06
ETOT 32 271.90603253330 1.832E+01 1.346E+00 1.954E+06
ETOT 33 200.61877473398 -7.129E+01 2.995E-03 1.015E+06
ETOT 34 5032.5847097360 4.832E+03 1.375E+00 9.652E+07
ETOT 35 224.61886674024 -4.808E+03 2.119E-01 1.791E+06
ETOT 36 227.69995936727 3.081E+00 1.253E-01 1.723E+06
ETOT 37 15296.093795337 1.507E+04 1.280E-01 2.326E+08
ETOT 38 5223.0681511120 -1.007E+04 1.322E-01 9.654E+07
ETOT 39 7013.8467579661 1.791E+03 2.093E-01 1.190E+08
ETOT 40 5127.5044482076 -1.886E+03 2.163E-01 2.276E+07
ETOT 41 12176.427816549 7.049E+03 3.550E-01 1.030E+08
ETOT 42 4792.8650717127 -7.384E+03 2.929E-01 1.573E+07
ETOT 43 9820.3788405442 5.028E+03 5.123E-01 1.922E+08
ETOT 44 19843.346953819 1.002E+04 4.651E-01 1.105E+08
ETOT 45 116773.46448778 9.693E+04 5.130E-01 9.038E+08
ETOT 46 44658.671016227 -7.211E+04 6.879E-01 7.252E+08
ETOT 47 495776.10187867 4.511E+05 9.725E-01 1.007E+10
ETOT 48 193046.68231029 -3.027E+05 9.193E-01 3.412E+09
ETOT 49 185262.33371659 -7.784E+03 9.928E-01 1.324E+09
ETOT 50 167471.33869366 -1.779E+04 1.373E+00 8.579E+08
ETOT 51 111988.76614857 -5.548E+04 1.281E+00 5.304E+08
ETOT 52 6843654.0750648 6.732E+06 1.397E+01 1.276E+11
ETOT 53 2967120.5122031 -3.877E+06 9.058E+00 4.647E+10
ETOT 54 219034.02837964 -2.748E+06 5.237E+00 1.100E+09
ETOT 55 9401427.1528353 9.182E+06 1.851E+01 1.263E+11
ETOT 56 5977247.3118967 -3.424E+06 3.140E+01 1.127E+11

Hello

Try setting usexcnhat=1 (default is -1). It should help for convergence issues with the conjugate gradient.

Boris

thank you Boris for your reply.
I tried but it does not converge

Hi!

Did you check whether:

1] Your 1st dataset is well converged,

2] You use the same input parameters in RF calculation as in your relaxation calculation (ecut, ecutsm, dilatmx, ngkpt, shiftk, nshiftk,...),

3] Try putting npulayit to higher value that default (nnsclo and nline, also).

Igor

lagoun brahim wrote:thank you Boris for your reply.
I tried but it does not converge

Hi again

It may be a parallelisation issue. How many cpus are you using for the calculation ?

Thanks to the symmetries, the number of kpts has been reduced to 48. If you use more than 48 cores, abinit will switch to band parallelisation, which does not work yet and the conjugate gradient fails.

Boris

Boris wrote:

lagoun brahim wrote:thank you Boris for your reply.
I tried but it does not converge

Hi again

It may be a parallelisation issue. How many cpus are you using for the calculation ?

Thanks to the symmetries, the number of kpts has been reduced to 48. If you use more than 48 cores, abinit will switch to band parallelisation, which does not work yet and the conjugate gradient fails.

Boris

thank you boris again
i use 12 cores only

ilukacevic wrote:Hi!

Did you check whether:

1] Your 1st dataset is well converged,

2] You use the same input parameters in RF calculation as in your relaxation calculation (ecut, ecutsm, dilatmx, ngkpt, shiftk, nshiftk,...),

3] Try putting npulayit to higher value that default (nnsclo and nline, also).

Igor

thank you for your reponse
1-yes
2-yes
3- i tried to increase nnsclo and nline but i have the same problem (i tried iscf 7 also)
thanks again