ABINIT Discussion Forums

Hello,
I am trying to do a response function calculation (following tuturial 1) on a cubic KNbO3 system.
With a 4x4x4 k point sampling grid the calculations work fine but for the subsequent finer grids
like 6x6x6 or 8x8x8 etc the log file shows the error


-P-0000 fxphas : BUG -
-P-0000 The eigenvector number 1 has zero norm.
-P-0000
-P-0000 leave_new : decision taken to exit ...

The output file shows a bunch of "NaN"s in the DDB.
In the data set2, where it calculates the derivative of the wavefunction with k it shows something
like

iter 2DEtotal(Ha) deltaE(Ha) residm vres2
ETOT*** -32.767253538900 -7.105E-15 6.328E-02 0.000E+00
ETOT*** -32.767253538900 0.000E+00 2.004E-01 0.000E+00
ETOT*** -32.767253538900 7.105E-15 1.258E-01 0.000E+00
ETOT*** -32.767253538900 0.000E+00 1.101E-01 0.000E+00
ETOT*** -32.767253538900 0.000E+00 2.801E-01 0.000E+00
ETOT*** -32.767253538900 0.000E+00 1.434E-01 0.000E+00
ETOT*** -32.767253538900 0.000E+00 4.474E-01 0.000E+00
ETOT*** -32.767253538900 0.000E+00 1.755E-01 0.000E+00
ETOT*** -32.767253538900 0.000E+00 5.574E-01 0.000E+00
ETOT*** -32.767253538900 0.000E+00 2.014E-01 0.000E+00
ETOT*** -32.767253538900 0.000E+00 6.931E-02 0.000E+00

scprqt: WARNING -
nstep= 5000 was not enough SCF cycles to converge;
maximum residual= 6.931E-02 exceeds tolwfr= 1.000E-20

The residual does not seem to converge but keeps on fluctuating.

I was wondering if anybody has any suggestions or pointers regarding this.
Thanks in advance,
Mohua

Read viewtopic.php?f=20&t=251

and provide more information, with a small input file which reproduces your problem.