ABINIT Discussion Forums

Hi,
So I ran a relaxation calculation with the xred in the P63/mmc (#194) space group and forced abinit to find a sol'n within this space group by using the spgroup variable and setting it equal to 194.

this calc returns xred in the 194 group. fine so far.

Now, 194 is a hexagonal space group. so, I do not define rprim, but I define angdeg as 90 90 120. And I supply the acell coords in the proper order.

So the I run a berry polarization using:
rfdir 1 1 1
berryopt -1 (since my occopt 1).

I use a kpt of 8 8 8, with a shiftk 0.5 0.5 0.5 (side note: a kpt of 6 6 6 ... leads to abinit hanging...)

I expect to get a polarization in x, y, and z all equal to 0.

I get:
Electronic berry phase: -0.217863665E-17 -0.690143852E-18 -0.210435482E-01
Expectation value (PAW only): 0.000000000E+00 0.000000000E+00 0.000000000E+00
Ionic: 0.626624034E-15 -0.888009271E-15 -0.996822686E-16
Total: 0.624445397E-15 -0.888699414E-15 -0.210435482E-01

So, all is "good" except the z dir polarization. It is too high in abs. val to be ignored, yes?

What could I be doing wrong?

My output xred coordinates from my relaxation calc (which are the input coords to my Polar. calc) are listed below:
0 0 0
0 0 0.5
0.33333333 0.66666667 0.75
0.66666667 0.33333333 0.25
0.17426253 -0.17426253 0.25
0.17426253 0.34852506 0.25
-0.34852506 -0.17426253 0.25
-0.17426253 0.17426253 0.75
-0.17426253 -0.34852506 0.75
0.34852506 0.17426253 0.75

The ntypat is 3 and typat is 2*1 2*2 6*3

thanks for looking into this,
John B.

I'm not sure whether this is ok or not (I'm the main developer currently working on berryopt with PAW). If you send me your input files I'll look into it more carefully, I can't tell from what you posted.