The ABINIT Users Mailing List ( CLOSED )

[Xie Yu <lazycle@gmail.com>]

Dear Fan,

When you use the ab initio package for a particular properties, you have to test your pseudopotential and the parameters very carefully.

There are some parameters related to the system besides the kpoints and ecut, i.e. smearing and occupation. You should test them before you do the real calculation.

In my point of view, you can calculate the frequencies at the gamma point with different parameters until the frequencies are converged . Then you can use the converged parameters for the whole phonon dispersion calculation.

PS: 17 Ha/ 34 Ry is quite small for a norm conserving pseudopotential.

Best,

Yu

On Sun, Dec 20, 2009 at 1:25 AM, jinhong zhang <zhangjinhong2009@gmail.com> wrote:

Dear Abiniters

 

I recently calculated phonon dispersion of B12P2 following the RF tutorial. I first optimized the geometry of the material by GGA and the tensor stress in the material reached e-7, which i believe is accurate enough. And the ngkpt I chose first time is 4 4 4 and ecut is 15. The result of phonon dispersion turned out to be bad, some low frequencies appeared negative and even reached -700cm-1 (but the acoustic frequencies are 0 at Gamma point). And then I changed the ngkpt to 8 8 8 and ecut to 17, and optimized and calculated again, the graph turned out to be better but still not good, only 3 acoustic frequencies represented negative beyond Gamma point this time, and the magnitude reached about -500cm-1max.

 

I was wondering the reason. Is the problem of ABINT or somewhere i set incorrect? (I preferred the later:P) or the stability of this material is not good? But my teacher had calculated the phonon dispersion of B12O2-a similar one-by Material Studio, which the result was good.

 

Hope any Abiniter can give me some advices and help me solve the problem, i would be very appreciated.

 

Best regards

 

Fan