The ABINIT Users Mailing List ( CLOSED )

[xenophon.krokidis@scienomics.com]

Dear Xavier,

Thank you for the support. I wanted also to know if the tolsym issue is also 
responsible for the problem I see when I use the conventional cell (using 
spgroup=203, spgorig=2 and checkprim=0). Indeed Abinit finds that the box 
should contain 144 (that a primitive cell) atoms and not 576 (that of the 
conventional cell). I should expect that Abinit would report the usual 
message that the cell is not a primitive but would continue because of 
checkprim=0.

Once again thanks for the very valuable and appreciated support.

Best regards,

Xenophon
---- Envoyé avec BlackBerry® d'Orange ----

-----Original Message-----
From: Xavier Gonze <xavier.gonze@uclouvain.be>
Date: Sat, 20 Mar 2010 18:27:26 
To: <forum@abinit.org>
Subject: Re: [abinit-forum] Problem with symmetry finder

Dear Xenophon,

It is a numerical precision issue.

On 19 Mar 2010, at 12:27, Xenophon Krokidis wrote:

> Dear All,
>
> I am attaching here the input and output files of an Abinit  
> calculation on the primitive cell of faujasite (space group FD-3,  
> origin 2) with Abinit 5.8.4 and 6.0.2. In both cases the space group  
> found is not correct. Furthermore, both Abinit versions differ:  
> 5.8.4 finds at least a space group with less operations, while 6.0.2  
> is not able to find a space group at all. In the second attachment  
> there is the input for the same system, but this time using the  
> conventional cell. Here Abinit fails with both versions to generate  
> the full list of atoms.

I had a look at the primitive cell case.
If one uses

  rprim    1.0000000000   0.0000000000   0.0000000000
           0.5000000000   sqrt(0.75d0)   0.0000000000
           0.5000000000   sqrt(0.083333333333d0)   
sqrt(0.6666666666666d0)

or (even better)

  rprim 0  sqrt(.5)  sqrt(.5)
        sqrt(.5) 0 sqrt(.5)
        sqrt(.5) sqrt(.5) 0

The space group is correctly recognized.
Interestingly, with your original rprim,

rprim    1.0000000000   0.0000000000   0.0000000000
              0.5000000000   0.8660254038   0.0000000000
              0.5000000000   0.2886751346   0.8164965809
(that is accurate up to nine digits)
I would have expected the use of tolsym to allow ABINIT
to recognize the Bravais lattice. But I tested tolsym 1.e-3,
without success.
I will try to see what is problematic.

Best wishes,
Xavier

>
> Best regards,
>
> Xenophon
>
> <FAU_conventional.zip><FAU_primitive.zip><Xenophon_Krokidis.vcf>