[Wien] [Wien2k User Group] Lattice stability in bcc, fcc and hcp and other crystal structures

[Ghosh SUDDHASATTWA]

Dear Stefaan, 
Thanks for the reply. Let me explain my position. Let us take an example of
uranium. 
The stable phases of uranium are C-centered orthorhombic (alpha), primitive
tetragonal(beta) and fcc (gamma). 
Getting the atomic positions of these phases from ICSD, I can calculate the
ENE (ground state energy). I need not do any min_lapw since these are stable
phases. 
Now I imagine uranium in hcp phase of any metal like any refractory element
(which are usually in hcp) (Blaha's papers on hcp metals are a good starting
point). I replace the hcp crystal structure with uranium only. Then I
calculate the ENE. 
This is the ENE of uranium in the metastable phase. I guess I am right. 

Now for this metastable phase, much work probably has been done before. We
can approach this problem in many ways and what are my requirements as you
said before are to be taken into consideration. 
Keeping in mind the time and my efforts, does min_lapw fulfill my
requirements? 
I ask this question again because of the large experience of Wien2k
developers so that all my efforts can be seen in a proper perspective. 
Since I want uranium to be in hcp phase, I should not be carrying out a
shape relaxation calculation. 
Is my approach right? 


Thanks in advance 
Suddhasattwa Ghosh 

-----Original Message-----
From: wien-bounces at zeus.theochem.tuwien.ac.at
[mailto:wien-bounces at zeus.theochem.tuwien.ac.at] On Behalf Of Stefaan
Cottenier
Sent: Friday, February 19, 2010 5:18 PM
To: A Mailing list for WIEN2k users
Subject: Re: [Wien] [Wien2k User Group] Lattice stability in bcc, fcc and
hcp and other crystal structures


> 1. I want optimized positions in case of experimentally known crystal
> structures. Does it really help? Is it worth doing? I may be wrong or may
be
> doing what is not required. 
> 
> 2. I want to do only internal coordinates optimization for metastable
> phases. Is it incomplete? 

It all depends on your problem -- I don't think this mailing list can 
answer that for you.

Don't put the horse behind the carriage: first define for yourself what 
is the question you want to answer. Then examine what is the 'best' 
computational strategy to answer that question (where 'best' is your 
compromise between speed, accuracy, effort and unavoidable limitations). 
Finally, carry out the calculations, and don't make interpretations that 
are not allowed under the approximations that you have adopted.

Stefaan

_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien