[Pw_forum] Access to 2nd order derivatives

[Marcel Mohr]

On Mon, 19 Mar 2007, Paolo Giannozzi wrote:

>
> On Mar 14, 2007, at 22:00 , Marcel Mohr wrote:
>
>> I wonder if it is possible to obtain the 2nd derivative of the total energy
>> for a user defined set of atomic coordinates (not a real but infinitely
>> small displacement from the ground state).
>
> in principle, it is, and once upon a time it was possible.
> Now there is the 'single mode calculation' (example 02),
> in which you can choose a specific displacement pattern,
> but only among those automatically computed by the code.
>
> P.
> ---

Thank you very much for your answer.

I summarize.
One can "move" one atom, and that in the directions x, y, and z.
The variable in pw.x modenum 3 is corresponds to a "displacement" of 
the first atom in direction z. 
ph.x computes then the dynamical matrix entries for this mode, from which 
one can make conclusions on the 2nd order derivative for the atom 3 in 
direction z. 
If I have the whole dynamical matrix I can diagonalize it and obtain 
frequencies and eigenvectors.
If there are some errors please correct me.
Otherwise thanks again.

Cheers
Marcel