[CP2K:477] Re: berry phase

[Teodoro Laino]

Hi Marc,

just to let you know that in the CVS of cp2k is now available the  
dipole (computed within FIST) also
for non-orthorhombic cells. Moreover there was also a bug fix for the  
old implementation of dipole (so
it is highly suggested, if you're interested in dipoles, to update  
cp2k). Let me know in case you (or anybody of
you out there ;-) ) should find possible problems.

Cheers,
teo

On 11 Dec 2007, at 11:24, marc wrote:

>
> OK, thank you very much for the information.  I'm going into it a
> little bit more and let you know how it evolves.
>
> until then,
>
> take care!
>
> marc
>
> On Dec 11, 12:22 am, Teodoro Laino <teodor... at gmail.com> wrote:
>> Ciao Marc,
>>
>> On 10 Dec 2007, at 17:41, marc wrote:
>>
>>
>>
>>> Hey all,
>>
>>> I have a problem concerning the calculation of dipole moments in
>>> cp2k.  I'm doing classical  molecular mechanics using force  
>>> fields, so
>>> can I use cp2k for this?  What we want to do is calulate an IR-
>>
>> Yes you can.
>>
>>> spectrum using M.D..  I already ran a calculation, and cp2k gives
>>> three dipole moments for each simulation step:
>>
>>> 1. MM DIPOLE BERRY PHASE ( A.U.)|
>>> 2. MM DIPOLE BERRY PHASE (Debye)|
>>
>> These are the dipoles.. while this one is the derivative of just one
>> of the two:
>>
>>> 3. MM DIPOLE (NON BERRY PHASE) DERIVATIVE (A.U.)|
>>
>>> Has the berry phase output any revelance when doing classical M.D.
>>> using force fields?
>>
>> Yes sure! Unless you work in free boundary conditions (i.e. isolated
>> systems), you can't
>> define the dipole moment as you would normally do i.e. (in latex
>> notation)
>>
>> d = - \int \rho(r) * r   (in au units)
>>
>> This equation looks very trivial, but it exploits an essential fact:
>> the density of any finite N charge system vanishes exponentially at
>> infinity.
>> Moreover, the problem with the expression above is that the integral
>> is ill-defined due to the unbounded nature of the position operator.
>> Pioneering work (only in the early 1990s!!) can be considered:
>>
>> (1) M. Posternak, A. Baldereschi, A. Catellani, R. Resta, Phys. Rev.
>> Lett.  64 (1990) 1777
>>
>> followed by real breakthrough in 1992, with the modern theory of
>> polarization:
>>
>> (2) R. Rest a, Ferroelectrics 136 (1992) 51;
>> (3) R. D. King-Smith, D. Vanderbilt, Phys. Rev. B 47 (1993) 1651;
>> (4) D. Vanderbilt, D. King-Smith, Phys. Rev. B 48 (1993) 4442;
>>
>> For more details based on the Berry-phase in the polarization field
>> you may have a look at:
>>
>> (5) R. Resta, J. Phys. Condens 12 (2000) R107.
>>
>> AN IMPORTANT WARNING: the version implemented in FIST gives the right
>> number only for orthorhombic cells.
>> obviously it can be expanded also to non-orthorhombic cells.
>>
>> Let me know if you need I extend it to non-orthorhombic cells.
>>
>> Have Fun! ;-)
>> Teo
>>
>>> I also was wondering what the berry phase
>>> actually has to do with dipole moments?  Can someone give a  
>>> reference
>>> which explains this berry phase related to classical M.D.?
>>> regards,
>>
>>> marc
> >