[gpaw-users] Dielectric Function

[Jun Yan]

Hi, Torsten,

   chi(r,r',w) is in principle a non-local function and a microscopic 
quantity, it means give an external field at r and get a response at r'.
   but what you are referring is a macroscopic dielectric function, 
which takes average on the microscopic response. that's why eps_xx, yy, 
zz are three values and chi is a matrix.

Best,
Jun

Torsten Hahn wrote:
> Hi Jun,
>
> but shouldn't chi be a 3x3 tensor or at least the chi_xx, chi_yy, 
> chi_zz values should be stored somewhere? Do i miss something here?
>
> Thx,
> Torsten. 
>
> Am 12.01.2011 um 09:22 schrieb Jun Yan:
>
>> Hi, Torsten,
>>
>>    chi is calculated by solving Dyson's equation (see 
>> https://wiki.fysik.dtu.dk/gpaw/documentation/tddft/dielectric_response.html#interacting-density-response-function)
>>    it has a dimension of (Nw, NG, NG) , where Nw is the number of 
>> frequency points and NG is the number of planewave used, which is 
>> related to the planewave cutoff (ecut) you choose. For response 
>> function, it can be quite low compared to ground state calculations 
>> since the coulomb interaction scales with 1/|G|**2 and thus decays fast.
>>
>> Best,
>> Jun
>>
>> Torsten Hahn wrote:
>>> Hi Jun,
>>>
>>> thanks for clarifying this. How do i get access to e.g. chi? Following the example on the website and using the df.get_chi() method i get back an array of dimensions (481, 169, 169). Can you tell me what is written in there ?
>>>
>>> Best regards,
>>> Torsten.
>>>
>>> Am 11.01.2011 um 11:31 schrieb Jun Yan:
>>>
>>>   
>>>> Hi, Torsten,
>>>>
>>>>   Use  df1, df2 = df.get_dielectric_function()
>>>>   where df1 and df2 are the dielectric function calculated with and without local field effects using RPA. The ALDA kernel is not implemented for optical excitations since it doesnt improve results.
>>>>   usually one use np.real(df2) and np.imag(df2) to get the real and imaginary part of dielectric function.
>>>>
>>>>   The df.get_dielectric_function() works in two ways:
>>>>   1) start a new calculation if this object is void. but then you have to define a df object by (in the following example I write the least parameters you have to input. Please also provide a gpw file that contains the pseudo-wavefunction. you can get it by using the 'all' option in the write method : calc.write('filename.gpw', 'all')):
>>>>       from gpaw.response.df import DF
>>>>       df = DF('filename.gpw', q = np.array([0.0001,0,0], w=np.linspace(0,20,201)), optical_limit=True )
>>>>       df1, df2 = df.get_dielectric_function()
>>>>   2 ) output the values if they are available, but you have to save a previous calculation using
>>>>      df.write('filename.traj')
>>>>     and using
>>>>     df.read('filename.traj')
>>>>     df1, df2 = df.get_dielectric_function()
>>>>     you can also get these values.
>>>>
>>>> Best,
>>>> Jun
>>>>
>>>>
>>>> Torsten Hahn wrote:
>>>>     
>>>>> Hi Jun,
>>>>>
>>>>> Am 11.01.2011 um 09:50 schrieb Jun Yan:
>>>>>
>>>>>  
>>>>>       
>>>>>> Hi, Torsten,
>>>>>>
>>>>>> Do you mean that you want to get the diagonal elements (eps_x,y,z) of a 9 component dielectric tensor ?   If so, you can find tutorial and manual at https://wiki.fysik.dtu.dk/gpaw/tutorials/dielectric_response/dielectric_response.html#df-tutorial
>>>>>> https://wiki.fysik.dtu.dk/gpaw/documentation/tddft/dielectric_response.html#df-theory
>>>>>> The current version can't support calculating the off-diagonal elements (fx: eps_xy, eps_xz) of the dielectric tensor   .
>>>>>>
>>>>>> The dielectric function, which is both wave-vector (q) and frequency (w) dependent, is calculated in a different way for excitations with photons (q-->0) and electrons (finite q).  If you are interest in optical excitations, you can follow the example 1 in the above tutorial.
>>>>>>
>>>>>>    
>>>>>>         
>>>>> sorry for the unclear expression. Yes i need the diagonal elements of the dielectric tensor. My question is rather technical: How do i access the values? The functions get_dielectric_function() etc. seems not to give access to the values ...
>>>>>
>>>>> Thx,
>>>>> Torsten.
>>>>>  
>>>>>       
>>>
>>>   
>>
>> _______________________________________________
>> gpaw-users mailing list
>> gpaw-users at listserv.fysik.dtu.dk 
>> <mailto:gpaw-users at listserv.fysik.dtu.dk>
>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20110112/01bbdcfe/attachment-0001.html