[Sax-project-users] how to calculate transition energy of arbitrary two levels

[marsamos at democritos.it]

Dear user, Let's first talk about gw. GW quasi particle (the band 
structure ...) gives you the total energy differences between 
E_s(N+1)-E(N) for the s unoccupied states and E(N)-E_s(N-1) for the s 
occupied states. So if you have a molecule, the GW energy of the top of 
the HOMO (top of valence) gives you minus the ionization potential. The 
GW energy of the LUMO (bottom of the conduction) gives you minus the 
electron affinity. So GW gives you how it cost to create a hole within 
the occupied states and put it in infinity or how it cost to take an 
electron from infinity and put it in an unoccupied state. So in a one 
particle picture the transition energy for driving an electron from 
occupied band 15 to the unoccupied level 18 is just the GW 
quasiparticle energy difference between these two levels. But if you do 
this, you are assuming that the hole created in the level 15 is not 
interacting with the electron created in level 18, and that is not true 
as the attractive coulomb interaction between them maybe very high. 
That is why is important in semi-conductors and insulators to include 
the electron-hole interaction, i.e. to solve the BSE. But the BSE is 
NOT a one-particle picture, BSE is two particle picture with many-body 
stuff. The energies (<eigenvalues tag in the bse file) of the bse are 
already transition energies that include the electron-hole interaction. 
But, these transitions are NOT build with two single particle levels, 
they are build with a COMBINATION of single particle levels. The tag 
<eigenvectors of the file bse, gives you the weight of any two single 
particle levels that contributes to the transition.

For understanding better all these theoretical points, which is 
fundamental to be able to understand the calculations them self, I 
suggest to read the tutorial (section Docs of 
www.qe-forge.org/projects/sax-project/) and also to read all the papers 
of Hedin and Lundqvist.

Bests

SaX developer Team

Quoting xiaoyinxie <xiaoyinxie at 126.com>:

> Dear developers:
>  I am sorry to misunderstood some basic concepts before you suggest 
> me to read some related papers.
>  However I still have some confusions.
> 1.How to calculate(By how to set parameters in bse.in or gw.in) 
> transition energy of arbitrary two levels.(assume there is an 
> electron at level 15 ,then it is excited onto level 18).
>
>
>
>
> --
>
> ???????,???????!
>
>
> ?2009-05-06,marsamos at democritos.it ??:
>> As attachment you can find the routine pw_pp_parameters_extern.f90. Put
>> this routine in the directory of sax sources (which should be
>> sax_1.0.0/sax/src, or sax-1.0.0/sax/src). This will replace the old
>> routine with this one. Then just type within the source directory: make
>> all. For the calculations just run as you always do. Please, once you
>> got the GW results can you just tell me if is in agreement with
>> experiments? thank you in advance.
>>
>> In the BSE there is no anymore the concept of HOMO-LUMO as the BSE give
>> the TRANSITION energy between levels but including the electron-hole
>> interaction. The solution of BSE gives excitation energies to form an
>> electron-hole pair. I put a tutorial (you can download it) on all this
>> stuff in the Docs menu of www.qe-forge.org/projects/sax-project.
>>
>> Bests
>>
>> Layla
>>
>> PS In oas.RPAEXC you have the imaginary part of the macroscopic
>> dielectric constant as a function of energy, or in other words, you
>> have the absorption spectrum. The peaks are the signature of optically
>> active excitons or excitonic resonnances. In x you have energies in Ry
>> in y you have the absorption in arbitrary units. Read, for instance, E.
>> Chang and S. Louie PRL ..do not remember the numbers but is very easy
>> to find in the aps web site.
>>
>> Quoting xiaoyinxie <xiaoyinxie at 126.com>:
>>
>>> Dear marsamos:
>>>  Thank you for your reply and kindly help.
>>>  please mail me the sax routine  pw_pp_parameters_extern.f90. which
>>> was mentioned in your previous email .and the procedure of compiling
>>> it.
>>>  PS : where can I find the eigenvilue of the energy level
>>> distributions(such as HOMO and LUMO) in the BSE (output of bse.x
>>> calculation ) or somewhere else ?
>>> Thank you !
>>>             Xie Xiaoyin
>>>                   2009/5/7
>>>
>>>
>>>
>>>
>>> --
>>>
>>> ???????,???????!
>>>
>>>
>>> ?2009-05-04,marsamos at democritos.it ??:
>>>> Dear User, to perform an EXX calculation with espresso you need first
>>>> to download the cvs version (go to www.qe-forge.org, select quantum
>>>> ESPRESSO and there select SCM and follow instruction for anonymous
>>>> connection). Then as always execute ./configure then add the -DEXX flag
>>>> to the DFLAGS line of make.sys:
>>>>
>>>> DFLAGS      = ...... -DEXX
>>>>
>>>> then you have to compile espresso and the post-processing programs
>>>> (pw_export.x). As attachment you can find an example of run with EXX
>>>> and then GW for Polyacetylene (30 k points).
>>>>
>>>> I have perform some calculations on Polyacetylene, polyethylene and PPV
>>>> with the new way of implementing Gigi-Baldereschi within the plasmon
>>>> pole and it works very well (the agreement with experiment is
>>>> impressive). If you want, instead of waiting for the release, I can
>>>> send you the new version of the sax routine
>>>> pw_pp_parameters_extern.f90. With this new version you will be able to
>>>> perform GW calculation with the plasmon pole on your polymers!!!
>>>>
>>>> best regards
>>>>
>>>> Layla
>>>>
>>>> Quoting xiaoyinxie <xiaoyinxie at 126.com>:
>>>>
>>>>> Dear marsamos:
>>>>> thank you for your kind help!
>>>>> your previous reply surgested me to implement a BSE calculation based
>>>>> on a Exact-exchange and I need to modify somewhere of pwscf.
>>>>> could you tell me which part I nend to modify in order to perform a
>>>>> EXX ,and what is the exact computational procedure?
>>>>> best wishes
>>>>>  Xie Xiaoyin
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> --
>>>>>
>>>>> ???????,???????!
>>>>>
>>>>>
>>>>> ?2009-04-20,marsamos at democritos.it ??:
>>>>>> Dear user, we where already aware of such 1D problems! I perform some
>>>>>> calculations on PPV, polyacetylene and polyethylene and found that for
>>>>>> PPV the agreement with experiment is good instead PA and PE seems to
>>>>>> behave abnormaly, in the way you already noticed. If you try a
>>>>>> calculation with cohsex option instead of plasmonpole you will see that
>>>>>> the results are completely different. We suspect that the problem comes
>>>>>> from the way plasmon-pole is interpolating. In fact in strongly
>>>>>> anisotropic systems the traditional way of doing plasmon-pole is not
>>>>>> consistent. We are now working in a modification of the plasmon pole. I
>>>>>> hope that for august (beginning of september) we will release a new sax
>>>>>> version. But if you need the optical properties, i.e. to solve Bethe
>>>>>> Salpeter equation and you prefer to have the results before I suggest
>>>>>> to you to do the following:
>>>>>>
>>>>>> Perform a DFT calculation with exact-exchange (this really give good
>>>>>> results for the electronic structure, I test this for PA and PE). Then
>>>>>> calculate with sax the screening (Polarizability and W), and then
>>>>>> perform the Bethe-Salpeter calculation. This procedure is theoretically
>>>>>> correct with the lesser approximation. As EXX is still experimental
>>>>>> within espresso you will have to modify some things within the code (is
>>>>>> very easy). Let me now and I give you the exact computational procedure.
>>>>>>
>>>>>>
>>>>>> Best regards
>>>>>>
>>>>>> Layla
>>>>>>
>>>>>>
>>>>>> of this We are now studying polyacetylene  Quoting xiaoyinxie
>>>>>> <xiaoyinxie at 126.com>:
>>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>> ----------------------------------------------------------------
>>>> This message was sent using IMP, the Internet Messaging Program.
>>>>
>>>
>>
>>
>>
>> ----------------------------------------------------------------
>> This message was sent using IMP, the Internet Messaging Program.
>>
>



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.