[CP2K-user] dimension inconsistency of the overlap and MO coefficients
Xiaoming Wang
wxia... at gmail.com
Mon Oct 15 13:32:00 UTC 2018
Dear Vladimir,
Thanks for you reply.
Actually, I need to calculate the transition dipole moment <Psi_i| r
|Psi_j> between two states, which can be evaluated using the MO_CUBE files.
However, as suggested by Prof. Hutter, using the MO coefficient should be
more efficient. Since the MO coefficients are not normalized, the overlap
matrix is needed, so the dipole is then <C_i|S*r|C_j>. Unfortunately, I
cannot use the diagonalization method. I will have to use ROKS with OT in my
applications. For OT, the ADDED_MOS are not allowed. So I cannot find a way
around by now. Maybe I need turn back to MO_CUBE files.
Best,
Xiaoming
On Monday, October 15, 2018 at 4:38:32 AM UTC-4, Vladimir Rybkin wrote:
>
> Dear Xiaoming,
>
> I believe that for no application does one need to check CSC=1. If you do
> it for fun than you might as well try diagonalization once.
>
> Yours,
>
> Vladimir
>
> пятница, 12 октября 2018 г., 17:47:43 UTC+2 пользователь Xiaoming Wang
> написал:
>>
>> Hi Vladimir,
>>
>> Thanks for your suggestions.
>>
>> Best,
>> Xiaoming
>>
>> For my applications, however, I do need the OT method.
>> So I cannot use ADDED_MOS.
>>
>> On Friday, October 12, 2018 at 10:04:18 AM UTC-4, Vladimir Rybkin wrote:
>>>
>>> Dear Xiaoming,
>>>
>>> you need to do the following:
>>> 1) use DIAGONALIZATION ins SCF
>>> 2) in SCF use ADDED_MOS #very large number
>>>
>>> Yours,
>>>
>>> Vladimir
>>>
>>> пятница, 12 октября 2018 г., 14:53:59 UTC+2 пользователь Xiaoming Wang
>>> написал:
>>>>
>>>> Hi Vladimir,
>>>>
>>>> I'm sorry I made a mistake. I mean could the overlap matrix be
>>>> transformed to cartesian coordinates? Because I only have the MO of LUMO
>>>> printed in cartesian coordinates.
>>>>
>>>> Best,
>>>> Xiaoming
>>>>
>>>> On Friday, October 12, 2018 at 8:48:45 AM UTC-4, Vladimir Rybkin wrote:
>>>>>
>>>>> Hi Xiaoming,
>>>>>
>>>>> overlap matrix is already in spherical Gaussians. The overlap is taken
>>>>> between basis functions (AO), not between MO(s). It is calculated before
>>>>> any MOs are obtained.
>>>>>
>>>>> Yours,
>>>>>
>>>>> Vladimir
>>>>>
>>>>> пятница, 12 октября 2018 г., 14:34:43 UTC+2 пользователь Xiaoming Wang
>>>>> написал:
>>>>>>
>>>>>> Hi Vladimir,
>>>>>>
>>>>>> Thanks for your confirmation. Yes, with Cartesian off, the dimensions
>>>>>> are consistent now. If cp2k does not print MO of vacant state, then is it
>>>>>> possible to transform the overlap matrix to spherical coordinates?
>>>>>>
>>>>>> Best,
>>>>>> Xiaoming
>>>>>>
>>>>>> On Friday, October 12, 2018 at 8:18:30 AM UTC-4, Vladimir Rybkin
>>>>>> wrote:
>>>>>>>
>>>>>>> Dear Xiaoming,
>>>>>>>
>>>>>>> thank you for the data. I don't see any inconsistency now. You get
>>>>>>> 52 with Cartesian off and 54 with Cartesian on. Turning Cartesian off you
>>>>>>> can now multiply your C with your S. Isn't that what you want?
>>>>>>> Unfortunately, I can't comment on why orbital #5 is not printed. Perhaps,
>>>>>>> that is due to the fact that it is vacant.
>>>>>>>
>>>>>>> Yours,
>>>>>>>
>>>>>>> Vladimir
>>>>>>>
>>>>>>> пятница, 12 октября 2018 г., 13:47:35 UTC+2 пользователь Xiaoming
>>>>>>> Wang написал:
>>>>>>>>
>>>>>>>> Hi Vladimir,
>>>>>>>>
>>>>>>>> Thanks for you reply.
>>>>>>>>
>>>>>>>> My input of the print part is:
>>>>>>>>
>>>>>>>> ------------
>>>>>>>>
>>>>>>>> &MO
>>>>>>>>
>>>>>>>> EIGENVALUES
>>>>>>>>
>>>>>>>> EIGENVECTORS
>>>>>>>>
>>>>>>>> CARTESIAN
>>>>>>>>
>>>>>>>> FILENAME mos
>>>>>>>>
>>>>>>>> MO_INDEX_RANGE 4 5
>>>>>>>>
>>>>>>>> OCCUPATION_NUMBERS
>>>>>>>>
>>>>>>>> &EACH
>>>>>>>>
>>>>>>>> QS_SCF 0
>>>>>>>>
>>>>>>>> &END EACH
>>>>>>>>
>>>>>>>> &END MO
>>>>>>>>
>>>>>>>> --------------
>>>>>>>>
>>>>>>>>
>>>>>>>> With CARTESIAN on, I get
>>>>>>>>
>>>>>>>>
>>>>>>>> ----------------
>>>>>>>>
>>>>>>>> MO EIGENVALUES, MO OCCUPATION NUMBERS, AND CARTESIAN MO
>>>>>>>> EIGENVECTORS
>>>>>>>>
>>>>>>>>
>>>>>>>> 4 5
>>>>>>>>
>>>>>>>> -0.260963 -0.034173
>>>>>>>>
>>>>>>>>
>>>>>>>> 2.000000 0.000000
>>>>>>>>
>>>>>>>>
>>>>>>>> 1 1 O 2s 0.000000 0.236782
>>>>>>>>
>>>>>>>> 2 1 O 3s 0.000000 -0.090781
>>>>>>>>
>>>>>>>> 3 1 O 4s 0.000000 -0.240258
>>>>>>>>
>>>>>>>> 4 1 O 3px -0.917807 -0.000000
>>>>>>>>
>>>>>>>> 5 1 O 3py -0.000000 -0.000000
>>>>>>>>
>>>>>>>> 6 1 O 3pz 0.000000 -0.238095
>>>>>>>>
>>>>>>>> 7 1 O 4px 0.020934 -0.000000
>>>>>>>>
>>>>>>>> 8 1 O 4py 0.000000 -0.000000
>>>>>>>>
>>>>>>>> 9 1 O 4pz -0.000000 0.013849
>>>>>>>>
>>>>>>>> 10 1 O 5px -0.003042 0.000000
>>>>>>>>
>>>>>>>> 11 1 O 5py 0.000000 0.000000
>>>>>>>>
>>>>>>>> 12 1 O 5pz -0.000000 0.099980
>>>>>>>>
>>>>>>>> 13 1 O 3dx2 -0.000000 0.000099
>>>>>>>>
>>>>>>>> 14 1 O 3dxy -0.000000 -0.000000
>>>>>>>>
>>>>>>>> 15 1 O 3dxz -0.005256 -0.000000
>>>>>>>>
>>>>>>>> 16 1 O 3dy2 -0.000000 0.000744
>>>>>>>>
>>>>>>>> 17 1 O 3dyz 0.000000 -0.000000
>>>>>>>>
>>>>>>>> 18 1 O 3dz2 0.000000 -0.000844
>>>>>>>>
>>>>>>>> 19 1 O 4dx2 -0.000000 0.000357
>>>>>>>>
>>>>>>>> 20 1 O 4dxy 0.000000 -0.000000
>>>>>>>>
>>>>>>>> 21 1 O 4dxz -0.022359 0.000000
>>>>>>>>
>>>>>>>> 22 1 O 4dy2 0.000000 0.004105
>>>>>>>>
>>>>>>>> 23 1 O 4dyz -0.000000 0.000000
>>>>>>>>
>>>>>>>> 24 1 O 4dz2 0.000000 -0.004463
>>>>>>>>
>>>>>>>> 25 1 O 5s -0.000000 -0.011663
>>>>>>>>
>>>>>>>> 26 1 O 5px 0.024750 -0.000000
>>>>>>>>
>>>>>>>> 27 1 O 5py -0.000000 0.000000
>>>>>>>>
>>>>>>>> 28 1 O 5pz 0.000000 -0.077018
>>>>>>>>
>>>>>>>>
>>>>>>>> 29 2 H 1s 0.000000 -0.116793
>>>>>>>>
>>>>>>>> 30 2 H 2s 0.000000 -0.003374
>>>>>>>>
>>>>>>>> 31 2 H 3s -0.000000 0.197029
>>>>>>>>
>>>>>>>> 32 2 H 2px 0.011575 -0.000000
>>>>>>>>
>>>>>>>> 33 2 H 2py 0.000000 0.008716
>>>>>>>>
>>>>>>>> 34 2 H 2pz -0.000000 -0.004748
>>>>>>>>
>>>>>>>> 35 2 H 3px 0.028901 -0.000000
>>>>>>>>
>>>>>>>> 36 2 H 3py 0.000000 0.013894
>>>>>>>>
>>>>>>>> 37 2 H 3pz -0.000000 -0.010138
>>>>>>>>
>>>>>>>> 38 2 H 4s -0.000000 0.421270
>>>>>>>>
>>>>>>>> 39 2 H 4px 0.026994 0.000000
>>>>>>>>
>>>>>>>> 40 2 H 4py 0.000000 0.041403
>>>>>>>>
>>>>>>>> 41 2 H 4pz -0.000000 -0.044975
>>>>>>>>
>>>>>>>>
>>>>>>>> 42 3 H 1s 0.000000 -0.116793
>>>>>>>>
>>>>>>>> 43 3 H 2s -0.000000 -0.003374
>>>>>>>>
>>>>>>>> 44 3 H 3s 0.000000 0.197029
>>>>>>>>
>>>>>>>> 45 3 H 2px 0.011575 -0.000000
>>>>>>>>
>>>>>>>> 46 3 H 2py 0.000000 -0.008716
>>>>>>>>
>>>>>>>> 47 3 H 2pz -0.000000 -0.004748
>>>>>>>>
>>>>>>>> 48 3 H 3px 0.028901 -0.000000
>>>>>>>>
>>>>>>>> 49 3 H 3py -0.000000 -0.013894
>>>>>>>>
>>>>>>>> 50 3 H 3pz -0.000000 -0.010138
>>>>>>>>
>>>>>>>> 51 3 H 4s 0.000000 0.421270
>>>>>>>>
>>>>>>>> 52 3 H 4px 0.026994 0.000000
>>>>>>>>
>>>>>>>> 53 3 H 4py 0.000000 -0.041403
>>>>>>>>
>>>>>>>> 54 3 H 4pz 0.000000 -0.044975
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Fermi energy: -0.260963
>>>>>>>>
>>>>>>>>
>>>>>>>> HOMO-LUMO gap: 0.226790 = 6.17 eV
>>>>>>>> ---------------------------------------
>>>>>>>>
>>>>>>>> However, with CARTESIAN OFF, I get only the orbital #4:
>>>>>>>>
>>>>>>>> ----------------------------------------
>>>>>>>>
>>>>>>>> MO EIGENVALUES, MO OCCUPATION NUMBERS, AND SPHERICAL MO
>>>>>>>> EIGENVECTORS
>>>>>>>>
>>>>>>>>
>>>>>>>> 4
>>>>>>>>
>>>>>>>> -0.260963
>>>>>>>>
>>>>>>>>
>>>>>>>> 2.000000
>>>>>>>>
>>>>>>>>
>>>>>>>> 1 1 O 2s 0.000000
>>>>>>>>
>>>>>>>> 2 1 O 3s 0.000000
>>>>>>>>
>>>>>>>> 3 1 O 4s 0.000000
>>>>>>>>
>>>>>>>> 4 1 O 3py -0.000000
>>>>>>>>
>>>>>>>> 5 1 O 3pz 0.000000
>>>>>>>>
>>>>>>>> 6 1 O 3px -0.917807
>>>>>>>>
>>>>>>>> 7 1 O 4py 0.000000
>>>>>>>>
>>>>>>>> 8 1 O 4pz -0.000000
>>>>>>>>
>>>>>>>> 9 1 O 4px 0.020934
>>>>>>>>
>>>>>>>> 10 1 O 5py 0.000000
>>>>>>>>
>>>>>>>> 11 1 O 5pz -0.000000
>>>>>>>>
>>>>>>>> 12 1 O 5px -0.003042
>>>>>>>>
>>>>>>>> 13 1 O 3d-2 -0.000000
>>>>>>>>
>>>>>>>> 14 1 O 3d-1 0.000000
>>>>>>>>
>>>>>>>> 15 1 O 3d0 0.000000
>>>>>>>>
>>>>>>>> 16 1 O 3d+1 -0.005256
>>>>>>>>
>>>>>>>> 17 1 O 3d+2 -0.000000
>>>>>>>>
>>>>>>>> 18 1 O 4d-2 0.000000
>>>>>>>>
>>>>>>>> 19 1 O 4d-1 -0.000000
>>>>>>>>
>>>>>>>> 20 1 O 4d0 0.000000
>>>>>>>>
>>>>>>>> 21 1 O 4d+1 -0.022359
>>>>>>>>
>>>>>>>> 22 1 O 4d+2 -0.000000
>>>>>>>>
>>>>>>>> 23 1 O 5s -0.000000
>>>>>>>>
>>>>>>>> 24 1 O 5py -0.000000
>>>>>>>>
>>>>>>>> 25 1 O 5pz 0.000000
>>>>>>>>
>>>>>>>> 26 1 O 5px 0.024750
>>>>>>>>
>>>>>>>>
>>>>>>>> 27 2 H 1s 0.000000
>>>>>>>>
>>>>>>>> 28 2 H 2s 0.000000
>>>>>>>>
>>>>>>>> 29 2 H 3s -0.000000
>>>>>>>>
>>>>>>>> 30 2 H 2py 0.000000
>>>>>>>>
>>>>>>>> 31 2 H 2pz -0.000000
>>>>>>>>
>>>>>>>> 32 2 H 2px 0.011575
>>>>>>>>
>>>>>>>> 33 2 H 3py 0.000000
>>>>>>>>
>>>>>>>> 34 2 H 3pz -0.000000
>>>>>>>>
>>>>>>>> 35 2 H 3px 0.028901
>>>>>>>>
>>>>>>>> 36 2 H 4s -0.000000
>>>>>>>>
>>>>>>>> 37 2 H 4py 0.000000
>>>>>>>>
>>>>>>>> 38 2 H 4pz -0.000000
>>>>>>>>
>>>>>>>> 39 2 H 4px 0.026994
>>>>>>>>
>>>>>>>>
>>>>>>>> 40 3 H 1s 0.000000
>>>>>>>>
>>>>>>>> 41 3 H 2s -0.000000
>>>>>>>>
>>>>>>>> 42 3 H 3s 0.000000
>>>>>>>>
>>>>>>>> 43 3 H 2py 0.000000
>>>>>>>>
>>>>>>>> 44 3 H 2pz -0.000000
>>>>>>>>
>>>>>>>> 45 3 H 2px 0.011575
>>>>>>>>
>>>>>>>> 46 3 H 3py -0.000000
>>>>>>>>
>>>>>>>> 47 3 H 3pz -0.000000
>>>>>>>>
>>>>>>>> 48 3 H 3px 0.028901
>>>>>>>>
>>>>>>>> 49 3 H 4s 0.000000
>>>>>>>>
>>>>>>>> 50 3 H 4py 0.000000
>>>>>>>>
>>>>>>>> 51 3 H 4pz 0.000000
>>>>>>>>
>>>>>>>> 52 3 H 4px 0.026994
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Fermi energy: -0.260963
>>>>>>>>
>>>>>>>> -----------------------------
>>>>>>>>
>>>>>>>>
>>>>>>>> Why the MO #5 is not printed in spherical coordinate?
>>>>>>>>
>>>>>>>>
>>>>>>>> Best,
>>>>>>>>
>>>>>>>> Xiaoming
>>>>>>>>
>>>>>>>> On Friday, October 12, 2018 at 5:51:08 AM UTC-4, Vladimir Rybkin
>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>> Hi Xiaoming,
>>>>>>>>>
>>>>>>>>> it is. I mean that by default it is not printed, only the basis
>>>>>>>>> is. So, please attach your input/output: this will help to stay on equal
>>>>>>>>> footing.
>>>>>>>>>
>>>>>>>>> Yours,
>>>>>>>>>
>>>>>>>>> Vladimir
>>>>>>>>>
>>>>>>>>> четверг, 11 октября 2018 г., 16:26:57 UTC+2 пользователь Xiaoming
>>>>>>>>> Wang написал:
>>>>>>>>>>
>>>>>>>>>> Hi Vladimir,
>>>>>>>>>>
>>>>>>>>>> Are you saying the option of printing MOs with eigenvector on is
>>>>>>>>>> not printing the MO coefficients?
>>>>>>>>>>
>>>>>>>>>> Best,
>>>>>>>>>> Xiaoming
>>>>>>>>>>
>>>>>>>>>> On Thursday, October 11, 2018 at 9:36:49 AM UTC-4, Vladimir
>>>>>>>>>> Rybkin wrote:
>>>>>>>>>>>
>>>>>>>>>>> Dear Xiaoming,
>>>>>>>>>>>
>>>>>>>>>>> by default CP2K does not print MO coefficients. It prints the
>>>>>>>>>>> basis, which is Cartesian (e.i. 6 d orbitals instead of 5). So,
>>>>>>>>>>> input/output would be helpful.
>>>>>>>>>>>
>>>>>>>>>>> Yours,
>>>>>>>>>>>
>>>>>>>>>>> Vladimir
>>>>>>>>>>>
>>>>>>>>>>> четверг, 11 октября 2018 г., 15:24:16 UTC+2 пользователь
>>>>>>>>>>> Xiaoming Wang написал:
>>>>>>>>>>>>
>>>>>>>>>>>> Hi,
>>>>>>>>>>>>
>>>>>>>>>>>> I just missed one thing. The MOs could also be printed with
>>>>>>>>>>>> spherical coordinates. Then, it should be consistent.
>>>>>>>>>>>>
>>>>>>>>>>>> Best,
>>>>>>>>>>>> Xiaoming
>>>>>>>>>>>>
>>>>>>>>>>>> On Thursday, October 11, 2018 at 9:21:56 AM UTC-4, Xiaoming
>>>>>>>>>>>> Wang wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>> Hi Vladimir,
>>>>>>>>>>>>>
>>>>>>>>>>>>> Thanks for your reply. I'm wondering why the eigenvectors and
>>>>>>>>>>>>> overlap are not using the same basis (or with same basis but different
>>>>>>>>>>>>> coordinates).
>>>>>>>>>>>>>
>>>>>>>>>>>>> Best,
>>>>>>>>>>>>> Xiaoming
>>>>>>>>>>>>>
>>>>>>>>>>>>> On Thursday, October 11, 2018 at 4:28:32 AM UTC-4, Vladimir
>>>>>>>>>>>>> Rybkin wrote:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Dear Xiaoming,
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> the 6 d orbitals correspond to the six components in the
>>>>>>>>>>>>>> Cartesian coordinates, whereas 5 d orbitals are the same expressed is
>>>>>>>>>>>>>> spherical coordinates. The coordinate transformation is involved in
>>>>>>>>>>>>>> between.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Yours,
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Vladimir
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> среда, 10 октября 2018 г., 16:16:56 UTC+2 пользователь
>>>>>>>>>>>>>> Xiaoming Wang написал:
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Hello,
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> I'd like to check the equation C^T S C = 1, where C is MO
>>>>>>>>>>>>>>> coefficient and S the overlap matrix.
>>>>>>>>>>>>>>> However, for my test calculation, C is 54*1 while S is
>>>>>>>>>>>>>>> 52*52. I checked the basis details, for the
>>>>>>>>>>>>>>> MO coefficient, there are 6 d orbitals
>>>>>>>>>>>>>>> (dx2,dxy,dxz,dy2,dyz,dz2), while for the overlap matrix,
>>>>>>>>>>>>>>> only 5 d orbitals (d-2,d-1,d0,d+1,d+2) are printed. Why is
>>>>>>>>>>>>>>> like that? Am I missing something?
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Best,
>>>>>>>>>>>>>>> Xiaoming
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20181015/68cfb305/attachment.htm>
More information about the CP2K-user
mailing list