hamiltonian matrix and overlap files

Top Page > Browsing

hamiltonian matrix and overlap files

Date: 2011/03/12 07:30
Name: xianghui: Dear developer,

I am wondering how the hamiltonian matrix elements are listed in the HS.out file. More specifically, I do not know the matrix element corresponds to which orbitals. Take the Pt as an example. I used the minimal basis, i.e., 5s,6s,5p,5d as valence electrons. As a result, I got a 10 x 10 matrix for each pair of atoms. What is unclear to me is the sequence of orbitals for each row and column. Are they like following?

5s 6s 5px 5py 5pz 5d3z2-r2 5dx2-y2 5dxy 5dzx 5dyz

6s

5px

5py

5pz

5d3z2-r2

5dx2-y2

5dxy

5dzx

5dyz

Page: [1]

Re: hamiltonian matrix and overlap files ( No.1 )

Date: 2011/03/12 08:01
Name: xianghui

Sorry, forgot thanks.

Any suggestion will be appreciated.

Best regards,

Xianghui

Re: hamiltonian matrix and overlap files ( No.2 )

Date: 2011/03/23 18:51
Name: T. Ozaki

Hi,

As written in 'analysis_example.c", your guess is exactly what OpenMX defines.

Best regards,

TO

Re: hamiltonian matrix and overlap files ( No.3 )

Date: 2011/03/25 20:12
Name: xianghui

Well, then it is somehow funning.
I used the minimal basis mentioned above for bulk Pt, a FCC structure with a lattice constant of 4.0 A. Version of PAO and VPS is DFT_DATA06. Hamiltonian matrix shows a very strange results such as following.

********************

glbal index=1 local index=19 (grobal=1, Rn=123)
0.0001076 -0.0273032 -0.0000000 0.0005491 0.0005491 0.0059115 -0.0102391 0.0000000 -0.0000000 0.0204782

-0.0273032 0.1744889 0.0000000 -0.0319570 -0.0319570 -0.0258023 0.0446909 0.0000000 0.0000000 -0.0893819

0.0000000 -0.0000000 0.0002835 0.0000000 0.0000000 0.0000000 -0.0000000 0.0066845 0.0066845 0.0000000

-0.0005491 0.0319570 0.0000000 -0.0013250 -0.0016084 -0.0127678 0.0087456 0.0000000 0.0000000 -0.0241756

-0.0005491 0.0319570 0.0000000 -0.0016084 -0.0013250 -0.0011900 0.0154301 0.0000000 0.0000000 -0.0241756

0.0059115 -0.0258023 -0.0000000 0.0127678 0.0011900 -0.0402591 -0.0401019 -0.0000000 -0.0000000 0.0160519

-0.0102391 0.0446909 0.0000000 -0.0087456 -0.0154301 -0.0401019 0.0060465 0.0000000 0.0000000 -0.0278027

-0.0000000 0.0000000 -0.0066845 -0.0000000 -0.0000000 -0.0000000 0.0000000 -0.0263739 -0.0370381 -0.0000000

-0.0000000 0.0000000 -0.0066845 -0.0000000 -0.0000000 -0.0000000 0.0000000 -0.0370381 -0.0263739 -0.0000000

0.0204782 -0.0893819 -0.0000000 0.0241756 0.0241756 0.0160519 -0.0278027 -0.0000000 -0.0000000 0.0662696

*********************

You can see that the overlap between 6s orbitals are positive i.e., 0.1744889, which should be negative since sssigma is negative by symmetry.
It is so strange that I do believe the sequence of orbitals that I guessed is correct.
How to understand that? Thanks a lot.
Best regards,

xianghui

Re: hamiltonian matrix and overlap files ( No.4 )

Date: 2011/04/06 22:55
Name: xianghui

Thanks for Prof. Taisuke Ozaki's private email. Now the problem is clear. The Hamiltonian matrix element between 5s and 6s orbitals can be positive for the out-most radial part of atomic orbital are positive and negative for 5s and 6s orbitals, respectively. So the resulting hopping parameters is positive by taking sign of Column interaction into account. Thanks a lot for Prof. Taisuke Ozaki's reply and remind me the nodes of atomic orbital.

Page: [1]