Re: Confusion about the index in HS.fileout Hamiltonian ( No.1 ) |
- Date: 2015/04/05 11:56
- Name: kzhoulatte
- What still confuse me is the way the FNAN SNAN is determined. In my calculations, my FNAN is 103 138 103, but my SNAN is always 0. From the manual and codes, I did not get many ideas about the truncation mechanism.
Could anyone help explain more about this? Thanks.
|
Re: Confusion about the index in HS.fileout Hamiltonian ( No.2 ) |
- Date: 2015/04/06 12:33
- Name: T. Ozaki
- Hi,
Only the non-zero matrix elements are stored, since the strictly localized basis functions are used in OpenMX. The explanation on the indexing can be found in analysis_example.c as
/********************************************************************** Example 3: Print overlap matrix OLP[ct_AN][h_AN][i][j] ct_AN: global index of atoms h_AN local index of neighbouring atoms for the atom ct_AN i: orbital index in the atom ct_AN j: orbital index in the atom h_AN NOTE: For instance, if the basis specification of the atom ct_AN is s2p2, then the obital index runs in order of s, s', px, py, pz, px', py', pz' Transformation of the local index h_AN to the grobal index Gh_AN is made as Gh_AN = natn[ct_AN][h_AN]; Also, the cell index is given by Rn = ncn[ct_AN][h_AN]; Each component l, m, or n (Rn = l*a + m*b + n*c) are given by l = atv_ijk[Rn][1]; m = atv_ijk[Rn][2]; n = atv_ijk[Rn][3]; ***********************************************************************/
If you still have questions, please let us know.
Regards,
TO
![メンテ](./img/mente.gif) |
Re: Confusion about the index in HS.fileout Hamiltonian ( No.3 ) |
- Date: 2015/04/07 06:18
- Name: kzhoulatte
- Hi, T. Ozaki,
Thanks for your response. I saw the explanations in the code as above. I see how the orbitals and how the index organizes in codes. But what confuses me is the way to determine to FNAN and SNAN. Is that meaning the calculation in openMX is nearest neighboring tight binding method? So the way the blocks of Hamiltonian are arranged confuses me. Basically, how you determine FNAN and SNAN? And I was assuming for the Hamiltonian I should get some hopping energy between principle layers, I do not know why I did not see that?
Thanks for any help.
|
Re: Confusion about the index in HS.fileout Hamiltonian ( No.4 ) |
- Date: 2015/04/07 06:29
- Name: kzhoulatte
- I mean: is that meaning the calculation in openMX of mine is only first nearest neighboring tight binding method? Since my calculation gave me:
myjob.5063215.out:<truncation> CpyCell= 1 ct_AN= 1 FNAN SNAN 26 0 myjob.5063215.out:<truncation> CpyCell= 1 ct_AN= 1 FNAN SNAN 26 0 myjob.5063215.out:<truncation> CpyCell= 1 ct_AN= 2 FNAN SNAN 26 0 myjob.5063215.out:<truncation> myid= 0 CpyCell= 1 ct_AN= 2 FNAN SNAN 26 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 1 ct_AN= 1 FNAN SNAN 26 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 1 ct_AN= 3 FNAN SNAN 26 0 myjob.5063215.out:TFNAN= 78 Average FNAN= 26.00000 myjob.5063215.out:<truncation> CpyCell= 2 ct_AN= 1 FNAN SNAN 72 0 myjob.5063215.out:<truncation> CpyCell= 2 ct_AN= 2 FNAN SNAN 69 0 myjob.5063215.out:<truncation> CpyCell= 2 ct_AN= 1 FNAN SNAN 69 0 myjob.5063215.out:<truncation> myid= 0 CpyCell= 2 ct_AN= 2 FNAN SNAN 69 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 2 ct_AN= 1 FNAN SNAN 72 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 2 ct_AN= 3 FNAN SNAN 69 0 myjob.5063215.out:TFNAN= 210 Average FNAN= 70.00000 myjob.5063215.out:<truncation> CpyCell= 3 ct_AN= 1 FNAN SNAN 99 0 myjob.5063215.out:<truncation> CpyCell= 3 ct_AN= 1 FNAN SNAN 118 0 myjob.5063215.out:<truncation> CpyCell= 3 ct_AN= 2 FNAN SNAN 99 0 myjob.5063215.out:<truncation> myid= 0 CpyCell= 3 ct_AN= 2 FNAN SNAN 99 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 3 ct_AN= 1 FNAN SNAN 118 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 3 ct_AN= 3 FNAN SNAN 99 0 myjob.5063215.out:TFNAN= 316 Average FNAN= 105.33333 myjob.5063215.out:<truncation> CpyCell= 4 ct_AN= 1 FNAN SNAN 103 0 myjob.5063215.out:<truncation> CpyCell= 4 ct_AN= 1 FNAN SNAN 138 0 myjob.5063215.out:<truncation> CpyCell= 4 ct_AN= 2 FNAN SNAN 103 0 myjob.5063215.out:<truncation> myid= 0 CpyCell= 4 ct_AN= 2 FNAN SNAN 103 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 4 ct_AN= 1 FNAN SNAN 138 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 4 ct_AN= 3 FNAN SNAN 103 0 myjob.5063215.out:TFNAN= 344 Average FNAN= 114.66667 myjob.5063215.out:<truncation> CpyCell= 5 ct_AN= 1 FNAN SNAN 103 0 myjob.5063215.out:<truncation> CpyCell= 5 ct_AN= 1 FNAN SNAN 138 0 myjob.5063215.out:<truncation> CpyCell= 5 ct_AN= 2 FNAN SNAN 103 0 myjob.5063215.out:<truncation> myid= 0 CpyCell= 5 ct_AN= 2 FNAN SNAN 103 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 5 ct_AN= 1 FNAN SNAN 138 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 5 ct_AN= 3 FNAN SNAN 103 0 myjob.5063215.out:TFNAN= 344 Average FNAN= 114.66667 myjob.5063215.out:<truncation> CpyCell= 5 ct_AN= 1 FNAN SNAN 103 0 myjob.5063215.out:<truncation> CpyCell= 5 ct_AN= 1 FNAN SNAN 138 0 myjob.5063215.out:<truncation> CpyCell= 5 ct_AN= 2 FNAN SNAN 103 0 myjob.5063215.out:<truncation> myid= 0 CpyCell= 5 ct_AN= 2 FNAN SNAN 103 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 5 ct_AN= 1 FNAN SNAN 138 0 myjob.5063215.out:<truncation> myid= 1 CpyCell= 5 ct_AN= 3 FNAN SNAN 103 0 myjob.5063215.out:TFNAN= 344 Average FNAN= 114.66667
But the bands still seem quite reasonable. Thanks.
![メンテ](./img/mente.gif) |
Re: Confusion about the index in HS.fileout Hamiltonian ( No.5 ) |
- Date: 2015/04/07 06:33
- Name: kzhoulatte
- Moreover, why the FNAN for different layers should be different? Because I believe they have the same atom distance between neighbors.
Thanks.
|
Re: Confusion about the index in HS.fileout Hamiltonian ( No.6 ) |
- Date: 2015/04/07 10:16
- Name: T. Ozaki
- Hi,
As you may know, OpenMX uses strictly localized basis functions with cutoff radii 'rc' as basis set, and the cutoff radius of each atomic species can be different from each other.
If the cutoff radii for basis functions are rcA and rcB for atoms A and B, respectively, the Kohn-Sham and overlap matrix elements between atoms A and B, whose distance is less than (rcA+rcB), only survive. Otherwise, the matrix elements become strictly zero.
FNAN is the number of atoms B, where distance between atoms A and B is less than (rcA+rcB), resulting in non-zero matrix elements.
SNAN is used to construct a truncated cluster used for O(N) methods, and ignored in the conventional calculations.
If you still have questions, please let us know.
Regards,
TO
|
Re: Confusion about the index in HS.fileout Hamiltonian ( No.7 ) |
- Date: 2015/04/08 05:26
- Name: kzhoulatte
- Hi, T. Ozaki,
Thanks for your answer. I was thinking about orderN.HoppingRanges before. Your response, especially "SNAN is used to construct a truncated cluster used for O(N) methods, and ignored in the conventional calculations." makes things much clear to me now.
Thanks.
|