2014
DOI: 10.1103/physreve.90.013306
|View full text |Cite
|
Sign up to set email alerts
|

Real-space finite-difference calculation method of generalized Bloch wave functions and complex band structures with reduced computational cost

Abstract: Generalized Bloch wave functions of bulk structures, which are composed of not only propagating waves but also decaying and growing evanescent waves, are known to be essential for defining the open boundary conditions in the calculations of the electronic surface states and scattering wave functions of surface and junction structures. Electronic complex band structures being derived from the generalized Bloch wave functions are also essential for studying bound states of the surface and junction structures, wh… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
34
0

Year Published

2016
2016
2021
2021

Publication Types

Select...
7

Relationship

2
5

Authors

Journals

citations
Cited by 10 publications
(34 citation statements)
references
References 68 publications
0
34
0
Order By: Relevance
“…However, an additional computational cost is incurred when we use the BiCG method for solving Eq. (25). If we use the BiCG method as the solver, it is difficult to set a reasonable number of right-hand side N rh owing to a variation in the eigenvalue distribution inside the ring-shaped region.…”
Section: B Robustness Of the Algorithmmentioning
confidence: 99%
See 1 more Smart Citation
“…However, an additional computational cost is incurred when we use the BiCG method for solving Eq. (25). If we use the BiCG method as the solver, it is difficult to set a reasonable number of right-hand side N rh owing to a variation in the eigenvalue distribution inside the ring-shaped region.…”
Section: B Robustness Of the Algorithmmentioning
confidence: 99%
“…To demonstrate speed-ups, we compare the computational time of our method with that of the OBM method, which is categorized as a WFM method. Although continuous improvements [24][25][26]56 have been made after the first study of the OBM method, the computation of the first and last r columns of (E − H l,l ) −1 and the 2r-dimensional generalized eigenvalue problem is still required. In this study, the matrix inversion is calculated using the CG method, 57 and the generalized eigenvalue problem is solved by the optimized LAPACK routine ZGGEV and SS method.…”
Section: Serial Performancementioning
confidence: 99%
“…Hence, they are given as integer multiples of the number of the grid points in an xy plane (for more details, see Ref. [35]). The subscript L (R) denotes that a quantity with the subscript is associated with the left (right) boundary.…”
Section: A Conventional Proceduresmentioning
confidence: 99%
“…(8) in Ref. [35], one can see that the sets of generalized Bloch wave functions in the boundary regions, Q i,M(N) = {q i M(N),1 ,q i M(N),2 , . .…”
Section: Appendix A: Self-consistent Equationmentioning
confidence: 99%
“…Different solutions have been found, e.g. a pivoting-like bandwidth minimization method in combination with a block-tridiagonalization algorithm [Wim09], the reverse Cuthill-McKee algorithm for the block-tridiagonalization of connected graphs [Mas11,Cut69], the Knitting algorithm, which uses the Dyson equation and the resulting decimation scheme to build up the system site by site [Kaz08], or accelerations by using singular value decompositions [Tsu14]. Besides the RGF, also other methods are often utilized.…”
Section: Introductionmentioning
confidence: 99%