메뉴 건너뛰기
Physical Review Letters
Volumn 69, Issue 24, 1992, Pages 3547-3550
Large scale electronic structure calculations
Author keywords

[No Author keywords available]
Indexed keywords

EID: 3843075351     PISSN: 00319007     EISSN: None     Source Type: Journal    
DOI: 10.1103/PhysRevLett.69.3547     Document Type: Article
Times cited : (332)
References (17)
  • 1
    • 84927427022 scopus 로고    scopus 로고
    • ]{slow} Present address.
  • 4
    • 84927427021 scopus 로고    scopus 로고
    • See, e.g., G. Galli and M. Parrinello, in Computer Simulation in Material Science, edited by M. Meyer and V. Pontikis (Kluwer, Dordrecht, 1991).
  • 9
    • 84927427020 scopus 로고    scopus 로고
    • In practice the theta function is smoothed for numerical convenience.
  • 10
    • 84927427019 scopus 로고    scopus 로고
    • We have used a PW cutoff of 12 Ry, separable nonlocal pseudopotentials, and a spherical LR of radius σ = 3.5 a.u. A was varied between 10-2 and 10-3 a.u. during the iterative procedure.
  • 11
    • 84927388173 scopus 로고
    • Another possible choice, not yet explored, would be to optimize also the positions of the localization centers, following a method similar to that introduced by M. Sprik and M.L. Klein [, ] in the context of floating Gaussian calculations. par
    • (1989) J. Chem. Phys. , vol.89 , pp. 1592
  • 12
    • 84927427018 scopus 로고    scopus 로고
    • See, e.g., Numerical Recipes (Cambridge Univ. Press, Cambridge, 1987), pp. 306 307. The number of iterations needed for the solution of Eq. (8) can be made as small as 1 to 3 if overline{φi} is stored at each step.
  • 13
    • 84927427017 scopus 로고    scopus 로고
    • The methods for inverting banded matrices are so efficient that for realistic sizes one can afford the evaluation of all Sij-1: This amounts to a tiny fraction of the global CPU time, and the practical scaling of the algorithm is still O(N).

* 이 정보는 Elsevier사의 SCOPUS DB에서 KISTI가 분석하여 추출한 것입니다.