Conference Presentation (Invited)

FZJ-2015-02284

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Enabling Large Scale LAPW DFT Calculations by a Scalable Iterative Eigensolver

Di Napoli, E. (Corresponding Author)FZJ* ; Wortmann, D.FZJ* ; Berljafa, M.

2015

Please use a persistent id in citations: http://hdl.handle.net/2128/8471

Abstract: In LAPW-based methods a sequence of dense generalized eigenvalue problems appears. Traditionally these problems were solved using direct eigensolvers from standard libraries like ScaLAPACK. We developed a subspace iteration method pre-conditioned with Chebyshev polynomials of optimal degree (ChASE). This algorithm is consistently competitive with direct eigensolvers and greatly enhance performance and scalability. ChASE is included in the FLEUR software and improves its scaling behaviour for calculations of large physical systems on modern supercomputers.

---

Contributing Institute(s):

1. Jülich Supercomputing Center (JSC)

Appears in the scientific report 2015

---

Database coverage:

---