Efficient solutions to the NDA-NCA low-order eigenvalue problem

Conference · Mon Jul 01 00:00:00 EDT 2013

OSTI ID:22212911

Willert, J. A. ^[1]; Kelley, C. T. ^[2]

---

1. North Carolina State University, Raleigh, NC 27606 (United States)
2. Department of Mathematics, North Carolina State University, Raleigh, NC 27606 (United States)

Recent algorithmic advances combine moment-based acceleration and Jacobian-Free Newton-Krylov (JFNK) methods to accelerate the computation of the dominant eigenvalue in a k-eigenvalue calculation. In particular, NDA-NCA [1], builds a sequence of low-order (LO) diffusion-based eigenvalue problems in which the solution converges to the true eigenvalue solution. Within NDA-NCA, the solution to the LO k-eigenvalue problem is computed by solving a system of nonlinear equation using some variant of Newton's method. We show that we can speed up the solution to the LO problem dramatically by abandoning the JFNK method and exploiting the structure of the Jacobian matrix. (authors)

• Research Organization: American Nuclear Society, 555 North Kensington Avenue, La Grange Park, IL 60526 (United States)
• OSTI ID: 22212911
• Resource Relation: Conference: M and C 2013: 2013 International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering, Sun Valley, ID (United States), 5-9 May 2013; Other Information: Country of input: France; 8 refs.; Related Information: In: Proceedings of the 2013 International Conference on Mathematics and Computational Methods Applied to Nuclear Science and Engineering - M and C 2013| 3016 p.
• Country of Publication: United States
• Language: English

Related Subjects

97 MATHEMATICAL METHODS AND COMPUTING
ACCELERATION
EIGENFUNCTIONS
EIGENVALUES
EQUATIONS
MATHEMATICAL SOLUTIONS
MATRICES
NEWTON METHOD
NONLINEAR PROBLEMS
VELOCITY