On Using Approximate Finite Differences in Matrix-Free Newton–Krylov Methods
✍ Scribed by Brown, Peter N.; Walker, Homer F.; Wasyk, Rebecca; Woodward, Carol S.
- Book ID
- 118190083
- Publisher
- Society for Industrial and Applied Mathematics
- Year
- 2008
- Tongue
- English
- Weight
- 257 KB
- Volume
- 46
- Category
- Article
- ISSN
- 0036-1429
No coin nor oath required. For personal study only.
📜 SIMILAR VOLUMES
This paper presents a new look at Davidson's method for the calculation of the rightmost eigenvalue(s). The combination of time-stepping by the Krylov exponential propagator and the Davidson method leads to a method that builds a Krylov space of the matrix exponential. The method is well-suited when
It is shown how the Lagrange Multiplier method for constrained minimization can be implemented in a molecular mechanics program using the common approximations to the full-matrix Newton-Raphson minimization. The method reduces the number of cycles to achieve convergence, and also stabilizes the refi