Abstract
We present a projection based multiscale optimization method for eigenvalue problems. In multiscale optimization, optimization steps using approximations at a coarse scale alternate with corrections by occasional calculations at a finer scale. We study an example in the context of electronic structure optimization. Theoretical analysis and numerical experiments provide estimates of the expected efficiency and guidelines for parameter selection.
Original language | American English |
---|---|
Pages (from-to) | 235-245 |
Number of pages | 11 |
Journal | Journal of Global Optimization |
Volume | 39 |
Issue number | 2 |
DOIs | |
State | Published - 2007 |
NREL Publication Number
- NREL/JA-530-39289
Keywords
- Approximation optimization
- Galerkin projection
- Multiscale optimization
- Reduced-order modeling