Solving Helmholtz boundary value problems with quasiseparable matrices

Brad Nelson

'13

Dartmouth


Quasiseparable matrices are a general class of matrices with low off-diagonal rank. These matrices are attractive for numerical computation, as storage scales linearly, allowing for large simulations, and algorithms for standard matrix operations are of linear complexity, allowing for quick manipulation. The use of quasiseparable matrices in the context of solving Helmholtz boundary value problems in one and two dimensions will be discussed, as well as problems with scaling to the three-dimensional case.

Back to ACMS schedule