| |
|
Item: REP-2006-109
|
| |
Item:REP-2006-109
|
Title:Dynamic supernodes in sparse Cholesky update/downdate and triangular solves
Timothy A. Davis
William W. Hager
Abstract:
The supernodal method for sparse Cholesky factorization represents the
factor L as a set of supernodes, each consisting of a contiguous set of
columns of L with identical nonzero pattern. A conventional supernode is
stored as a dense submatrix. While this is suitable for sparse Cholesky
factorization where the nonzero pattern of L does not change, it is not
suitable for methods that modify a sparse Cholesky factorization after a
low-rank change to A (an update/downdate, A = A +/- WW^T).
Supernodes merge and split apart during an update/downdate.
Dynamic supernodes are introduced, which allow a sparse Cholesky
update/downdate to obtain performance competitive with conventional
supernodal methods. A dynamic supernodal solver is shown to exceed the
performance of the conventional (BLAS-based) supernodal method for solving
triangular systems. These methods are incorporated into CHOLMOD, a sparse
Cholesky factorization and update/downdate package, which forms the basis of
x=A in MATLAB when A is sparse and symmetric positive
definite.
Supporting File:here
|
|
| |
|
Select Departmental Reports
|
| |
|
|
| |
|
Departmental Report Categories
|
| |
|
|
To submit items for this page, please log in on the
CISE submit page.
|
|