Abstract
We present a new parallel algorithm for computing arbitrary functions of triangular matrices. The presented algorithm is the first one to date requiring polylogarithmic time, and computes an arbitrary function of an n×n triangular matrix in O(log3 n) time using O(n6) processors. The algorithm requires the eigenvalues of the input matrix be distinct, and makes use of the commutativity relationship between the input and output matrices.
Original language | English (US) |
---|---|
Pages (from-to) | 85-92 |
Number of pages | 8 |
Journal | Computing (Vienna/New York) |
Volume | 57 |
Issue number | 1 |
DOIs | |
State | Published - Jan 1 1996 |
Externally published | Yes |
Keywords
- Divide and conquer
- Matrix functions
- Parlett's algorithm
ASJC Scopus subject areas
- Software
- Theoretical Computer Science
- Numerical Analysis
- Computer Science Applications
- Computational Theory and Mathematics
- Computational Mathematics