Approximate MCSP (Matrix Chain Multiplication)
Revision as of 10:17, 15 February 2023 by Admin (talk | contribs) (Created page with "{{DISPLAYTITLE:Approximate MCSP (Matrix Chain Multiplication)}} == Description == The Matrix Chain Scheduling Problem (or MCSP) is an optimization problem� where the goal is to find the product sequence for evaluating a chain of matrix products and the processor schedule for the sequence such that the evaluation time is minimized on a parallel system. In the approximation problem, the matrix multiplication carried out with the output result will use a number of opera...")
Description
The Matrix Chain Scheduling Problem (or MCSP) is an optimization problem� where the goal is to find the product sequence for evaluating a chain of matrix products and the processor schedule for the sequence such that the evaluation time is minimized on a parallel system. In the approximation problem, the matrix multiplication carried out with the output result will use a number of operations that has some sort of upper bound based on the optimal solution.
Related Problems
Generalizations: Matrix Chain Scheduling Problem
Related: Matrix Chain Ordering Problem, Approximate MCOP
Parameters
$P$: number of processors $n$: number of matrices
Table of Algorithms
| Name | Year | Time | Space | Approximation Factor | Model | Reference |
|---|---|---|---|---|---|---|
| Heejo Lee; Jong Kim; Sungje Hong; and Sunggu Lee | 1997 | $O(n^{2})$ | $O(n^{2})$? | "near optimal" | Deterministic | Time |
References/Citation
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.56.222&rep=rep1&type=pdf