Title :
A sparse knapsack algo-tech-cuit and its synthesis
Author :
Andonov, Rumen ; Rajopadhye, Sanjay
Author_Institution :
IRISA, Rennes, France
Abstract :
We systematically derive an improved algorithm (called the sparse algorithm) for the general knapsack problem which has better average case performance than the standard (dense) dynamic programming algorithm. The derivation is based on transformation of the standard recurrences into stream functional programs, and cannot be achieved by the usual space-time mapping techniques because the dependencies are statically unpredictable. Furthermore such a sparse algorithm for the general knapsack problem has not been proposed in the literature, to the best of our knowledge. We also implement the sparse algorithm on a linear asynchronous array with constant size memory on each PE (i.e., a wavefront array processor). Using LPGS partitioning, the algorithm can run on an arbitrary size ring and has optimal time speedup
Keywords :
combinatorial mathematics; computational complexity; dynamic programming; operations research; optimisation; parallel algorithms; LPGS partitioning; NP-complete combinatorial optimization problem; PE; arbitrary size ring; average case performance; constant size memory; dynamic programming algorithm; general knapsack problem; linear asynchronous array; optimal time speedup; space-time mapping techniques; sparse algorithm; sparse knapsack algo-tech-cuit; standard recurrences; statically unpredictable; stream functional programs; wavefront array processor; Application specific processors; Boundary conditions; Computer science; Cost function; Difference equations; Dynamic programming; Heuristic algorithms; Partitioning algorithms; Polynomials;
Conference_Titel :
Application Specific Array Processors, 1994. Proceedings. International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-8186-6517-3
DOI :
10.1109/ASAP.1994.331794