Distributed coloration neighborhood search
This talk will examine the performance of several distributed coloration neighborhood search algorithms for general graph coloring. Our main results are: (1) running local optimization within a rejection-free annealing framework can be considerably more efficient and effective than using the original annealing scheme, (2) maintaining a population of the best colorations seen so far and repeatedly attempting to improve them in parallel is more efficient and effective than concurrently following independent search threads, (3) random G{sub n, p} graphs with p {approximately} 0.5 seem to represent the hardest instances for these algorithms, and (4) the algorithms are competitive on dense random graphs, sparse random graphs and structured graphs. We will also discuss the effort required to tune the search parameters and will indicate how the process can be automated.
- OSTI ID:
- 36304
- Report Number(s):
- CONF-9408161--
- Country of Publication:
- United States
- Language:
- English
Similar Records
A parallel graph coloring heuristic
Quantum Local Search with the Quantum Alternating Operator Ansatz