Exact parallel algorithms for some members of the traveling salesman problem family
The traveling salesman problem and its many generalizations comprise one of the best known combinatorial optimization problem families. Most members of the family are NP-complete problems so that exact algorithms require an unpredictable and sometimes large computational effort. Parallel computers offer hope for providing the power required to meet these demands. A major barrier to applying parallel computers is the lack of parallel algorithms. The contributions presented in this thesis center around new exact parallel algorithms for the asymmetric traveling salesman problem (ATSP), prize collecting traveling salesman problem (PCTSP), and resource constrained traveling salesman problem (RCTSP). The RCTSP is a particularly difficult member of the family since finding a feasible solution is an NP-complete problem. An exact sequential algorithm is also presented for the directed hamiltonian cycle problem (DHCP). The DHCP algorithm is superior to current heuristic approaches and represents the first exact method applicable to large graphs. Computational results presented for each of the algorithms demonstrates the effectiveness of combining efficient algorithms with parallel computing methods. Performance statistics are reported for randomly generated ATSPs with 7,500 cities, PCTSPs with 200 cities, RCTSPs with 200 cities, DHCPs with 3,500 vertices, and assignment problems of size 10,000. Sequential results were collected on a Sun 4/260 engineering workstation, while parallel results were collected using a 14 and 100 processor BBN Butterfly Plus computer. The computational results represent the largest instances ever solved to optimality on any type of computer.
- Research Organization:
- Carnegie-Mellon Univ., Pittsburgh, PA (United States)
- OSTI ID:
- 6043348
- Resource Relation:
- Other Information: Thesis (Ph.D)
- Country of Publication:
- United States
- Language:
- English
Similar Records
The cost-constrained traveling salesman problem
Comparative Study of Variations in Quantum Approximate Optimization Algorithms for the Traveling Salesman Problem
Related Subjects
PARALLEL PROCESSING
ALGORITHMS
COMPUTER CALCULATIONS
OPTIMIZATION
ROAD TRANSPORT
ROUTING
TASK SCHEDULING
DATA PROCESSING
LAND TRANSPORT
MATHEMATICAL LOGIC
PROCESSING
PROGRAMMING
TRANSPORT
990200* - Mathematics & Computers