A new polynomial-time algorithm for linear programming
Combinatorica
An experimental procedure for simulation response surface model identification
Communications of the ACM
Journal of Computer and System Sciences - 26th IEEE Conference on Foundations of Computer Science, October 21-23, 1985
Design of frequency-domain experiments for discrete-valued factors
Applied Mathematics and Computation
Driving frequency selection for frequency domain simulation experiments
Operations Research
Simulation factor screening using harmonic analysis
Management Science
Computers and Intractability: A Guide to the Theory of NP-Completeness
Computers and Intractability: A Guide to the Theory of NP-Completeness
The Analysis of Local Search Problems and Their Heuristics
STACS '90 Proceedings of the 7th Annual Symposium on Theoretical Aspects of Computer Science
| Hi-index | 0.00 |
Frequency domain methodology has been applied to discrete-event simulations to identify terms in a polynomial metamodel of the simulation output. This paper studies the problems associated with optimally selecting input frequencies to drive the frequency domain method. This Frequency Selection Problem (FSP) can be formulated as a large mixed integer linear program (MILP) or as a large set of small linear programs (LPs). The number of such LPs with nonzero optimal values is shown to be exponential in the number of input parameters to the model. Each such LP solution corresponds to a local optimum of the MILP formulation. The LP formulation is also used to show that this local search version of FSP is polynomially solvable. A new NP-complete problem, the Indicator Function Spacing Problem (IFSP), is presented. This problem is then used to show that FSP is NP-easy, hence no harder than NP-complete problems. These results suggest that FSP is difficult, and that researchers addressing this problem should focus their attention on the construction of heuristic procedures rather than polynomial time algorithms.