Abstract
We present a new method for constructing nearly orthogonal Latin hypercubes that greatly expands their availability to experimenters. Latin hypercube designs have proven useful for exploring complex, high-dimensional computational models, but can be plagued with unacceptable correlations among input variables. To improve upon their effectiveness, many researchers have developed algorithms that generate orthogonal and nearly orthogonal Latin hypercubes. Unfortunately, these methodologies can have strict limitations on the feasible number of experimental runs and variables. To overcome these restrictions, we develop a mixed integer programming algorithm that generates Latin hypercubes with little or no correlation among their columns for most any determinate run-variable combination—including fully saturated designs. Moreover, many designs can be constructed for a specified number of runs and factors—thereby providing experimenters with a choice of several designs. In addition, our algorithm can be used to quickly adapt to changing experimental conditions by augmenting existing designs by adding new variables or generating new designs to accommodate a change in runs.
- Aarts, E. and Lenstra, J.K. 1997. Local Search in Combinatorial Optimization. Wiley. Google ScholarDigital Library
- Ang, J. K. 2006. Extending orthogonal and nearly orthogonal Latin hypercube designs for computer simulation and experimentation. Master's thesis. Naval Postgraduate School, Monterey, CA.Google Scholar
- Barton, R. R. 1998. Simulation metamodels In Proceedings of the Winter Simulation Conference. D. J. Medeiros, E. F. Watson, J. S. Carson, and M. S. Manivannan, Eds., vol 1, IEEE, 167--174. Google ScholarDigital Library
- Bazaraa, M. S., Jarvis, J. J., and Sherali, H. D. 2004. Linear Programming and Network Flows 4th Ed. Wiley. Google ScholarDigital Library
- Bertsimas, D. and Tsitsiklis, J. N. 1997. Introduction to Linear Optimization. Athena Scientific, Nashua, NH. Google ScholarDigital Library
- Body, H. and Marston, C. 2007. The peace support operations model: origins, development, philosophy and use. J. Defense Model. Simul. Appl. Methodol. Technol. 8, 2, 69--77.Google Scholar
- Buyske, S. and Trout, R. 2001. Advanced Design of Experiments. Statistics 591 Lecture Series, Rutgers University.Google Scholar
- Cioppa, T. M. 2002. Efficient nearly orthogonal and space-filling experimental designs for high-dimensional complex models. Doctoral dissertation. Monterey, CA: Naval Postgraduate School.Google Scholar
- Cioppa, T. M. and Lucas, T. W. 2007, March. Efficient nearly orthogonal and space-filling Latin hypercubes. Technometrics 49, 1, 45--55.Google ScholarCross Ref
- Fang, K. T. 2011. The uniform design. http://www.ath.hkbu.edu.hk/UniformDesign/.Google Scholar
- Fang, K. T., Lin, D. K. J., Winker, P., and Zhang, Y. 2000a. Uniform design: Theory and application. Technometrics 42, 3, 237--248.Google ScholarCross Ref
- Fang, K. T., Ma, C. X., and Winker, P. 2000b. Centered L2-discrepancy of random sampling and Latin hypercube design, and construction of uniform designs. Math. Comput. 71, 237, 275--296. Google ScholarDigital Library
- Fang, K. T., Ma, C. X., and Winker, P. 2002. Centered L2-discrepency of random sampling and Latin hypercube design. Math. Comput. 71, 275--296. Google ScholarDigital Library
- Florian, A. 1992. An efficient sampling scheme: Updated Latin hypercube sampling. Probabil. Engin. Mech. 7, 123--130.Google ScholarCross Ref
- Hernandez, A. S. 2008. Breaking barriers to design dimensions in nearly orthogonal Latin hypercubes. Doctoral dissertation. Naval Postgraduate School, Monterey, CA.Google Scholar
- Hickernell, F. J. 1998. A generalized discrepancy and quadrature error bound. Math. Comput. 67, 299--322. Google ScholarDigital Library
- Hillier, F. S. and Lieberman, G. J. 2005. Introduction to Operations Research. McGraw-Hill. Google ScholarDigital Library
- Iman, R. L. and Conover, W. J. 1982. A distribution-free approach to inducing rank correlation among input variables. Commun. Statist. Simul. Comput. 11, 3, 311--334.Google ScholarCross Ref
- IBM. 2012. IBM ILOG Cplex Optimizer. http://www-01.ibm.com/software/integration/optimization/cplex-optimizer/.Google Scholar
- Joseph, V. R. and Hung, Y. 2008. Orthogonal-maximin Latin hypercube designs. Statistica Sinica 18, 171--186.Google Scholar
- Kim, L. and Loh, H. 2003. Classification trees and bivariate linear discriminant node models. J. Graphical Statist. 12, 512--530.Google ScholarCross Ref
- Kleijnen, J. P. C., Sanchez, S. M., Lucas, T. W., and Cioppa, T. M. 2005. A user's guide to the brave new world of designing simulation experiments. INFORMS J. Comput. 17, 3, 263--289. Google ScholarDigital Library
- Koehler, J. R. and Owen, A. B. 1996. Computer experiments. Handbook Statist. 13, 261--308.Google ScholarCross Ref
- L'E cuyer, P. 2009. Quasi-Monte Carlo methods with applications in finance. Finance Stochastics 13, 3, 307--349.Google ScholarCross Ref
- Marlin, B. 2009. Ascertaining validity in the abstract realm of PMESII simulation models: An analysis of the peace support operations model PSOM. Masters thesis. Naval Postgraduate School, Monterey, CA.Google Scholar
- McKay, M. D., Beckman, R. J., and Conover, W. J. 1979. A comparison of three methods for selecting values of input variables in the analysis of output from a computer code. Technometrics 21, 2, 239--245.Google Scholar
- Moon, H., Dean, A., and Santner, T. J. 2011. Algorithms for generating maximin Latin hypercube and orthogonal designs. J. Statist. Theory Pract. 5, 1, 81--98.Google ScholarCross Ref
- Morris, M. D. and Mitchell, T. J. 1995. Exploratory designs for computer experiments. J. Statist. Planning Inference 43, 381--402.Google ScholarCross Ref
- Montgomery, D. C. 2005. Design and Analysis of Experiments. Wiley. Google ScholarDigital Library
- Owen, A. B. 1994. Controlling correlations in Latin hypercube samples. J. Amer. Statist. Assoc. Theory Methods 89, 428, 1517--1522.Google Scholar
- Owen, A. B. 1998. Latin supercube sampling for very high-dimensional simulations. ACM Trans. Model. Comput. Simul. 8, 1, 71--102. Google ScholarDigital Library
- Pang, F., Liu, M. Q., and Lin, D. K. J. 2009. A construction method for orthogonal Latin hypercube designs with prime power levels. Statistica Sinica 19, 1721--1728.Google Scholar
- Patterson, H. D. 1954. The errors of lattice sampling. J. Royal Statist. Soc. Series B Methodol. 16, 1, 140--149.Google Scholar
- Rardin, R. L. 1998. Optimization in Operations Research. Prentice Hall, Inc.Google Scholar
- Sanchez, S. M. 2012. NOLH designs spreadsheet and code. http://harvest.nps.edu.Google Scholar
- Sanchez, S. M., Lucas, T. W., Sanchez, P. J., Nannini, C. J., and Wan, H. 2012. Designs for large-scale simulation experiments, with application to defense and homeland security. The Design and Analysis of Computer Experiments. Vol. 3: Special Designs and Applications, K. Hinkelmann, Ed., Wiley, 413--441.Google Scholar
- Sanchez, S. M. and Sanchez, P. J. 2005. Very large fractional factorial and central composite designs. ACM Trans. Model. Comput. Simul. 15, 4, 362--377. Google ScholarDigital Library
- Santner, T. J., Williams, B. J., and Notz, W. 2003. The Design and Analysis of Computer Experiments. Springer.Google Scholar
- Stinstra, E., Stehouwer, P., Den Hertog, D., and Vestjens, A. 2003. Constrained maximin designs for computer experiments. Technometrics 45, 4, 340--346.Google ScholarCross Ref
- Steinberg, D. M. and Lin, D. K. J. 2006. A construction method for orthogonal Latin hypercube designs. Biometrika 93, 2, 279--288.Google ScholarCross Ref
- Sun, F., Liu, M. Q., and Lin, D. K. J. 2009. Construction of orthogonal Latin hypercube designs. Biometrika 96, 971--974.Google ScholarCross Ref
- Tang, B. 1998. Selecting Latin hypercubes using correlation criteria. Statisca Sinica 8, 965--977.Google Scholar
- Wolsey, L. A. 1998. Integer Programming. Wiley.Google Scholar
- Ye, K. Q. 1998. Orthogonal column Latin hypercubes and their application in computer experiments. J. Amer. Statist. Assoc. Theory Methods 93, 444, 1430--1439.Google Scholar
Index Terms
- Constructing nearly orthogonal latin hypercubes for any nonsaturated run-variable combination
Recommendations
A Census of Small Latin Hypercubes
We count all latin cubes of order $n\le6$ and latin hypercubes of order $n\le5$ and dimension $d\le5$. We classify these (hyper)cubes into isotopy classes and paratopy classes (main classes). For the same values of $n$ and $d$ we classify all $d$-ary ...
Constructions of transitive latin hypercubes
A function f : { 0 , , q - 1 } n { 0 , , q - 1 } invertible in each argument is called a latin hypercube. A collection ( π 0 , π 1 , , π n ) of permutations of { 0 , , q - 1 } is called an autotopism of a latin hypercube f if π 0 f ( x 1 , , x n ) = f ( ...
The first infinite family of orthogonal Steiner systems S(3,5,v)
AbstractThe research on orthogonal Steiner systems S( t , k , v ) was initiated in 1968. For ( t , k ) ∈ { ( 2 , 3 ) , ( 3 , 4 ) }, this corresponds to orthogonal Steiner triple systems (STSs) and Steiner quadruple systems (SQSs), ...
Comments