ENERGY SCIENCE AND TECHNOLOGY SOFTWARE CENTER
U.S. Department of Energy
Find the latest in DOE-sponsored Scientific and Technical Software

# SOFTWARE ABSTRACT

PACKAGE ID 001728WKSTN00LHS UNIX 2004 Latin Hypercube Sampling (LHS) UNIX Library/Standalone Gregory Wyss, Sharon Daniel, Michael Eldrds, Shannon Brown, Laura Swiler COPY/LIM 05/13/2004 05/13/2004 The LHS UNIX Library/Standalone software provides the capability to draw random samples from over 30 distribution types. It performs the sampling by a stratified sampling method called Latin Hypercube Sampling (LHS). Multiple distributions can be sampled simultaneously, with user-specified correlations amongst the input distributions, LHS UNIX Library/ Standalone provides a way to generate multi-variate samples. The LHS samples can be generated either as a callable library (e.g., from within the DAKOTA software framework) or as a standalone capability. LHS UNIX Library/Standalone uses the Latin Hypercube Sampling method (LHS) to generate samples. LHS is a constrained Monte Carlo sampling scheme. In LHS, the range of each variable is divided into non-overlapping intervals on the basis of equal probability. A sample is selected at random with respect to the probability density in each interval, If multiple variables are sampled simultaneously, then values obtained for each are paired in a random manner with the n values of the other variables. In some cases, the pairing is restricted to obtain specified correlations amongst the input variables. Many simulation codes have input parameters that are uncertain and can be specified by a distribution, To perform uncertainty analysis and sensitivity analysis, random values are drawn from the input parameter distributions, and the simulation is run with these values to obtain output values. If this is done repeatedly, with many input samples drawn, one can build up a distribution of the output as well as examine correlations between input and output variables. Media Directory; Software Abstract; Media includes Source Code; Sample Problem Input Data; Installation Instructions and User Guide;/1 CD ROM Y 1 MLT-PLTFM Most recent Linux/Unix systems (RedHat, IRIX, AIS, Solaris, etc.) Sampling should be very fast, at most a few seconds for large sample sets with lots of correlations Sviler, L.P. and G.D. WYss. (2004) "A Users Guide to Sandias Latin Hypercube Sampling Software: LHS UNIX Library/Standalone Version". SAND04-XXXX(tbd). Iman, R. L. and Shortencarier, M. J. (1984) "A Fortran 77 Program and Users Guide for the Generation of Latin Hypercube and Random Samples for Use with Computer Models". SAND83-2365. Wyss, G. 0. and Jorgensen, K. H. (1998) "A Users Guide to LHS: Sandias Latin Hypercube Sampling Software" SAND98-0210. 1MB storate required for program software, and another 3MB required for object files. At least 10MB is recommended Released As-Is 11/29/2006 Sandia National Laboratories AS-IS