pRIPPLE: a parallel version of a polynomial-time piecewise linear estimation algorithm

Authors:
Manjula A. Iyer;Layne T. Watson;Jeffrey B. Birch
Affiliations:
Virginia Polytechnic Institute and State University, Blacksburg, Virginia;Virginia Polytechnic Institute and State University, Blacksburg, Virginia;Virginia Polytechnic Institute and State University, Blacksburg, Virginia
Venue:
SpringSim '07 Proceedings of the 2007 spring simulation multiconference - Volume 3
Year:
2007

Citing 3
Cited 0

Multivariate interpolation of large sets of scattered data

ACM Transactions on Mathematical Software (TOMS)
Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography

Communications of the ACM
A two-dimensional interpolation function for irregularly-spaced data

ACM '68 Proceedings of the 1968 23rd ACM national conference

Quantified Score

Hi-index	0.00

Visualization

Abstract

The response functions in many engineering problems are piecewise smooth functions of continuous variables. High quality interpolatory approximations to such functions can be constructed without requiring a large number of expensive function evaluations by using linear local approximations. Most robust estimation algorithms have factorial complexity (in the number of data points) as a result of which their performance degrades rapidly in higher dimensions. RIPPLE (residual initiated polynomial-time piecewise linear estimation) was developed to produce robust estimates of data obtained from high dimensional piecewise linear functions in polynomial time. Its performance was found to be comparable to other robust estimation techniques. A parallel version of RIPPLE has been developed in order to take advantage of the additional computational power of parallel systems. This paper presents the parallel version of RIPPLE and results obtained from a piecewise linear test function.