Generalized Enhanced Multivariance Product Representation for Data Partitioning: Constancy Level

TUNGA M. A., Demiralp M.

International Conference on Numerical Analysis and Applied Mathematics (ICNAAM), Halkidiki, Greece, 19 - 25 September 2011, vol.1389 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume: 1389
  • Doi Number: 10.1063/1.3637819
  • City: Halkidiki
  • Country: Greece
  • Istanbul Technical University Affiliated: Yes


Enhanced Multivariance Product Representation (EMPR) method is used to represent multivariate functions in terms of less-variate structures. The EMPR method extends the HDMR expansion by inserting some additional support functions to increase the quality of the approximants obtained for dominantly or purely multiplicative analytical structures. This work aims to develop the generalized form of the EMPR method to be used in multivariate data partitioning approaches. For this purpose, the Generalized HDMR philosophy is taken into consideration to construct the details of the Generalized EMPR at constancy level as the introductory steps and encouraging results are obtained in data partitioning problems by using our new method. In addition, to examine this performance, a number of numerical implementations with concluding remarks are given at the end of this paper.