The Annals of Statistics

Nonlinear principal components I. Absolutely continuous random variables with positive bounded densities

Ernesto Salinelli

Full-text: Open access


Nonlinear principal components for an absolutely continuous random vector X with positive bounded density are defined as the solution of a variational problem in a suitable function space. In this way transformations depending on all the components of X are obtained. Some properties of nonlinear principal components are proved: in particular, it is shown that the set of nonlinear principal transformations of X is an orthonormal basis for the function space associated with the optimal problem. The spectral decomposition of X and its covariance matrix with respect to this basis are given. A notion of marginal nonlinear principal components is sketched and the relations with nonlinear principal components are shown. Finally, treating the case of random vectors distributed on unbounded domains, the existence problem is shown to be related to the global existence of the moment generating function of X. Since it is not restrictive, definitions and results are stated in terms of a uniformly distributed random vector.

Article information

Ann. Statist., Volume 26, Number 2 (1998), 596-616.

First available in Project Euclid: 31 July 2002

Permanent link to this document

Digital Object Identifier

Mathematical Reviews number (MathSciNet)

Zentralblatt MATH identifier

Primary: 62H25: Factor analysis and principal components; correspondence analysis
Secondary: 35J20: Variational methods for second-order elliptic equations

Nonlinear principal components moment generating function Sobolev spaces Laplacian


Salinelli, Ernesto. Nonlinear principal components I. Absolutely continuous random variables with positive bounded densities. Ann. Statist. 26 (1998), no. 2, 596--616. doi:10.1214/aos/1028144850.

Export citation


  • Adams, R. A. (1975). Sobolev Spaces. Academic Press, New York. Adams, R. A. and Fournier, J. (1971a). Some imbedding theorem for Sobolev spaces. Canada J. Math. 23 517-530. Adams, R. A. and Fournier, J. (1971b). A compact imbedding theorem for functions without compact support. Canad. Math. Bull. 14 305-309.
  • Arfken, G. (1966). Mathematical Methods for physicists. Academic Press, New York.
  • Bandle, C. (1980). Isoperimetric Inequalities and Applications. Pitman, Boston.
  • Berger, G. (1991). Different eigenvalue asy mptotics for Dirichlet and Neumann boundary conditions of elliptic operators of second order in some irregular domains. Math. Nachr. 152 159-171.
  • Courant, R. and Hilbert, D. (1989). Methods of Mathematical physics 1. Wiley, New York.
  • Darroch, J. N. (1965). An optimal property of principal components. Ann. Math. Statist. 36 1579-1582.
  • De Leeuw, J. and Van Rijekevorsel, H. and Van der Wouden, H. (1981). Nonlinear principal components analysis with B-splines. Methods Oper. Res. 33, 379-393.
  • Donnell, D. J., Buja, A. and Stuetzle, W. (1994). Analy sis of additive dependencies and concurtivities using smallest additive principal components, Ann. Statist. 22 1635-1668.
  • Duchamp, T. and Stuetzle, W. (1993). The geometry of principal curves in the plane. Technical
  • Report 250, Dept. Statistics, Univ. Washington (revised in August 1995). Duchamp, T. and Stuetzle, W. (1995a). Extremal properties of principal curves in the plane. Preprint. Duchamp, T. and Stuetzle, W. (1995b). Geometric properties of principal curves in the plane. Robust Statistics, Data Analy sis, and Computer Intensive Methods 135-152. Springer, Berlin.
  • Egorov, Yu. V. and Shubin, M. A., eds. (1992). Partial differential equations I. Ency clopedia of Mathematical Sciences 30. Springer, Berlin.
  • Flury, B. D. (1994). Discussion of "Analy sis of additive dependencies and concurtivities using smallest additive principal components," by D. J. Donnell, A. Buja and W. Stuetzle, Ann. Statist. 22 1668-1670.
  • Gifi, A. (1990). Nonlinear Multivariate Analy sis. Wiley, New York.
  • Gottlieb, H. P. W. (1985). Eigenvalues of the Laplacian with Neumann boundary conditions. J. Austral. Math. Soc. Ser. B 26 293-309.
  • Gnanadesikan, R. (1977). Methods for Statistical Data Analy sis of Multivariate Observations. Wiley, New York.
  • Hastie, T. and Stuetzle, W. (1989). Principal curves, J. Amer. Statist. Assoc., 84 502-516.
  • Hotelling, H. (1933). Analy sis of a complex of statistical variables into principal components. Journal of Educational Psy chology 26 417-441, 498-520.
  • Koy ak, R. A. (1987). On measuring internal dependence in a set of random variables. Ann. Statist. 15 1215-1228.
  • Jolliffe, I. T. (1986). Principal Component Analy sis. Springer, Berlin.
  • Mikhailov, V. P. (1978). Partial Differential Equations. MIR, Moscow.
  • Mizuta, M. (1984). Generalized principal components analysis invariant under rotations of a coordinate sy stem. J. Japan Statist. Soc. 14 1-9.
  • Ne cas, J. (1967). Les m´ethodes directes en Th´eorie des ´Equations Elliptiques. Masson, Paris.
  • Okamoto, M. and Kanazawa, M. (1968). Minimization of eigenvalues of a matrix and optimality of principal components. Ann. Math. Statist. 39 850-863.
  • Pearson, K. (1901). On lines and planes of closest fit to sy stems of points in space. Philosophical Magazine 2 559-572.
  • Rao, C. R. (1973). Linear statistical inference and its applications. Wiley, New York.
  • Sleeman, B. D. and Zay ed, E. M. E. (1984). Trace formulae for the eigenvalues of the Laplacian. Z. Angew. Math. Phy s. 35 106-115.
  • Treves, F. (1975). Basic Linear Partial Differential Equations. Academic Press, New York.