SciELO - Scientific Electronic Library Online

 
vol.6 issue3Evaluation algorithm for a decomposed simplicial piecewise-linear formulationRobust generic model feedback under model uncertainties: application of a tubular reactor for the treatment of industrial plating wastewater author indexsubject indexsearch form
Home Pagealphabetic serial listing  

Services on Demand

Journal

Article

Indicators

Related links

  • Have no similar articlesSimilars in SciELO

Share


Journal of applied research and technology

On-line version ISSN 2448-6736Print version ISSN 1665-6423

J. appl. res. technol vol.6 n.3 Ciudad de México Dec. 2008

 

Alternative methods of calculation of the pseudo inverse of a non full-rank matrix

 

M. A. Murray-Lasso

 

Facultad de Ingeniería UNAM.

 

ABSTRACT

The calculation of the pseudo inverse of a matrix is intimately related to the singular value decomposition which applies to any matrix be it singular or not and square or not. The matrices involved in the singular value decomposition of a matrix A are formed with the orthogonal eigen vectors of the symmetric matrices ATA and AAT associated with their nonzero eigenvalues which forms a diagonal matrix. If instead of using the eigenvectors, which are difficult to calculate, we use any set of vectors that span the same spaces, which are easier to obtain, we can get simpler expressions for calculating the pseudoinverse, although the diagonal matrix of eigenvalues is filled. All numerical work to obtain the pseudo inverse whose components are rational numbers when the original matrix is also rational reduces to elementary row operations. We can, thus, generalize the least-squares/ minimum-length normal equations for full-rank matrices and solve said problems and obtain the pseudo inverse in terms of A and AT. without solving any eigen problems or factoring matrices.

Key words: pseudo inverse, singular values, normal equations, least-squares, minimum-length.

 

RESUMEN

El cálculo de la seudo inversa de una matriz está íntimamente relacionado con la descomposición de valores singulares aplicable a cualquier matriz, singular o no y cuadrada o no. Las matrices involucradas en la descomposición en valores singulares de una matriz A están formadas con los vectores característicos ortogonales de las matrices simétricas ATA y AAT asociados con los valores característicos no nulos, los cuales forman una matriz diagonal. Si, en lugar de usar los vectores característicos, los cuales son difíciles de calcular, se usa cualquier conjunto de vectores que generan los mismos espacios, que son más fáciles de obtener, se pueden obtener expresiones más simples para el cálculo de la seudo inversa, no obstante que la matriz diagonal se llena. Todo el trabajo numérico se reduce a operaciones elementales de filas obteniéndose seudo inversas con componentes racionales cuando la matriz original tiene componentes racionales. De esta manera podemos generalizar las ecuaciones normales de mínimos cuadrados / longitud mínima de matrices de rango completo, resolver el problema y obtener la seudo inversa en términos de A y AT sin resolver problemas de vectores característicos o factorizar matrices.

Palabras clave: seudo inversa, valores singulares, ecuaciones normales, mínimos cuadrados, longitud mínima.

 

DESCARGAR ARTÍCULO EN FORMATO PDF

 

13. REFERENCES

[1] Dahlquist, G. & A. Bjork (1974): Numerical Methods. Prentice - Hall, Inc., Englewood Cliffs, NJ.         [ Links ]

[2] Golub, G. H. & C. Reinsch (1971): "Singular Value Decomposition and Least Square Solutions." Contribution 1/10 in Wilkinson, J. H. & C. Reinsch: Handbook for Automatic Computation. Volume II, Linear Algebra. Springer - Verlag, New York, pp. 134 - 151.         [ Links ]

[3] Noble, B. (1969): Applied Linear Algebra. Prentice - Hall, Inc., Englewood Cliffs, NJ.         [ Links ]

[4] Murray-Lasso, M. A. (2007): "Linear Vector Space Derivation of New Expressions for the Pseudo Inverse of Rectangular Matrices." Journal of Applied Research and Technology, Vol. 5, No. 3, pp. 150 - 159.         [ Links ] .

[5] Zadeh, L. & C. Desoer (1963): Linear System Theory: The State Space Approach. Mc Graw - Hill Book Company, Inc., New York        [ Links ]

[6] Lanczos, C. (1961): Linear Differential Operators. D. Van Nostrand Company Limited, Londres.         [ Links ]

[7] Wolfram, S. (1991): Mathematica: A System for Doing Mathematics by Computer. Second Edition. Addison - Wesley Publishing Company, Inc., Redwood City, CA.         [ Links ]

[8] Moore, E. H. (1920): "On the Reciprocal of the General Algebraic Matrix." Bulletin of the American Mathematical Society, Vol. 26, pp. 394 - 395.         [ Links ]

[9] Penrose, R. (1955): "A Generalized Inverse for Matrices." Proceedings of the Cambridge Philosophical Society, Vol. 51, pp. 406 - 413.         [ Links ]

[10] Davis, P. J. (1963): Interpolation and Approximation. Blaisdell Publishing Company, New York.         [ Links ]

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License