Summary
A dual approach for matrix by matrix derivatives is proposed, based on the work ofBalestra and the idea ofDwyer/MacPhail. The two derivative concepts are calledB-type derivative for the form ∂B/∂A = (∂b kl /∂A), because it was studied fully inBalestra, andA-type derivative for the form ∂B/ /∂A = (∂B/∂a ij ). Both derivative concepts are linked by permutation matrices, which also reveil the duality aspect more clearly, and can be transformed to each other very easily. The derivatives are applied to least squares estimates and posterior means in the general regression model and the “seemingly unrelated regression” (SUR)-system, introduced byZellner. The derivatives with respect to the covariance matrix is related to the local sensitivity concept ofLeamer [1978], while the derivatives with respect to the data-matrix, also called local resistance, is linked with the robustness concept ofTukey. The newly definedB-derivative enables an easier interpretation of the results.
Similar content being viewed by others
References
Balestra, P.: La Dérivation Matricelle. Collection de L'IME No.12, Sirey, Paris 1976.
Chamberlain, G., andE. Leamer: Matrix Weighted Averages and Posterior Bounds. Journal of the Royal Statistical Society. Series B,38, 1976, 73–84.
Dwyer, P.S.: Some Applications of Matrix Derivatives in Multivariate Analysis, J. o. Am. Stat. Ass.,62, 1967, 607–625.
Dwyer, P.S., andM.S. MacPhail: Symbolic Matrix Derivatives. Annals Math. Stat.19, 1948, 517–534.
Leamer, E.E.: A Class of Informative Prior and Distributed Lag Analysis. Econometrica42, 1972, 1059–1081.
—: Specification Searches. J. Wiley, New York 1978.
MacRae, E.Ch.: Matrix Derivatives with an Application to an Adaptive Linear Decision Problem. Annals of Statistics. Vol.2, 1974, 337–346.
Magnus, J.R., andH. Neudecker: The Elimination Matrix: Some Theorems and Applications. SIAM J. Alg. and Disc. Math., 1980, 422–449.
Magnus, J.R., andH. Neudecker: The Commutation Matrix: Some Properties and Applications. Annals of Statistics, 1979, 381–394.
Mosteller, F., andJ.W. Tukey: Data Analysis and Regression. Addison-Wesley, New York 1977.
Neudecker H., andT. Wansbeck: Some Results on Super Commutation Matrices with Statistical Application. Report 80.04/AE12/80, University of Amsterdam, 1980.
Polasek, W.: Local Sensitivity Analysis in the General Linear Model. University of Southern California, MRG#8006, 1980.
Polasek, W.: Three Methods of Matrix Derivatives and Local Sensitivity Analysis, 1985, to appear.
Rogers, G.S.: Matrix Derivatives. Marcel Dekker, New York-Basel 1980.
Tukey, J.W.: Exploratory Data Analysis. Addison-Wesley, New York 1977.
Zellner, A.: An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests of Aggregation Bias. Journal of the American Statistical Association,57, 1962, 348–368.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Polasek, W. A dual approach for matrix-derivatives. Metrika 32, 275–292 (1985). https://doi.org/10.1007/BF01897818
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF01897818