Abstract
The complex Wishart distribution has ample applications in science and engineering. In this paper, we give explicit expressions for and , respectively, for particular values of g, h, i, j, , where W follows a complex Wishart distribution. For specific values of g, h, i, j, we first write and in terms of zonal polynomials and then by using results on integration evaluate resulting expressions. Several expected values of matrix-valued functions of a complex Wishart matrix have also been derived.
Keywords:
complex matrix; moments; multivariate gamma function; random matrix; trace; zonal polynomials; Wishart distribution MSC:
62H10
1. Introduction
The complex Wishart distribution has applications in various fields, including statistics, finance, physics, and engineering. Properties such as marginal and conditional distributions of sub-matrices, distribution of diagonal elements, and distribution of the determinant of the complex Wishart matrix W, including, of course, its moments, are therefore of broad interest. The motivation of the present paper is therefore to derive expressions for expected values of functions of W to further enrich the already existing literature on the complex Wishart distribution. For some early work on the complex Wishart distribution the reader is referred to Goodman [1], James [2], Khatri [3,4], Krishnaiah [5], Srivastava [6], Shaman [7], and Tan [8]. Some related material is considered in Alfano et al. [9], Carmeli [10], Cunden, Dahlqvist, and O’Connell [11], Deng et al. [12], Ermolova and Tirkkonen [13], Gomez et al. [14], Kumar [15], Nielsen, Skriver, and Conradsen [16], Shakil and Ahsanullah [17], Tague and Caldwell [18], and Tralli and Conti [19]. Systematic treatment of the complex Wishart distribution is available in Andersen et al. [20].
If , , , then its p.d.f. is
where determinant of A, denotes the conjugate transpose of A, means that A is Hermitian positive definite, and the complex multivariate gamma function is defined by
The complex Wishart distribution can be derived as the joint distribution of sample variances and covariances from a complex multivariate normal population (Goodman [1]). The parameter n need not be an integer, but, when n is not an integer, W can no longer be interpreted as a matrix of sample variances and covariances.
The complex Wishart distribution frequently arises in multivariate analysis as distributions of complex random matrices (for example, see Gupta and Nagar [21]) and hence plays a pivotal role in various branches of science and engineering. For properties and different variations of the Wishart distribution the reader is referred to James [2], Nagar, Gupta and Sánchez [22], Latec and Massam [23], Nagar, Roldán-Correa, and Gupta [24], Di Nardo [25], Dharmawansa and McKay [26], Hillier and Can [27], and Tralli and Conti [19].
Several authors have derived expected values of functions of a complex Wishart matrix. In her doctoral thesis, Grace Wahba [28] gave expressions for . Shaman [7] derived several expected values including . For any constant square matrix A of order m, Tague and Caldwell [18] derived , , and many other useful results. Sultan and Tracy [29] gave expressions for , , and . Maiwald and Kraus [30] gave approximations for , , , and . Hélène Massam and her co-workers (see Graczyk, Letac, and Massam [31]) gave a number of results which include , , , and . Several results on expected values of functions of a complex Wishart matrix including , , , , , and are available in Nagar and Gupta [32]. A recurrence relation for , where k is a non-negative integer, and several special cases of , are available in Pielaszkiewicz, von Rosen, and Singull [33]. A collection of moments of the Wishart distribution is given in Holgersson and Pielaszkiewicz [34].
In this article, we compute expected values of functions of complex Wishart and inverted complex Wishart matrices. By definition
and
where g, h, i, and j are non-negative integers, and .
In Section 3 and Section 4, we give explicit expressions for and , respectively, for specific values of , and j. More precisely, we give expressions for , , , , , , , , , , , , , , , , , , , , , , , , and . Several special cases of these results for are given in Section 5. Further, these special cases have been used to evaluate expected values of matrix-valued functions of a complex Wishart matrix and inverted complex Wishart matrices. Finally, Section 6 closes the paper with concluding remarks.
2. Some Known Definitions and Results
The Pochhammer symbol, denoted by , is defined as for , and . The ordered partition of k is defined by , , . The complex generalized hypergeometric coefficient , for an ordered partition of k, is given as
where is the number of non-zero s. Using (5), the computation of can be carried out for ordered partitions of k. These coefficients are in Table 1 for .
Table 1.
Values of .
For an ordered partition of k, we define and as
and
respectively. The coefficients for are in Table 2. We denote by the zonal polynomial (James [2]) of an complex symmetric matrix X corresponding to the ordered partition . For small values of k, explicit formulas for derived in Khatri [35] are listed in Appendix A.
Table 2.
Values of .
Next, we give two integrals involving zonal polynomials. These results will be used to evaluate integrals in Section 3 and Section 4.
Lemma 1.
Let Z be a Hermitian positive definite matrix of order m and let T be an Hermitian matrix. Then, for , we have
Lemma 2.
Let Z be a Hermitian positive definite matrix of order m and let T be an Hermitian matrix. Then, for , we have
Lemmas 1 and 2 are in James [2] and Khatri [4].
3. Expressions for
Consider the expected value of as
Writing in terms of zonal polynomials (see Appendix A) and integrating the resulting expression by using (6), we obtain
Now, substituting for , , and from Table 1 and Appendix A in the above expression, we have
Writing in terms of zonal polynomials by using results given in Appendix A and integrating the resulting expression by applying (6), we obtain
Now, substituting for , , , and above, we have
Following the procedure described above, we evaluate explicitly for selected values of g, h, i, j given by
4. Expressions for
The expected value of can be derived as
Expressing in terms of zonal polynomials (see Appendix A) and integrating the resulting expression with the aid of (7), we can write
Now, substituting for , from Table 2 and for and from Appendix A, we have
Writing in terms of zonal polynomials (see Appendix A) and applying (7) to integrate the resulting expression, we arrive at
Now, substituting for , , , , and above, we have
Similarly, following the procedure described above, we obtain
5. Special Cases
By substituting into results given in Section 3 and Section 4, several special cases can be derived for . Thus, substituting appropriately, we obtain
Further, using the unitary invariance (Khatri, Khattree, and Gupta [36], Gupta, Nagar, and Vélez-Carvajal [37], Nagar and Gupta [32], and Shaman [7]) of , we can easily derive
6. Discussion and Conclusions
In Section 3 and Section 4, we evaluated as many as twenty-six expected values of the type and , . In Section 5, these expected values were simplified for . Further, using the unitary invariance of , we also computed a number of expected values of matrix-valued functions of W.
For , explicit formulas for available in Khatri [35] are reproduced in Appendix A. By using these expressions for zonal polynomials, through a linear equation system, we expressed traces such as , in terms of zonal polynomials. These results are also summarized in Appendix A. Further, using these expressions, for specific values of g, h, i, j, and Lemmas 1 and 2 adequately, we computed several expected values given in Section 3 and Section 4. Clearly, the method applied in this paper is simple, straightforward, gives explicit expressions for expected values of functions of W and , and does not require any advanced mathematical tool. The present method requires results on zonal polynomials and two lemmas. To find moments of higher order, for example , we will require zonal polynomials for .
Author Contributions
Conceptualization, D.K.N., A.R.-C. and S.N.; methodology, D.K.N., A.R.-C. and S.N.; writing—original draft preparation, D.K.N., A.R.-C. and S.N.; writing—review and editing, D.K.N., A.R.-C. and S.N. All authors have read and agreed to the published version of the manuscript.
Funding
This research received no external funding.
Data Availability Statement
Not applicable.
Conflicts of Interest
The authors declare no conflict of interest.
Appendix A
For small values of k, explicit formulas for are available in Khatri [35] as
By using the above expressions for zonal polynomials it is straightforward, through a linear equation system, to express traces such as in terms of zonal polynomials. Thus, after some algebra, we obtain
References
- Goodman, N.R. Statistical analysis based on a certain multivariate complex Gaussian distribution (An introduction). Ann. Math. Stat. 1963, 34, 152–177. [Google Scholar] [CrossRef]
- James, A.T. Distributions of matrix variate and latent roots derived from normal samples. Ann. Math. Stat. 1964, 35, 475–501. [Google Scholar] [CrossRef]
- Khatri, C.G. Classical statistical analysis based on a certain multivariate complex Gaussian distribution. Ann. Math. Stat. 1965, 36, 98–114. [Google Scholar] [CrossRef]
- Khatri, C.G. On certain distribution problems based on positive definite quadratic functions in normal vectors. Ann. Math. Stat. 1966, 37, 468–479. [Google Scholar] [CrossRef]
- Krishnaiah, P.R. Some recent developments on complex multivariate distributions. J. Multivar. Anal. 1976, 6, 1–30. [Google Scholar] [CrossRef]
- Srivastava, M.S. On the complex Wishart distribution. Ann. Math. Stat. 1965, 36, 313–315. [Google Scholar] [CrossRef]
- Shaman, P. The inverted complex Wishart distribution and its application to spectral estimation. J. Multivar. Anal. 1980, 10, 51–59. [Google Scholar] [CrossRef]
- Tan, W.Y. Some distribution theory associated with complex Gaussian distribution. Tamkang J. 1968, 7, 263–302. [Google Scholar]
- Alfano, G.; Tulino, A.; Lozano, A.; Verdú, S. Random Matrix Transforms and Applications via Non-Asymptotic Eigenanalysis; International Zurich Seminar on Communications: Zurich, Switzerland, 2006; pp. 18–21. [Google Scholar]
- Carmeli, M. Statistical Theory and Random Matrices in Physics; Marcel Dekker: New York, NY, USA, 1983. [Google Scholar]
- Cunden, F.D.; Dahlqvist, A.; O’Connell, N. Integer moments of complex Wishart matrices and Hurwitz numbers. Ann. Inst. Henri Poincaré D 2021, 8, 243–268. [Google Scholar] [CrossRef]
- Deng, X.; López-Martínez, C.; Chen, J.; Han, P. Statistical Modeling of Polarimetric SAR Data: A Survey and Challenges. Remote Sens. 2008, 9, 348. [Google Scholar] [CrossRef]
- Ermolova, N.Y.; Tirkkonen, O. Distribution of diagonal elements of a general central complex Wishart matrix. IEEE Commun. Lett. 2012, 16, 1373–1376. [Google Scholar] [CrossRef]
- Gomez, L.; Alvarez, L.; Mazorra, L.; Frery, A.C. Classification of complex Wishart matrices with a diffusion-reaction system guided by stochastic distances. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 2015, 373, 20150118. [Google Scholar] [CrossRef] [PubMed]
- Kumar, S. Eigenvalue statistics for the sum of two complex Wishart matrices. EPL 2014, 107, 60002. [Google Scholar] [CrossRef]
- Nielsen, A.A.; Skriver, H.; Conradsen, K. Complex Wishart Distribution Based Analysis of Polarimetric Synthetic Aperture Radar Data. In Proceedings of the 2007 International Workshop on the Analysis of Multi-Temporal Remote Sensing Images, Leuven, Belgium, 18–20 July 2007; pp. 1–6. [Google Scholar]
- Shakil, M.; Ahsanullah, M. Some inferences on the distribution of the Demmel condition number of complex Wishart matrices. Spec. Matrices 2017, 5, 127–138. [Google Scholar] [CrossRef]
- Tague, J.A.; Caldwell, C.I. Expectations of useful complex Wishart forms. Multidimens. Syst. Signal Process. 1994, 5, 263–279. [Google Scholar] [CrossRef]
- Tralli, V.; Conti, A. Noncentral complex Wishart matrices: Moments and correlation of minors. Random Matrices Theory Appl. 2022, 11, 2250009. [Google Scholar] [CrossRef]
- Andersen, H.H.; Højbjerre, M.; Sørensen, D.; Eriksen, P.S. Linear and Graphical Models; For the Multivariate Complex Normal Distribution, Lecture Notes in Statistics, 101; Springer: New York, NY, USA, 1995. [Google Scholar]
- Gupta, A.K.; Nagar, D.K. Matrix Variate Distributions; Chapman and Hall/CRC: Boca Raton, FL, USA, 2000. [Google Scholar]
- Nagar, D.K.; Gupta, A.K.; Sánchez, L.E. A class of integral identities with Hermitian matrix argument. Proc. Am. Math. Soc. 2006, 134, 3329–3341. [Google Scholar] [CrossRef]
- Letac, G.; Maasam, H. The noncentral Wishart as an exponential family, and its moments. J. Multivar. Anal. 2008, 99, 1393–1417. [Google Scholar] [CrossRef]
- Nagar, D.K.; Roldán-Correa, A.; Gupta, A.K. Extended matrix variate gamma and beta functions. J. Multivar. Anal. 2013, 122, 53–69. [Google Scholar] [CrossRef]
- Di Nardo, E. On a symbolic representation of non-central Wishart random matrices with applications. J. Multivar. Anal. 2014, 125, 121–135. [Google Scholar] [CrossRef]
- Dharmawansa, P.; McKay, M.R. Extreme eigenvalue distributions of some complex correlated non-central Wishart and gamma-Wishart random matrices. J. Multivar. Anal. 2011, 102, 847–868. [Google Scholar] [CrossRef]
- Hillier, G.; Kan, R. Properties of the inverse of a noncentral Wishart matrix. Econom. Theory 2022, 38, 1092–1116. [Google Scholar] [CrossRef]
- Wahba, G. Cross-Spectral Distribution Theory for Mixed Spectra and Estimation of Prediction Filter Coefficients. Ph.D. Thesis, Stanford University, Stanford, CA, USA, 1966. [Google Scholar]
- Sultan, S.A.; Tracy, D.S. Moments of the complex Wishart distribution. Int. J. Math. Stat. Sci. 1999, 8, 51–62. [Google Scholar]
- Maiwald, D.; Kraus, D. Calculation of moments of complex Wishart and complex inverse Wishart distributed matrices. IEE Proc.-Radar Sonar Navig. 2000, 147, 162–168. [Google Scholar] [CrossRef]
- Graczyk, P.; Letac, G.; Massam, H. The complex Wishart distribution and the symmetric group. Ann. Stat. 2003, 31, 287–309. [Google Scholar] [CrossRef]
- Nagar, D.K.; Gupta, A.K. Expectations of functions of complex Wishart matrix. Acta Appl. Math. 2011, 113, 265–288. [Google Scholar] [CrossRef]
- Pielaszkiewicz, J.; von Rosen, D.; Singull, M. On , where . Commun. Stat.—Theory Methods 2017, 46, 2990–3005. [Google Scholar] [CrossRef]
- Holgersson, T.; Pielaszkiewicz, J. A Collection of Moments of the Wishart Distribution. In Recent Developments in Multivariate and Random Matrix Analysis; Holgersson, T., Singull, M., Eds.; Springer: Cham, Switzerland, 2020; pp. 147–162. [Google Scholar]
- Khatri, C.G. On the moments of traces of two matrices in three situations for complex multivariate normal populations. Sankhyā Indian J. Stat. Ser. A 1970, 32, 65–80. [Google Scholar]
- Khatri, C.G.; Khattree, R.; Gupta, R.D. On a class of orthogonal invariant and residual independent matrix distributions. Sankhyā Indian J. Stat. Ser. B 1991, 53, 1–10. [Google Scholar]
- Gupta, A.K.; Nagar, D.K.; Vélez-Carvajal, A.M. Unitary invariant and residual independent matrix distributions. Comput. Appl. Math. 2009, 28, 63–86. [Google Scholar] [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).