An Introduction to Matrix Variate Stochastics推荐.pdfVIP

Chapter 10 An Introduction to Matrix Variate Stochastics In this chapter, we introduce the reader to matrix variate stochastics. It is intended to set the scene for Wishart processes, which will be covered in the next chapter. We begin by recalling notation and introducing some basic functions used throughout both chapters. This will bring us in a position to discuss matrix valued random vari- ables, matrix valued stochastic processes, and matrix valued stochastic differential equations. To illustrate these concepts, we apply them to the matrix valued version of the Ornstein-Uhlenbeck process and a multidimensional version of the MMM. The main references for this chapter are Gupta and Nagar (2000) and Pfaffel (2008). 10.1 Basic Definitions and Functions In this section, we fix primarily notation. Definition 10.1.1 We employ the following notation: • we denote by Mm,n () the set of all m ×n matrices with entries in . If m = n, we write Mn () instead; • we write GL(p) for the group of all invertible matrices of Mp (); • let Sp denote the linear subspace of all symmetric matrices of Mp (); • let S + (S −) denote the set of all symmetric positive (negative) definite matrices p p of Mp (); • denote by S + the closure of S + in Mp (), that is the set of all symmetric positive p p semidefinite matrices of Mp (). The next definition provides a one-to-one relationship between vectors and ma- trices. J. Baldeaux, E. Platen, Functionals of Multidimensional Diffusions with Applications 243 to Finance, Bocconi Springer Series 5, DOI 10.1007/978-3-319-00747-2_10, © Springer International Publishing Switzerland 2013 244 10 An Introduction to Matrix Variate Stochastics Definition 10.1.2 Let A ∈ Mm,n () with columns ai ∈ m , i = 1, . . . , n. Define the function vec : Mm,n () → mn via ⎛a 1 ⎞