CHAPTER 7 - Conditional Expectation and Martingale Theory

International audience

We develop a nonlinear martingale theory for time discrete processes $(Y_n)_{n\in \NN_0}$. These processes are defined on any probability space $(\O,\F,\F_n,\P)_{n\in\NN_0}$ and have values in a metric space (N,d) of nonpositive curvature (in the sense of A. D. Alexandrov). The defining martingale property for such processes is \[ \E(Y_{n+1}|\F_n)=Y_n, \qquad \P\mbox{-a.s.,} \] where the conditional expectation on the left-hand side is defined as the minimizer of the functional \[ Z\mapsto\E d^2(Z,Y_{n+1}) \] within the space of $\F_n$-measurable maps $Z\dvtx \O\to N$. We give equivalent characterization of N-valued martingales (using merely the usual linear conditional expectations) and derive fundamental properties of these martingales, for example, a martingale convergence theorem. Finally, we exploit the relation with harmonic maps. It turns out that a map $f\dvtx M\to N$ is harmonic w.r.t. a given Markov kernel p on M if and only if it maps Markov chains $(X_n)_{n\in\NN}$ (with transition kernel p) on M onto martingales $(f(X_n))_{n\in\NN}$ with values in $N$. The nonlinear heat flow $f\dvtx \N_0\times M\to N$ of a given initial map $f(0,\cdot)\dvtx M\to N$ at time n is obtained as the filtered expectation, \[ f(n,x) := \E_x [ f(X_n) |\!|\!| (\F_k)_{k\ge 0}] \] of the random map $f(X_n)$. Similarly, the unique solution to the Dirichlet problem for a given map $g\dvtx M\to N$ and a subset $D\subset M$ is obtained as \[ f(x) := \E_x [ g(X_{\tau(D)})|\!|\!| (\F_k)_{k\ge 0}]. \] In both cases, a crucial role is played by the notion of expectation $\E_x [\cdot |\!|\!| (\F_k)_{k\ge 0}]$ which will be discussed in detail. Moreover, we prove Jensen's inequality for expectations and expectations and we prove (weak and strong) laws of large numbers for sequences of i.i.d. random variables with values in N. Our theory is an extension of the classical linear martingale theory and of the nonlinear theory of martingales with values in manifolds as developed, for example, in Emery and Kendall. The goal is to extend the previous framework towards processes with values in metric spaces. This will lead to a stochastic approach to the theory of (generalized) harmonic maps with values in such singular spaces as developed by Jost and Korevaar and Schoen.

On abstract conditional expectations

Convex Functions, Partial Orderings, and Statistical Applications

On estimation of random variables via the martingale convergence theorem

The similarities between martingale convergence theory and pointwise ergodic theory are now well known [5, 7, 9, 10]. In [5] the similarity between the proofs of the Hopf– Dunford–Schwartz individual ergodic theorem and the martingale convergence theorem is systematically exploited to produce very general ” maximal ergodic ” inequalities for certain sequences of contractions on L1-spaces. A different approach by Rota [10] and Rao [9] leads to a unified convergence theory for martingales and Abel limits. Bishop [1] has produced ” upcrossing” inequalities which yield both theChacon-Ornstein theorem [4] and the martingale convergence theorem.

We discuss the construction of stopping lines in the branching random walk and thus the existence of a class of supermartingales indexed by sequences of stopping lines. Applying a method of Lyons (1997) and Lyons, Pemantle and Peres (1995) concerning size biased branching trees, we establish a relationship between stopping lines and certain stopping times. Consequently we develop conditions under which these supermartingales are also martingales. Further we prove a generalization of Biggins' Martingale Convergence Theorem, Biggins (1977a) within this context.

Four types of convergence for sequences of convex sets are investigated. Their interrelationships are explored.

We investigate a possible definition of expectation and conditional expectation for random variables with values in a local field such as the $p$-adic numbers. We define the expectation by analogy with the observation that for real-valued random variables in $L^2$ the expected value is the orthogonal projection onto the constants. Previous work has shown that the local field version of $L^\infty$ is the appropriate counterpart of $L^2$, and so the expected value of a local field-valued random variable is defined to be its ``projection'' in $L^\infty$ onto the constants. Unlike the real case, the resulting projection is not typically a single constant, but rather a ball in the metric on the local field. However, many properties of this expectation operation and the corresponding conditional expectation mirror those familiar from the real-valued case; for example, conditional expectation is, in a suitable sense, a contraction on $L^\infty$ and the tower property holds. We also define the corresponding notion of martingale, show that several standard examples of martingales (for example, sums or products of suitable independent random variables or ``harmonic'' functions composed with Markov chains) have local field analogues, and obtain versions of the optional sampling and martingale convergence theorems.

We discuss the conditional expectations and martingales in relevance with G-strongly quasi-invariant states on a C*-algebra A, where G is a separable locally compact group of ∗-automorphisms of A. In the von Neumann algebra A of the GNS representation, we define a unitary representation of the group and a group Ĝ of ∗-automorphisms of A, which is homomorphic to G. For the case of compact G, we find a Ĝ-invariant state on A and define a conditional expectation with range the Ĝ-fixed subalgebra. When G is the union of increasing compact groups, we construct a sequence of conditional expectations and thereby construct (backward) martingales, which have limits by the martingale convergence theorem. As an example we consider S∞ the group of local permutations which acts on a C*-algebra of infinite tensor product of finite dimensional C*-algebras. We also find an application in classical spin systems.

In this article we quantify almost sure martingale convergence theorems in terms of the tradeoff between asymptotic almost sure rates of convergence (error tolerance) and the respective modulus of convergence.For this purpose we generalize an elementary quantitative version of the first Borel-Cantelli lemma on the statistics of the deviation frequencies, which was recently established by the authors.First we study martingale convergence in L2, and in the setting of the Azuma-Hoeffding inequality.In a second step we study the strong law of large numbers for martingale differences.Applications are the tradeoff for the multicolor generalized Plya urn processes, the generalized Chinese restaurant process, statistical M-estimators, as well as excursion frequencies of the Galton-Watson branching process.

Conditional expectations and submartingale sequences of random Schwartz distributions

Martingales are a class of stochastic processes which has had profound influence on the development of probability and stochastic processes. There are few areas of the subject untouched by martingales. We will survey the theory and applications of discrete time martingales and end with some recent developments in mathematical finance. Here is what to expect in this chapter: Absolute continuity and the Radon-Nikodym Theorem. Conditional expectation. Martingale definitions and elementary properties and examples. Martingale stopping theorems and applications. Martingale convergence theorems and applications. The fundamental theorems of mathematical finance.

Martingale Convergence of Generalized Conditional Expectations

1 Geometric properties. -1.1 Introduction. -1.2 Uniformly convex spaces. -1.3 Strictly convex Banach spaces. -1.4 The modulus of convexity. -1.5 Uniform convexity, strict convexity and reflexivity. -1.6 Historical remarks. -2 Smooth Spaces. -2.1 Introduction. -2.2 The modulus of smoothness. -2.3 Duality between spaces. -2.4 Historical remarks. -3 Duality Maps in Banach Spaces. -3.1 Motivation. -3.2 Duality maps of some concrete spaces. -3.3 Historical remarks. -4 Inequalities in Uniformly Convex Spaces. -4.1 Introduction. -4.2 Basic notions of convex analysis. -4.3 p-uniformly convex spaces. -4.4 Uniformly convex spaces. -4.5 Historical remarks. -5 Inequalities in Uniformly Smooth Spaces. -5.1 Definitions and basic theorems. -5.2 q-uniformly smooth spaces. -5.3 Uniformly smooth spaces. -5.4 Characterization of some real Banach spaces by the duality map. -5.4.1 Duality maps on uniformly smooth spaces. -5.4.2 Duality maps on spaces with uniformly Gateaux differentiable norms. -6 Iterative Method for Fixed Points of Nonexpansive Mappings. -6.1 Introduction. -6.2 Asymptotic regularity. -6.3 Uniform asymptotic regularity. -6.4 Strong convergence. -6.5 Weak convergence. -6.6 Some examples. -6.7 Halpern-type iteration method. -6.7.1 Convergence theorems. -6.7.2 The case of non-self mappings. -6.8 Historical remarks. -7 Hybrid Steepest Descent Method for Variational Inequalities. -7.1 Introduction. -7.2 Preliminaries. -7.3 Convergence Theorems. -7.4 Further Convergence Theorems. -7.4.1 Convergence Theorems. -7.5 The case of Lp spaces, 1 2. -7.6 Historical remarks. 8 Iterative Methods for Zeros of F -Accretive-Type Operators. -8.1 Introduction and preliminaries. -8.2 Some remarks on accretive operators. -8.3 Lipschitz strongly accretive maps. -8.4 Generalized F -accretive self-maps. -8.5 Generalized F -accretive non-self maps. -8.6 Historical remarks. -9 Iteration Processes for Zeros of Generalized F -Accretive Mappings. -9.1 Introduction. -9.2Uniformly continuous generalized F -hemi-contractive maps. -9.3 Generalized Lipschitz, generalized F -quasi-accretive mappings. -9.4 Historical remarks. -10 An Example Mann Iteration for Strictly Pseudo-contractive Mappings. -10.1 Introduction and a convergence theorem. -10.2 An example. -10.3 Mann iteration for a class of Lipschitz pseudo-contractive maps. -10.4 Historical remarks. -11 Approximation of Fixed Points of Lipschitz Pseudo-contractive Mappings. -11.1 Lipschitz pseudo-contractions. -11.2 Remarks. -12 Generalized Lipschitz Accretive and Pseudo-contractive Mappings. -12.1 Introduction. -12.2 Convergence theorems. -12.3 Some applications. -12.4 Historical remarks. -13 Applications to Hammerstein Integral Equations. -13.1 Introduction. -13.2 Solution of Hammerstein equations. -13.2.1 Convergence theorems for Lipschitz maps. -13.2.2 Convergence theorems for bounded maps. -13.2.3 Explicit algorithms. -13.3 Convergence theorems with explicit algorithms. -13.3.1 Some useful lemmas. -13.3.2 Convergence theorems with coupled schemes for the case of Lipschitz maps. -13.3.3 Convergence in Lp spaces, 1 2: . -13.4 Coupled scheme for the case of bounded operators. -13.4.1 Convergence theorems. -13.4.2 Convergence for bounded operators in Lp spaces, 1 2:. -13.4.3 Convergence theorems for generalized Lipschitz maps. -13.5 Remarks and open questions. -13.6 Exercise. -13.7 Historical remarks. -14 Iterative Methods for Some Generalizations of Nonexpansive Maps. -14.1 Introduction. -14.2 Iteration methods for asymptotically nonexpansive mappings. -14.2.1 Modified Mann process. -14.2.2 Iteration method of Schu. -14.2.3 Halpern-type process. -14.3 Asymptotically quasi-nonexpansive mappings. -14.4 Historical remarks. -14.5 Exercises. -15 Common Fixed Points for Finite Families of Nonexpansive Mappings. -15.1 Introduction. -15.2 Convergence theorems for a family of nonexpansive mappings. -15.3 Non-self mappings. -16 Common Fixed Po