Mathematics Foundations/21.4 Product Measures

21.4 Product Measures Introduction Product measures extend the concept of measurement to multidimensional spaces by combining measures from individual spaces. They are fundamental in probability theory, statistical mechanics, and functional analysis, allowing us to work with joint distributions and multidimensional integrals. Definition of Product Measures Let and be two measure spaces. The product measure is defined on the product σ-algebra (the smallest σ-algebra containing all rectangles where and ). For any rectangle where and , the product measure is defined as: Construction of Product Measures The construction of product measures typically follows these steps: 1. Define the measure on rectangles as shown above 2. Extend to finite disjoint unions of rectangles 3. Use the Carathéodory Extension Theorem to extend to the entire product σ-algebra Uniqueness A key result is that there exists a unique measure on that agrees with the product measure on all rectangles. This uniqueness is crucial for ensuring that the construction is well-defined. Properties of Product Measures Product measures inherit several important properties from their component measures: 1. If and are σ-finite measures, then is the unique measure on such that for all and . 2. If and are probability measures, then is also a probability measure. 3. If and are complete measures, the product measure may not be complete. However, it can be completed in the standard way. Measurable Functions If is -measurable and is -measurable, then is -measurable. Fubini's Theorem Fubini's Theorem is one of the most important results involving product measures, allowing us to compute integrals over product spaces by iterating integrals over the component spaces. Statement of Fubini's Theorem Let and be σ-finite measure spaces, and let be -measurable. If either: , or (i.e., is integrable) Then: Interpretation Fubini's Theorem states that for suitable functions, we can compute the integral over the product space by first integrating with respect to one variable and then with respect to the other, in either order. Example Consider with Lebesgue measure, and . Then: Tonelli's Theorem Tonelli's Theorem is a variant of Fubini's Theorem for non-negative functions, without requiring integrability. Statement of Tonelli's Theorem Let and be σ-finite measure spaces, and let be -measurable. Then: 1. For almost all , the function is -measurable. 2. The function is -measurable. 3. Similar statements hold with the roles of and interchanged. Infinite Product Measures The concept of product measures extends naturally to countably many measure spaces. Definition Let be a sequence of measure spaces for . The product measure is defined on the product σ-algebra . For measurable rectangles where and for all but finitely many , the product measure is defined as: where the product is well-defined because all but finitely many terms equal 1. Existence and Uniqueness The existence and uniqueness of infinite product measures is guaranteed under certain conditions, most importantly when all component measures are probability measures or when they are σ-finite and the product is also σ-finite. Kolmogorov Extension Theorem The Kolmogorov Extension Theorem is a fundamental result that guarantees the existence of stochastic processes with given finite-dimensional distributions, using infinite product measures. Applications Product measures have numerous applications across mathematics: Probability Theory In probability theory, product measures represent joint distributions of independent random variables. If and are independent random variables with distributions and , then their joint distribution is the product measure . Statistical Mechanics In statistical mechanics, product measures are used to model systems of independent particles, where the state space of the entire system is the product of the state spaces of individual particles. Multiple Integration Product measures provide the theoretical foundation for multiple integration in calculus and analysis. Exercises 1. Let with Lebesgue measure. Compute using Fubini's Theorem. 2. Prove that if and are probability measures, then is also a probability measure. 3. Let for and . Show that the iterated integrals and exist but are not equal. Explain why this does not contradict Fubini's Theorem. 4. Let be a measure space and with counting measure . Show that can be identified with copies of . 5. Let with Lebesgue measure, and let (the unit disk). Compute . 6. Prove that if and are measurable sets with or , then . 7. Let be Lebesgue measure on . Compute where . 8. Let with counting measure. Describe the product measure explicitly. 9. Let with Lebesgue measure. Find a measurable set such that the sections have measure zero for all , but . 10. Let be independent random variables, each uniformly distributed on . Using product measures, calculate the probability that .