Characteristic Functions
The characteristic function of a random variable is the expectation of a complex exponential, the Fourier transform of its distribution; it always exists, determines the distribution uniquely, and converts independence into multiplication.
Definition
The characteristic function of a random variable is the expected value of the complex exponential of the variable times a real argument, equivalently the Fourier transform of its distribution, which exists for every distribution and determines it uniquely.
Scope
The topic covers the definition and elementary properties of the characteristic function, its uniqueness and inversion theorems, the factorization of the characteristic function of a sum of independent variables, the relation between smoothness of the function and moments of the distribution, Bochner's characterization of which functions are characteristic functions, and Levy's continuity theorem linking pointwise convergence to convergence in distribution.
Core questions
- Why does every distribution possess a characteristic function when moments may not exist?
- How does the characteristic function determine and allow recovery of the distribution?
- Why does the characteristic function of a sum of independent variables factorize?
- How does convergence of characteristic functions relate to convergence of distributions?
Key concepts
- Fourier transform of a measure
- uniqueness and inversion
- Levy continuity theorem
- Bochner's theorem
- moments from derivatives
Key theories
- Uniqueness and inversion
- Distinct distributions have distinct characteristic functions, and an inversion formula recovers the distribution from its characteristic function, so the transform is a faithful and invertible encoding of the law of a random variable.
- Levy continuity theorem
- A sequence of distributions converges in distribution if and only if their characteristic functions converge pointwise to a function continuous at the origin, which is then the characteristic function of the limit; this is the standard route to limit theorems.
- Factorization for sums of independent variables
- Because expectation factorizes over independent variables, the characteristic function of a sum of independent variables is the product of their characteristic functions, replacing convolution of distributions with ordinary multiplication.
Clinical relevance
Characteristic functions are the principal tool for proving the central limit theorem and other limit laws, they make sums of independent random variables analytically tractable in fields from signal processing to actuarial science, and their inversion underlies numerical methods for option pricing where the characteristic function is known in closed form.
History
Characteristic functions were used by Laplace and Cauchy and were made the systematic instrument of probability by Paul Levy, whose continuity theorem turned the proof of limit theorems into the study of pointwise convergence of these transforms; Bochner characterized exactly which functions arise this way.
Key figures
- Paul Levy
- Aleksandr Lyapunov
- Salomon Bochner
- Eugene Lukacs
Related topics
Seminal works
- feller1971
Frequently asked questions
- How does the characteristic function differ from the moment generating function?
- The characteristic function uses an imaginary exponent and therefore exists for every distribution, whereas the moment generating function uses a real exponent and may fail to exist for heavy-tailed distributions; the characteristic function is the more robust tool.
- Why is convergence checked only at the origin in the continuity theorem?
- Continuity of the limit at the origin rules out an escape of probability mass to infinity, ensuring the limiting function is itself a genuine characteristic function rather than that of a defective distribution.