Section 5 Maxwell’s distribution
5.1
Density: Radon-Nikodym derivative
5.5
Bochner’s theorem for characteristic functions
A function g is the characteristic function of a probability density distribution function f, if and only if g satisfies
(1) g is continuous,
(2) g(0) = 1
(3) g is positive definite: that is, for any positive integer n and complex numbers
{c_k} (k = 1, ..., n) \sum_{i,j} c_i^* g(z_i - z_j) c_j >= 0.
This theorem may be used, e.g., to model autocorrelation functions in nonequilibrium statistical mechanics.
Proof: A proof may found in, e.g.,
K Yosita, Functional Analysis (6th ed) (Springer 1980) p346
W Feller, An introduction of probability theory and its applications (Wiley, 1966, 1971) Vol. II Section XIX 2 p620-.