The complex-valued function phi_X(t) = E[exp(i t X)] defined for a random variable X that encodes its distribution via the expectation of a complex exponential kernel.

Principle

Represent a probability distribution by its moments under complex exponentials so that convolution and translation correspond to multiplication and phase shifts of the function.

Demonstration

For a normal random variable with mean mu and variance sigma^2, the characteristic function equals exp(i mu t − (sigma^2 t^2)/2), which compactly encodes mean and variance information.

Misapplication

Confusing this function with the real-valued moment-generating function and applying moment-based reasoning where the moment-generating function does not exist can lead to invalid conclusions.

Consequence

Uniquely determines the underlying distribution and turns sums of independent variables into products of characteristic functions, facilitating limit theorems and distributional analysis.

Reversal

The distribution (cdf) is the inverse transform object that recovers probabilities from the characteristic function; the two perspectives are dual.

Boundary

The characteristic function always exists for any probability law; practical inversion or estimation in continuous multivariate settings requires careful analytic or numerical techniques.

Semantic Tension

Versus moment-generating functions: characteristic functions always exist but take complex values, while moment-generating functions are real-valued where they exist but may not be defined for all distributions.

Synthesis

A universal complex-valued transform of a probability distribution obtained by taking the expectation of an exponential kernel; it encodes distributional information and converts convolution into multiplication.