Gexpectation

Gexpectation, or G-expectation, is a nonlinear, sublinear expectation operator developed by Shige Peng as part of the theory of sublinear expectations to model uncertainty in probability and finance. It generalizes the classical linear expectation by allowing the value to reflect ambiguity in the underlying probability law, particularly volatility. The operator, denoted Ê, is defined on a space of random variables and satisfies monotonicity, constant preservation, sub-additivity, and positive homogeneity.

Under G-expectation, random variables may have a distribution called the G-normal distribution, and time evolution is described by a G-Brownian motion, a process with stationary, independent increments whose variance is itself uncertain. The uncertainty is encoded in a set Γ of possible volatility matrices, and the function G is defined by G(A) = 1/2 sup_{γ∈Γ} tr(γ γ^T A).

A central representation is that Ê[X] equals the worst-case expectation over a family of probability measures that reflect volatility ambiguity. Conditional G-expectation Ê_t extends this to a time-consistent, adapted framework, akin to a conditional expectation under model uncertainty.

Analytically, the G-expectation is linked to fully nonlinear partial differential equations: for a suitable function φ, the function u(t,x) = Ê[φ(x + B_T − B_t)] solves the nonlinear heat equation ∂_t u + G(D^2 u) = 0 with terminal condition u(T,x) = φ(x). This connection provides a bridge between stochastic analysis and nonlinear PDE theory.

Applications of G-expectation include robust pricing of financial derivatives, risk assessment under volatility uncertainty, and the study of stochastic processes under model ambiguity. The framework has influenced probability, PDEs, and mathematical finance since its introduction in the early 2000s.