Driftsfunktion

Driftsfunktion, or drift function in English, is a mathematical concept used to describe the deterministic part of the evolution of a system modeled by stochastic or dynamical equations. It specifies the expected instantaneous change of the state variable, separating predictable trends from random fluctuations.

In continuous-time stochastic processes, the drift function μ governs the deterministic trend in a diffusion process. A common framework is the Itô stochastic differential equation dX_t = μ(X_t, t) dt + σ(X_t, t) dW_t, where X_t is the state, μ: R^n × R_+ → R^n is the drift function and σ: R^n × R_+ → R^{n×m} is the diffusion coefficient. The drift describes where the process tends to move in the absence of noise, while the diffusion term captures random variation.

In deterministic dynamical systems, the drift corresponds to the vector field f(x, t) in the ordinary differential equation dx/dt = f(x, t). There, the drift dictates the system’s trajectory in the absence of stochastic disturbance.

Examples include the Ornstein–Uhlenbeck process with μ(x) = −θ x, illustrating a restoring drift toward zero, and geometric Brownian motion with μ S_t as the drift term for asset prices. In finance, the drift represents the expected return under a chosen measure.

Properties relevant to drifts include regularity conditions (such as Lipschitz continuity and linear growth) that ensure existence and uniqueness of solutions. Estimation from data often involves separating drift and diffusion, using high-frequency observations or regression-based methods.

See also: stochastic differential equation, Itô calculus, diffusion process, drift coefficient.