Some random paths with angle constraints (Q2041791)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Some random paths with angle constraints |
scientific article |
Statements
Some random paths with angle constraints (English)
0 references
23 July 2021
0 references
Starting from random elements \(U_{1},U_{2},\ldots\) one can consider the random walk \(S(n)=\sum_{i=1}^{n}U_{i}\) and then investigate the behavior of the piece-wize linear interpolation of this walk, i.e., \(X_{t}^{(n)}=S(\left \lfloor nt\right\rfloor )+(nt-\left\lfloor nt\right\rfloor )U_{\left\lfloor nt\right\rfloor +1},0\leq t\leq 1\). In the case of independent increments and finite variances, the asympotic behaviour is well known and leads to Brownian motion. In the present paper the authors consider the following construction.\ Let \((\Theta _{i})_{i\geq 2}\) denote i.i.d. real random variables and let \(U_{1}\) denote a r.v. uniformly distributed on the unit circle. For \(j=2,3,\ldots\) define \(U_{j}=\exp (i\Theta _{j})U_{j-1}\). In this case the \(U_{i}\) are uniformly distributed on the unit circle, but they are not independent in general. Clearly for \(t\in \left[ 0,1\right] \), \(X_{t}^{(n)}\) takes values in \(\mathbb R^{2}\) . The authors prove the following Theorem 1. If the \(\Theta _{i}\) are uniformly distributed in \(\left[ -\alpha ,\alpha \right] \), where \(\alpha \in (0,\pi ]\), then as \(n\rightarrow \infty \), \(n^{-1/2}X_{n}\Longrightarrow \sigma (\alpha )B^{(2)}\), where \(\Longrightarrow \) denotes weak convergence of probability measures, \(B^{(2)}\) denotes the standard bivariate Brownian motion and \(\sigma ^{2}(\alpha )=0.5(1+(\sin \alpha )/\alpha )/(1-(\sin \alpha )/\alpha ))\). In the case \(\alpha =2\), this is Donsker's theorem. The proof of the theorem involves tightness, relative compactness, convergence of finite dimensional laws and a new central limit theorem for dependent random varables. The authors consider also the case where the \(\Theta _{i}\) may vary with \(n\). In section 3, the construction starts with \(U_{1,n}\) uniformly drawn from the unit circle and then \(U_{j,n}=\exp (i\Theta _{j,n})U_{j-1,n},j\geq 2,n\geq 1\), where \(\Theta _{j,n}\) are real random variables. In section 4, the authors introduce a type of Markovian condition on the \(\Theta _{j,n}\). The limit theorems are now of a different (interesting) nature.
0 references
Brownian motion
0 references
central limit theorem for dependent variables
0 references
curvature
0 references
random angles
0 references
random walk
0 references
0 references