Exact bounds for the rate of convergence in general stochastic approximation procedures (Q4393907)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 1161998
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Exact bounds for the rate of convergence in general stochastic approximation procedures |
scientific article; zbMATH DE number 1161998 |
Statements
Exact bounds for the rate of convergence in general stochastic approximation procedures (English)
0 references
18 November 1998
0 references
almost sure convergence
0 references
nonlinear difference equation
0 references
weighted sums of independent random variables
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0.8475247025489807
0 references
0.8275191187858582
0 references
The authors consider the nonlinear difference equation \[ X_n= X_{n-1}-a_n f(X_{n-1}) +b_n {\mathcal V}_n, \quad n\geq 1, \quad X_0 \in\mathbb{R}, \] where \(a_n>0\), \(\lim_{n\to \infty} a_n=0\), \(\sum^\infty_{n=1} a_n= \infty\), \(\{b_n, n\geq 1\} \subset \mathbb{R}\), and \(\{{\mathcal V}_n\), \(n\geq 1\}\) is a sequence of random variables and the equation \(f(x) =0\) has the solution \(\Theta\). Assume \(X_n\to \Theta\) a.s. as \(n\to \infty\). The authors determine a sequence of numbers \(\{c_n\), \(n\geq 1\}\) such that \[ \limsup_{n\to \infty} c_n | X_n-\Theta |=1 \text{ a.s.} \] Moreover, an almost sure representation for the deviation \(X_n- \Theta\) by weighted sums of independent random variables is also presented.
0 references