Asymptotic Global Behavior for Stochastic Approximation and Diffusions with Slowly Decreasing Noise Effects: Global Minimization via Monte Carlo (Q4722928)
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 3996751
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Asymptotic Global Behavior for Stochastic Approximation and Diffusions with Slowly Decreasing Noise Effects: Global Minimization via Monte Carlo |
scientific article; zbMATH DE number 3996751 |
Statements
Asymptotic Global Behavior for Stochastic Approximation and Diffusions with Slowly Decreasing Noise Effects: Global Minimization via Monte Carlo (English)
0 references
1987
0 references
global optimization
0 references
stochastic approximation
0 references
simulated annealing
0 references
Monte Carlo methods
0 references
0.7792266607284546
0 references
0.7775408625602722
0 references
0.7768937945365906
0 references
The asymptotic behavior of the systems \(X_{n+1}=X_ n+a_ nb(X_ n,\xi_ n)+a_ n\sigma (X_ n)\psi_ n\) and \(dy=\bar b(y)dt+\sqrt{a(t)}\sigma (y)dw\) is studied, where \(\{\psi_ n\}\) is i.i.d. Gaussian, \(\{\xi_ n\}\) is a (correlated) bounded sequence of random variables and \(a_ n\approx A_ 0/\log (A_ 1+n)\). Without \(\{\xi_ n\}\), such algorithms are versions of the ''simulated annealing'' method for global optimization. When the objective function values can only be sampled via Monte Carlo, the discrete algorithm is a combination of stochastic approximation and simulated annealing. Our forms are appropriate. The \(\{\psi_ n\}\) are the ''annealing'' variables, and \(\{\xi_ n\}\) is the sampling noise. For large \(A_ 0\), a full asymptotic analysis is presented, via the theory of large deviations: Mean escape time (after arbitrary time n) from neighborhoods of stable sets of the algorithm, mean transition times (after arbitrary time n) from a neighborhood of one stable set to another, approximate asymptotic invariant measures, and location of the values of \(\{X_ n\}\) or y(\(\cdot)\), the case where \(Eb(x,\xi)=\bar b(x)\) is the (negative) of a gradient of a function \(\bar B(x)\), and application to global function minimization via Monte Carlo methods.
0 references