The effect of the Markov chain condition on the prediction of extreme values
β Scribed by A. Naess
- Publisher
- Elsevier Science
- Year
- 1984
- Tongue
- English
- Weight
- 883 KB
- Volume
- 94
- Category
- Article
- ISSN
- 0022-460X
No coin nor oath required. For personal study only.
β¦ Synopsis
The purpose of this paper is to study the effect of correlation on extreme value estimates of a narrow-band, stationary Gaussian process. This is done by introducing a Markov chain condition on the sequence of peak values of the process. It is shown that this leads to extreme value estimates that are smaller than those obtained by standard order statistics. Under specified conditions, an explicit formula is obtained for the extreme value estimates resulting from the introduction of the Markov chain condition. Due to the close connection between extreme value statistics and first passage time statistics, a discussion of the impact of the results on the first passage time problem is also given. ~3 So IR(r)I dr<oo, one then defines the (one-sided) spectral dcnsity function as {!So l R(r) coscordr, ~o~0 S(to) = <0J" (3)
π SIMILAR VOLUMES
In this paper we study the question of the conditions under which a hidden Markov chain itself exhibits Markovian behaviour. An insightful method to answer this question is based on a recursive ΓΏltering formula for the underlying chain.
Let Q be a convex solid in R", partitioned mto two volumes u and t' by an area s. We show that s > min(u, o)/diam Q, and use this inequality to obtain the lower bound n -'/' on the conductance of order Markov chains, which describe nearly uniform generators of linear extensions for posets of size n.