Stability analysis of stochastic programs with second order dominance constraints (Q2434984): Difference between revisions
From MaRDI portal
Removed claim: reviewed by (P1447): Item:Q260631 |
Changed an Item |
||
Property / reviewed by | |||
Property / reviewed by: Ioan M. Stancu-Minasian / rank | |||
Normal rank |
Revision as of 10:45, 12 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Stability analysis of stochastic programs with second order dominance constraints |
scientific article |
Statements
Stability analysis of stochastic programs with second order dominance constraints (English)
0 references
3 February 2014
0 references
The authors consider the following stochastic program with stochastic second-order dominance (SSD) constraints: \[ \begin{aligned} &\underset{x}{\min} \,\,\mathbb{E}_{P}\left[ f\left( x,\xi \left( \omega \right) \right) \right] \\ &\text{s.t.}\;G\left( x,\xi \left( \omega \right) \right) \succeq _{2}Y\left( \xi \left( \omega \right) \right) , \\ & x\in X, \end{aligned} \] where \(X\) is a nonempty compact subset of \(\mathbb{R}^{n}\), \(\xi :\Omega \rightarrow \Xi \) is a vector of random variables defined of the probability space \(\left( \Omega ,\mathcal{F},P\right) \;\)with the support set \(\Xi \subset \mathbb{R}^{m}\), \(f,G: \mathbb{R}^{n}\times \Xi \rightarrow \mathbb{R}\) are Lipschitz continuous functions and for every \(\xi \in \Xi \), \(G\left( \cdot ,\xi \right) : \mathbb{R}^{n}\rightarrow\mathbb{R}\) is concave; \(Y\left( \xi \left( \omega \right) \right) \) is a random variable, \(\succeq _{2}\) is the relation of domination between two random variables,\ and \(\mathbb{E}_{P}\left[ \cdot \right] \) denotes the expected value with respect to the probability \((P)\) distribution of \(\xi \). \ It is assumed that \(f\) and \(G\) are \(P\)-integrable. The authors present a stability analysis of the SSD problem. They consider the case when the underlying probability measure \(P\;\)is approximated by a set of probability measures under pseudometric. By exploiting a result on error bounds in semi-infinite programming, the authors show under the Slater condition that the feasible solution set mapping is Lipschitz continous, the optimal solution set mapping is upper semicontinuous and the optimal value function is Lipschitz-like (calm). Also, the authors consider a special case when the probability measure \(P\) is approximated by an empirical probability measure and present a detailed analysis on the convergence of the optimal solution and the stationary point obtained from solving the sample average approximate optimization problems as the sample size increases. Finally, the analysis is extended to the stationary points.
0 references
second-order dominance
0 references
stochastic semi-infinite programming
0 references