CVaR-constrained stochastic programming reformulation for stochastic nonlinear complementarity problems (Q457217)

The authors consider the stochastic complementarity problem (SCP), i.e., the problem of finding a vector \(x\in \mathbb{R}^{n}\) such that \[ x\geq 0, \;F\left( x,\xi \left( \omega \right) \right) \geq 0, \;x^{T}F\left( x,\xi \left( \omega \right) \right) =0, \;\omega \in \Omega \text{ a.s., }\tag{1} \] where \(F: \mathbb{R}^{n}\times \Xi \rightarrow \mathbb{R}^{n}\) is a random vector-valued function and \(\xi =\xi \left( \omega \right) \;\)is an \(m\)-dimensional random vector\(\;\)defined on the probability space \( \left( \Omega ,\mathcal{F},\mathcal{P}\right) \) and supported on the set \( \Xi \subset \mathbb{R}^{m}\). The stochastic complementarity problem (1) is reformulated as a problem that minimizes the CVaR (conditional value-at-risk) of the loss function \(\left\| \Phi \left( x,\xi \right) \right\| \), namely \[ \min_{x\in \mathbb{R}^{n}}\text{CVaR}_{\alpha }\left( \left\| \Phi \left( x,\xi \right) \right\| \right) , \] where the CVaR of a random loss \(Z\left( x,\xi \right) \) at level \(\alpha \) is defined as \[ \text{CVaR}_{\alpha }\left( Z\left( x,\xi \right) \right) :=\left\{ \beta + \frac{1}{\alpha }\mathbb{E}\left[ \left( Z\left( x,\xi \right) -\beta \right) _{+}\right] \right\}. \] By applying the smoothing technique and the penalty method, they propose a penalized smoothing sample average approximation algorithm to solve the CVaR-constrained stochastic programming. The authors show that the optimal solution of the penalized smoothing sample average approximation problem converges to the solution of the corresponding nonsmooth CVaR-constrained stochastic programming problem almost surely. Finally, they report some preliminary numerical test results.