1: \begin{abstract}{Design optimization under uncertainty is notoriously difficult 

2: when the objective function is expensive to evaluate. State-of-the-art techniques, 

3: e.g, stochastic optimization or sampling average approximation, fail to learn 

4: exploitable patterns from collected data and require an excessive number of 

5: objective function evaluations. There is a need for techniques that alleviate 

6: the high cost of information acquisition and select sequential simulations 

7: optimally. In the field of deterministic single-objective unconstrained global 

8: optimization, the Bayesian global optimization (BGO) approach has been relatively 

9: successful in addressing the information acquisition problem. BGO builds a 

10: probabilistic surrogate of the expensive objective function and uses it to 

11: define an information acquisition function (IAF) whose role is to quantify 

12: the merit of making new objective evaluations. Specifically, BGO iterates 

13: between making the observations with the largest expected IAF and rebuilding 

14: the probabilistic surrogate, until a convergence criterion is met. In this 

15: work, we extend the expected improvement (EI) IAF to the case of design 

16: optimization under uncertainty wherein the EI policy is reformulated to 

17: filter out parametric and measurement uncertainties.  To increase the 

18: robustness of our approach in the low sample regime, we employ a fully 

19: Bayesian interpretation of Gaussian processes by constructing a particle 

20: approximation of the posterior of its hyperparameters using adaptive 

21: Markov chain Monte Carlo. We verify and validate our approach by solving 

22: two synthetic optimization problems under uncertainty and demonstrate it 

23: by solving the oil-well-placement problem with uncertainties in the 

24: permeability field and the oil price time series.}

25: \end{abstract}

26: