Polynomial chaos expansion for sensitivity analysis of model output with dependent inputs

In this paper, we discuss the sensitivity analysis of model response when uncertain inputs are not independent one other. case, two different kinds indices can be evaluated: (i) that account for dependence/correlation an input or group with remainder and (ii) do dependence. We argue distinction applies to any global measure. present work, focus on estimation variance-based which based second-order moment interest. particular, derive new strategies computationally efficient methods assess them, rely polynomial chaos expansion. Several numerical exercises carried out demonstrate performance methods, including a drainage posterior its statistical calibration. • An method models dependent inputs. Bayesian expansions adapted several transformations. The generalise Arbitrary dependence structures accommodated. show analytical examples calibration example.