“…In the analysis of functional data, more complex dependence models arise, involving conditional distributions in abstract spaces. We refer to the reader to the recent contribution by Chaouch, Laib and Louani (2017), on kernel conditional mode estimation, from functional station-ary ergodic data, in the context of random elements in semi-metric abstract spaces (see also Ling, Liu and Vieu, 2017). This paper considers the problem of linear functional multiple regression estimation, when the response takes values in an abstract separable Hilber space H, and the regressors are operators on H. The temporal dependence of the errors is represented, in terms of an ARH(1) time series model.…”