“…Unlike other standard results available in the literature, our result holds at the exact proximal-gradient point ŷ(k,0) , of which we have no knowledge if τ > 0, rather than at the point ỹ(k,0) that is actually computed in the algorithm. This is because such an upper bound is unlikely to hold at an inexact point computed according to (28), as discussed in [36, section 3.2]. Indeed, even if it is possible to obtain an upper bound for the distance between the exact point ŷ(k,0) and its approximation ỹ(k,0) (see inequality (16) in lemma 5), a similar relation between the subgradients at ŷ(k,0) and those at ỹ(k,0) can not be proved.…”