“…Often, it is formulated as a numerical optimization problem—commonly a non‐convex one. The challenges of non‐convexity are dealt with, for instance, by omitting the non‐convex constraints [KNP07, BZK09, KMZ11, MZ13, ESCK16, ZCZ∗18] or by conservative convexification [Lip12, BCE∗13, CBK15, BCW17, HCW19]. In the former case, results can be invalid (violating local injectivity requirements) [BZK09, EBCK13], in the latter case valid solutions are excluded—in the worst case leading to an infeasible problem.…”