We consider estimation and control in linear dynamical systems from the perspective of regret minimization. Unlike most prior work this area, we focus on problem designing causal state estimators controllers which compete against a clairvoyant noncausal policy, instead best policy selected hindsight some fixed parametric class. show that regret-optimal filters can be derived state-space form us...

We consider the problem of efficiently estimating size join a collection preprocessed relational tables from perspective instance optimality analysis. The running time optimal algorithms is comparable to minimum needed verify correctness solution. Previously, were only known when was small (as one component their linear in size). give an algorithm for all instances, including large, by removing...