Beyond thermodynamic uncertainty relations: nonlinear response, error-dissipation trade-offs, and speed limits

From a recent geometric generalization of thermodynamic uncertainty relations (TURs) we derive novel upper bounds on the nonlinear response of an observable of an arbitrary system undergoing a change of probabilistic state. Various relaxations of these bounds allow to recover well known bounds such as (strengthenings of) Cramer-Rao's and Pinsker's inequalities. In particular we obtain a master inequality, named symmetric response intensity relation, which recovers several TURs as particular cases. We employ this set of bounds for three physical applications. First, we derive a trade-off between thermodynamic cost (dissipated free energy) and reliability of systems switching instantly between two states, such as one-bit memories. We derive in particular a lower bound of 2.8k (B) T per Shannon bit to write a bit in such a memory, a bound distinct from Landauer's one. Second, we obtain a new family of classic speed limits which provide lower bounds for non-autonomous Markov processes on the time needed to transition between two probabilistic states in terms of a thermodynamic quantity (e.g. non-equilibrium free energy) and a kinetic quantity (e.g. dynamical activity). Third, we provide an upper bound on the nonlinear response of a system based solely on the 'complexity' of the system (which we relate to a high entropy and non-uniformity of the probabilities). We find that 'complex' models (e.g. with many states) are necessarily fragile to some perturbations, while simple systems are robust, in that they display a low response to arbitrary perturbations.