The decoupling technique was originally developed for information-theoretical purposes. It describes the conditions under which the correlations in a bipartite state disappear if one part undergoes an evolution separated from the other. In the past years there has been enormous progress in understanding the foundations of statistical mechanics from first principles of quantum mechanics. By use of the decoupling technique we are able to reproduce and generalize major results of this development and to approach open problems. As a first application of the decoupling technique we generalize the result of [Popescu et al., Nat. Phys. 2, 754-758 (2006)] about the apparent validity of the postulate of equal a priori probabilities to states which may be correlated to a reference. We express it in a form which allows to apply recent results about random two-qubit interactions. We give a criterion for the apparent validity of the postulate which is tight up to differences between different entropy measures. Similarly, we generalize the result of [Linden et al., Phys. Rev. E 79, 061103 (2009)] about the independence of the temporal average of a quantum mechanical system of its initial state to initial states which may be correlated to a reference. We develop a criterion for whether a quantum-mechanical system has “forgotten” about its initial state which is tight up to differences between different entropy measures. We find that comparing two local entropies of just one particular state tells us whether generic initial states of the system have already evolved to the same state or not. After developing new bounds on the times which are necessary for entropy changes, we are able to provide lower bounds on the times which are necessary for a system to become independent of its initial state. We discover an intimate connection between a system becoming independent of its initial state and a loss of an observer’s knowledge about the state of the system. As a further application of the decoupling technique we find sufficient conditions under which a system stays close to its initial state for all times, thereby extending the result of [Gogolin et al., Phys. Rev. Lett. 106, 040401 (2011)].
Modern quantum engineering techniques enabled successful foundational tests of quantum mechanics. Yet, the universal validity of quantum postulates is an open question. Here we propose a new theoretical framework of Q-data tests, which recognizes the established validity of quantum theory, but allows for more general—‘post-quantum’—scenarios in certain physical regimes. It can accommodate a large class of models with modified quantum wave dynamics, correlations beyond entanglement or general probabilistic postulates. We discuss its experimental implementation suited to probe the nature of strong nuclear interactions. In contrast to the present accelerator experiments, it shifts the focus from high-luminosity beam physics to individual particle coherent control.