Title:Generalized quantum fluctuation-dissipation identities

Abstract:In this article, we derive the fluctuation-dissipation theorem from Bayesian statistics in the language of quantum statistical mechanics, using density matrices representing the respective state of knowledge about the system. This theorem, which we have called the quantum-fluctuation-dissipation theorem (Q-FDT), allows us to obtain expectation identities between quantum observables depending on a continuous parameter $\gamma \in \mathbb{R}$. In this sense, we consider parameters contained in the observables (e.g. a perturbative parameter), or parameters such as Lagrange multipliers only, excluding parameters that modify the underlying Hilbert space. Using both canonical and grand canonical density matrices in the Q-FDT, we explore the resulting fluctuation-dissipation identities. In this sense, we found that some of the expectation identities in common use in traditional quantum statistics and quantum chemistry, such as the thermodynamical fluctuation-dissipation theorem, the Ehrenfest and the Hellmann-Feynman theorems, among others, are particular instances of the Q-FDT. Lastly, using a \textit{generalized} density matrix arising from a Maximum-Entropy procedure, we derive generalized fluctuation-dissipation identities: these generalized identities allow us to group all the previous cases in a unitary scheme.