Process tomography for systems in a thermal state

We propose a new method for implementing process tomography which is based on the information extracted from temporal correlations between observables, rather than on state preparation and state tomography. As such, the approach is applicable to systems that are in mixed states, and in particular thermal states. We illustrate the method for an arbitrary evolution described by Kraus operators, as well as for simpler cases such as a general Gaussian channels, and qubit dynamics.