We introduce a novel hybrid quantum-classical algorithm for the near-term computation of expectation values in quantum systems at finite temperatures.
This is based on two stages: on the first one, a mixed state approximating a fiducial truncated density matrix is prepared through Variational Quantum Eigensolving (VQE) techniques;
this is then followed by a reweighting stage where the expectation values for observables
of interest are computed. These two stages can then be iterated again with different hyperparameters to achieve arbitrary accuracy.
Resource and time scalability of the algorithm is discussed with a near-term perspective.