With his latest publication in the New Journal of Physics (DOI: 10.1088/1367-2630/aa54ab), George has proposed experimental tests which can further constrain the extent to which the wavefunction can be epistemic (and thus representing only a lack of knowledge as to the true physical state of reality).

There exist thought experiments capable of showing that the theory of quantum mechanics is entirely ontic (a true part of physical reality). But these require the use of an infinite number of states and completely error-free experiments, neither of which are accessible to a real experiment. While it may never be possible to experimentally demonstrate that quantum mechanics is entirely ontic, the extent to which the wavefunction can be epistemic can be bounded by a proportionality constant k. If k = 1, the quantum state can be interpreted as a state of knowledge. If k=0, it must be thought of as a part of reality (should such exist!).

Existing experiments suggest that k<1, but finding a set of states and measurements which can be used to further bound the epistemic-fraction of reality is not easy. By setting up the search as an alternating convex problem, numerical algorithms for the optimal set of states and measurements (for given Hilbert spaces) are less likely to become trapped in local optima, thereby enabling more efficient searches and better results. Using this new description of the optimisation problem George has been able to find states and measurements which provide better theoretical bounds on the epistemic-fraction and therefore demonstrate improved epistemic-bounds in future experiments.

For the first time George has also considered the usefulness of mixed states for performing such tests of the epistemic-fraction and shown that using mixed states could further push down the epistemic-fraction which can be seen in experiments, even though mixed states are usually thought of as less “quantum" and less useful for experiments.