Two definitions of maximally $ψ$-epistemic ontological model and preparation non-contextuality

Abstract: An ontological model is termed as maximally $\psi$-epistemic if the overlap between any two quantum states is fully accounted for by the overlap of their respective probability distributions of ontic states. However, in literature, there exists the two different mathematical definitions (termed here as 1M$\psi$E and 2M$\psi$E) that capture the equivalent notion of maximal $\psi$-epistemicity. In this work, we provide three theorems to critically examine the connections between preparation non-contextuality and the aforementioned two definitions of maximal $\psi$- epistemicity. In Theorem-1, we provide a simple and direct argument of an existing proof to demonstrate that the mixed state preparation non-contextuality implies the first definition of maximal $\psi$-epistemicity. In Theorem-2, we prove that the second definition of maximal $\psi$-epistemicity implies pure-state preparation non-contextuality. If both the definitions capture the equivalent notion of maximal $\psi$-epistemicity then from the aforementioned two theorems one infers that the mixed-state preparation non-contextuality implies pure-state preparation non-contextuality. But, in Theorem-3, we demonstrate that the mixed-state preparation non-contextuality in an ontological model implies pure-state contextuality and vice-versa. This leads one to conclude that 1M$\psi$E and 2M$\psi$E capture inequivalent notion of maximal $\psi$-epistemicity. The implications of our results and their connections to other no-go theorems are discussed.