I have already commented on how the above problem could be solved using the linear algebra of a hyperspace of even dimension and a discrete version of the Legendre Transformation involving higher derivatives of the intercept with respect to the gradient in 2D space by the process of a descent to the lower dimension and ascent to the higher dimension without loss of information. It is almost self evident that a successor prime must depend exclusively on its antecedents. [the number two  is in an excluded class of its own, itself].
The difficulty I have with the Riemann zeros is that I do not understand their significance in relationship to the sequence: [3, 5, 7, 11, ... ]. Perhaps if the Riemann zeros were computed in a for series where every even number was suppressed, then this would clarify matters.
I have an e-mail address, if someone wants to communicate with me on the matter of hyperspaces, Legendre Transformations and dimension reductions. This is: email@example.com