Experiment 5A and 5B both use a greatly modified version of the famous Delayed Choice Quantum Eraser (DCQE) published in 2000 by Youn-Ho Kim et.al. [http://xxx.lanl.gov/pdf/quant-ph/9903047 Phys.Rev.Lett. 84 1-5 (2000).]
This experiment needs to use photons as the particle to create the signal and idler photon pair produced by the original experimental setup.
First, adjust the parameters (e.g., slit size, slit width, distance ds between the slits and the result screen, and wavelength of laser light) of the sixth experiment such that the wave and particle patterns found on the result screen (and recorded by R0) are deconflicted as much as possible (overlap as little as possible).
Secondly, run a sufficient number of particles (more is better) through the sixth experimental setup to generate an algorithm that, given an X value from a particle impact point on D0, computes the probability that any particle (photon) hitting this impact point on D0 would be part of a diffraction or wave pattern (was absorbed after transmission at BS1 or BS2) or part of a two-bar particle pattern (was captured by D3 or D4). [For a simple example. Look at the K nearest impact points to position x on D0, if N of those points (N≤K) were part of a wave (diffraction) then Pw = N/K]. If M= K-N of those points were part of a particle pattern, then Pp = M/K where N+M = K and Pw + Pp = 1.
Next, test and optimize the algorithm accuracy.
Use the algorithm to predict (compute probability of) whether a given impact’s point x value is likely to be part of a diffraction pattern (will be absorbed after transmission at BS1 or BS2) or part of a two-bar pattern (will be recorded by D3 or D4) [The computation must be fast enough to make this prediction before the particle reaches either BS1 or BS2 (or at least start the deterministic computation process before then).