Looks familiar, also validated true for +E,X,N. Here's a new formula for ya, related to some work I'm doing with T right now. Given a valid cell (E,X,N), to predict the next X in it's "chain",
next x = sqrt((2 * n * b) - e)
This let's you step up X, where first cell's B becomes the new cell's A. There are many cells that have multiple "chains", going to work on that tomorrow.
Not so much predicting, as it is deriving. Not sure how you would get the previous cell's X other than,
previous x = x - (sqrt((2 * n * b) - e) - x)
I'm quite sure that this negates some of my previous statements on T. If you can find one valid cell in an (E,N), you can search between the valid cell and the next cell in it's chain, which should reveal all of the other chains, thus unlocking the entire cell.