Take yoru B and T cell, for example

The doc wants a reasonable demonstration of tissue concentrations of the various B cells under perturbed conditions.  This is exactly what you want because we have a vertical and can color in that surface, and get the 3D distribution.  Counting out this stuff is not bad for the computers, these are large events in stable tissue, and there is lots of hard data about cultures in tissue. We should be able to get a very good idea bout why the B cells we want are late to the ball game.

The computer simulator.

It is not really constrained except for the number of items in the color operator, and twenty is well within its range, It will chew through the equations, and step the operator.   Multiple paths fine work,  just set an artificial error bound, or put if then statements into your python code. The computer is simply doing step by step differential equations, using,essentially, e difference in differences ikind of thing, the the self sampling is implied from the color mix in the operator.. Time is removed, as is space, but there are mostly obvious verticals to give you surface.

Set B cell concentration as one color in the operator, give it the estimate count of operations it can do,. Then on transition, each of the finite, possible paths can calculate and return the various outcomes in the adjacent surface area. But the reaction with the cytokines and the various cells can be broken out separately. B cells will undergo a transaction matrix in each complete pass, that returns a deviation count, how often does this color have to adjust in the estimated transaction. So the model is simply counting patches of surface, as implied in the transaction matrices. One complete area of surface can undergo these transactions in one step of the model.

One transaction in the model  is the cell cytokine reaction.    How many of those interactions will generally keep one unit of surface in balance.   The model really gives the missing cytokine missed, that is the error term carried in the color operator.    Your model ends up constantly having to deposit the one color that is imbalanced, until you can adjust some conditionals.  This is essentially an interactive path engine. You look fo the coloring paths that merging back on themselves, there is the problem and solution, the congested path.

The surface is the boundary point in that at that position, skin or tissue is as usual.  This is not so, and the conversion of tissue into food happens at each step of the operator, but outside the operator. And any effect of tissue damage has to be treated in the adjustments to B and T cell probabilities, at the step.  But this is the boundary condition, there is a boundary, the model does not have a null at infinity. There is no kinetic energy inside the operator, but it is implied in the count deviations. The researchers can make the obvious transformation to a vertical.

A specific cultured B cell inhaled will attack the spike and give the rest of the system a chance.