For:
d($t_a$) doors (not already picked or opened at time $t_a$ ), $c_{tot}$ cars (initially), p($t_b$) picks (which depends on d($t_b$) == doors available at time of choosing picks),
o($t_c$) doors opened (revealing mixtures of cars and/or nothing, thus making accessible cars a 'function' of o($t_c$) =>c$(t_{c})$, and
s($t_d$) switched (which of course is constrained by accessible doors d($t_d$) not already picked or opened).
Where all $t_*$ are in the same Phase of Game play T=($t_a,t_b,t_c,t_d$) time order of updating event; where t_a<t_b<t_c<t_d , and a set of game play actions Q$_n$ = {$d,p,o,s$},
are related by
W$_n$ = Q$_n$R$_n$T == mapping logical$^{[1]}$ orders of actions onto the time order for a give Phase iteration n .
How would one go about constructing a statistical model for this for some phase n with rules$^{[2]}$? Any help clearing this question for readability, and disambiguate its motifs, is much appreciated.
$^{[1]}$ Logical: actions in orders which make sense. Including: no repeat actions during a phase, no run away addition of doors, time order of update says nothing about the amount an action/event is changed (or not change) simply a causal connective.
$^{[2]}$ Rules:
- Game is over when a floor on switching possibilities is met ( this does not require switching ), c = 0, and/or a limit on number of phases is met.