| markov {transition}+ | [Macro] |
Creates a Markov item stream. Each transition is a rule in the form:
({past}* -> {{outcome} | ({outcome} {weight})}+
The left side of the rule (the values preceding ->)
declare the ordered past outcomes that trigger the rule. Values
closer to the -> are more recent choices. The number of left-side values
determines the Markov order of the pattern; any number of values
may be specified but every transition rule should
have the same number of values.
Values to the right of the -> define
the set of outcomes. By default each outcome has the same probability of being selected;
to weight an outcome differently, specify it in a list together with its weight. For example,
(c -> a (b 2) c)defines a 1st order Markov transition rule whose outcomes A, B and C are triggered if C was the last value generated; B has twice the probability of being selected as A or C.
The next example
(q w a -> e r (d 3) g)describes a third-order transition rule, where A is the most recent outcome, and
(* x a -> foo bas zuz)demonstrates the "wildcard" * that matches any value in its position. The wildcard token allows multiple transitions with identical outcomes to be collapsed to a single rule. Finally,
(-> a (b .1) c)describes a 0th order Markov transition, equivalent to weighted random selection. Only one transition rule should be specified in a 0th order Markov pattern.
Example:
;;;
;;; first order Markov of Happy Birthday
;;;
(markov (c4 -> (c4 0.4) (d4 0.3) (f4 0.1) (g4 0.1) (c5 0.1))
(d4 -> (c4 0.75) (f4 0.25))
(e4 -> (c4 0.667) (d4 0.333))
(f4 -> (e4 0.6) (c4 0.2) (f4 0.2))
(g4 -> (f4 1.0))
(a4 -> (f4 1.0))
(c5 -> (a4 1.0)))
See Also:
graph, markov-analyze