StatBalNormStream | SuperCollider 3.13.0 Help

Stream class implementing Charles Ames' method of statistical feedback. This class implements all the full possibilities of the method as presented in "Statistics and Compositional Balance" by Charles Ames. This includes a normalisation step. Charles Ame's original seems incorrect, so I've added my own normalisation that maintains the sum of statistics as precisely zero. There is a touch of controversy here; Charles has communicated to me personally that he doesn't consider computer music his field any more, so his comments on my correction are unavailable. At some point a deeper mathematical analysis should help to resolve this.

Some normalisation step is necessary for situations where a weight may be zero for some time. Otherwise, the statistics will force unrealistic catchup when that weight becomes non zero again. Note that the set of weights, heterogeneity and accenture of a given test can be updated at any point. You could even change the set of states as long as the number of them remained the same. See StatBalStream for the simplified version without any controversial normalisation. The Pstatbalnorm class is also provided for use with Patterns classes.

References (all Ames, Charles):

"Statistics and Compositional Balance" Perspectives of New Music 1989
"A Catalogue of Sequence Generators" Leonardo Music Journal 1992
"Quantifying Musical Merit" Interface: Journal of New Music Research 1992

Class Methods

.new

Arguments:

states	An array of the states to return
weights	An array of weightings, one for each state. Weights may be zero.
heterogeneity	any positive float. Whilst weightings will be approached in the long term, how quickly this occurs relative to serialism and the tolerance within which a running histogram of results fluctuates depends on the ratio of heterogeneity to accent (by default 1). Heterogeneity 0.0 gives fully deterministic results, like serialism. Increasing heterogeneity increases the shuffling of order of appearance of states, that is, less predictability in the short term. If heterogenity is really large then floating point problems will eventually occur.
length	Number of calls to next before returning nil for sequence end. Default is inf for forever.