[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: VMs: Follow-up
On Tuesday 22 June 2004 11:58, Diego Rybski wrote:
> 2) correlations depend on the order of the symbols. If you shuffle the
> symbols, the distribution remains, but the correlation structure vanishes.
> Correlations mean, that there is a dependence between the elements of the
> series. In numbers, a big is more probably follwed by another big, and a
> small by another small one. The often mentioned long-range correlations
> are a special kind of correlation. Unfortunately I have not yet found
> the references related to this topic. The correlation function decays like
> a power-law, thus the mean correlation length diverges. It is simple to
> create long-range correlated sequences with gaussian distribution, but
> even today it is a rather difficult task to create long-range correlated
> power-low distributed sequences. I know that it is possible for
> exponential distributed.
As you are aware, symbolic sequences are not good candidates to correlation
analysis because symbols do not have a quantity associated.
There are some ways round, though. I did some long correlation analysis of the
vms text using Voss' method of applying fft to binary "same-symbol" sequences
(and of other texts).
The reference is: Landini, G. Evidence of linguistic structure in the Voynich
Manuscript using spectral analysis. Cryptologia, October 2001, Volume XXV,
Number 4, (pp.275-295), 2001.
I can send you a pdf preprint if you cannot get the journal.
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying: