Monday, April 22, 2019

Back to sequences

Huffman encode one or more complete sequences, and you get a compression graph which has rank equal to channel  size, approximately.   This works for short sequences, 256 samples, with a most stable precision value, minimize round off error.  That has a mapping to the number of check out counters, in the abstract, we see the encoding graph root being of rank size, the number of stable Poisson queues the counter keeps.  A lot of this stuff becomes trivially true when examining systematic node by node alterations.  There will always be a graph shape preserved in any transformation. That has to be true because maximum entropy is not a multiple equilibrium.  But the variations from typical still captured on the uncertainty of quantization, the bit error.

This is me hand waving, I am a blogger.

No comments: