Talk:Universal code (data compression)

Rice codes = Rice coding? --Abdull 16:00, 9 March 2006 (UTC)[reply]

symbol of golden ratio

why is gamma used for the golden ratio instead of phi? --WhiteDragon 17:43, 12 May 2006 (UTC)[reply]

Is the Fibonacci coding asymptotically optimal or Byte Coding

I think that there is a mistake in this page. as far as I know the Fibonacci coding method is universal yet it is not asymptotically optimal.

How come? Honnza (talk) 13:19, 8 December 2009 (UTC)[reply]

If Fibonacci coding is not asymptotically optimal. The same sources say that even all finite higher orders of Fibonacci are also not optimal for the same reason. But BYTE CODING where you code all nibbles in base 15 and use the last nibble as 4 ones for a stop code will at one point always use more bits than a higher order Fibonacci code that use 8 ones for a stop. So therefore the BYTE_CODING is not optimal for the same reason as Fibonacci

http://www.ics.uci.edu/~dan/pubs/DC-Sec3.html

"Fraenkel and Klein prove that the Fibonacci code of order 2 is universal, with c1=2 and c2=3 [Fraenkel and Klein 1985]. It is not asymptotically optimal since c1>1. Fraenkel and Klein also show that Fibonacci codes of higher order compress better than the order 2 code if the source language is large enough (i.e., the number of distinct source messages is large) and the probability distribution is nearly uniform. However, no Fibonacci code is asymptotically optimal." 99.69.170.177 (talk) 00:39, 24 December 2009 (UTC)[reply]