Talk:Arithmetic coding/As a generalized change of radix: Difference between revisions
Talk:Arithmetic coding/As a generalized change of radix (view source)
Revision as of 21:05, 28 January 2016
, 8 years ago→Goof?
No edit summary |
(→Goof?) |
||
Line 39:
::::::: But, I guess that answers my question about why people are using cumulative frequencies in the encoding implementations: (a) the algorithm itself is basically useless, so no one should care about using it, and (b) the description in wikipedia uses an example where cumulative frequencies are easily confused with the letter encoding values.
::::::: All in all, it's a mess. --[[User:Rdm|Rdm]] ([[User talk:Rdm|talk]]) 19:57, 28 January 2016 (UTC)
:::::::: The example used on Wikipedia is indeed very confusing. Now I understand your point; you're suggesting using the corresponding byte values in encoding (or the alphabetical positions for alphabetical strings), instead of the cumulative frequencies. I'm not sure if this is better or worse, but if you choose to do that in the encoding process, you also have to do the same in the decoding process, otherwise you will get different results. For example, "ABRACADABRA" encoded with positional values, with result into "4860830 * 10^4" (same output as J). If we decode it back, using cumulative frequency instead, we'll get "AARAAAARAAA", which is incorrect.
:::::::: After all, the Wikipedia page seems to be contradictory, as it mentions both mapping the string to digits (not saying which digits), and bellow referring to <math>\scriptstyle C_i</math> as the cumulative frequency of the current symbol. I agree, it's a mess... --[[User:Trizen|Trizen]] ([[User talk:Trizen|talk]]) 21:04, 28 January 2016 (UTC)
| |||