Anonymous user
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 21:35, 30 January 2016
, 8 years agono edit summary
mNo edit summary |
No edit summary |
||
Line 46:
::::::::: I do not think you read what I wrote.
::::::::: That said, using cumulative frequencies in place of letter values for your encoding means words like THE, ONE, TWO, DOG, BAD, etc. etc. all encode to the same number. --[[User:Rdm|Rdm]] ([[User talk:Rdm|talk]]) 18:45, 30 January 2016 (UTC)
:::::::::: Yup, they SHOULD all encode to the same number. That's the point, because they have the same set of frequencies. It's the dictionary that allows you to decode them. In Huffman Coding, all of those should also similarly encode to the same binary code, and it's the dictionary that allows you to decode. --[[User:Spoon!|Spoon!]] ([[User talk:Spoon!|talk]]) 21:35, 30 January 2016 (UTC)
:::::::: The purpose of all this is to divide the interval from 0 to 1 (or in this case, the interval from 0 to (length of string) since they want to use all integers) fully into intervals for each of the characters, where the length of the interval for each character is proportional to its frequency. So we want an interval of length 1 for A, length 2 for B, and length 3 for D. The implementations here accomplish that by assigning them from 0 upwards to the characters in ascending character order. So A gets the interval 0-1, B gets the interval 1-3, and D gets the interval 3-6. So each character gets an interval represented by the start being its "cumulative frequency", and the length of the interval being its frequency. --[[User:Spoon!|Spoon!]] ([[User talk:Spoon!|talk]]) 01:15, 29 January 2016 (UTC)
::::::::: Thank you, (and actually someone else adjusted that setting on my behalf), that worked around the problem with PNG rendering. --[[User:Rdm|Rdm]] ([[User talk:Rdm|talk]]) 18:45, 30 January 2016 (UTC)
| |||