Huffman Encoding Table First Try

Рет қаралды 18,898

TheVimeagen

Күн бұрын

Пікірлер: 26

@tmanley1985 7 ай бұрын

We need a full network programming tutorial. I'm way too interested in this now.

@IQof2 7 ай бұрын

I work in networking and it's a really interesting bridge for me

@FunkyELF 7 ай бұрын

First try really is a moving target. Back in the day first try meant it compiled and ran correctly but now with IDEs and LSPs he's guaranteed zero compile errors.

@SnowDaemon 7 ай бұрын

First Try on the 3rd go around, lets go

@fgf80 7 ай бұрын

Huffman coding is the most efficient way to encode arbitrarily large streams of data by just growing the word size to be arbitrarily large. Other compression algorithms exist because growing the word size has exponential cost in the size of the Huffman tree, so more optimal solutions are possible even for quite large finite data streams if you can avoid the cost of transferring the encoding table itself. Limiting your encoding scheme to a fixed word size of 7 or 8 bits will hurt potential for optimization down the line. I could imagine an optimal ASCII encoding involving a pre-shared Huffman tree baked into the compression and decompression tools that’s generated from a histogram of a really large body of input data, like the library of congress, with several byte word sizes.

@dough-pizza 7 ай бұрын

Poor grug no understand what this human say! Grug understand only simpler terms

@idf3da 7 ай бұрын

Grug hear me, Huffman coding good for making data small. But if Grug make word size too big, tree get too big too. Other ways better for big data, no need big tree. If Grug use 7 or 8 bit words, it not best. Maybe best if Huffman tree already made from big data, like library of congress, and used by all tools.@@dough-pizza

@fgf80 7 ай бұрын

@@dough-pizza if your data is big enough, you get better compression out of a Huffman code if each code represents, e.g., two bytes instead of one. The problem is that makes the Huffman tree 256x bigger, so you have to be compressing a lot more data for the better compression ratio to be worth the bigger table size. If you have ENOUGH data, it’s actually been proven you can’t (in general) outperform a Huffman coding scheme where each code represents a large enough number of bytes. I was saying Prime here is limiting his possibilities for improved compression ratios by designing it around having each code correspond to exactly one byte.

@the_duda 7 ай бұрын

@@dough-pizza if data big huffman good

@dough-pizza 7 ай бұрын

@@fgf80 Grug understood! If one Huffman code represent multiple bytes then compression good! Prime here use Huffman to represent one byte so compression bad 😞