Answers to the Compression Challenge, 2004

Back to Compression Challenge | ITPRNN | ITILA

Name Methodcompressed size
James Kermode huffman code to encode the run-lengths of zeros, with associated probabilities (1-f)**n * f 837 bits (plus a couple of bytes)
James Kermode Block huffman code, block size 10 bits roughly 1200 bits
B.J. Lowing splitting the file into short blocks and then writing the absolute position of the ones within the blocks 880 bits (no uncompressor yet)
Jeremy Taylor Golomb compression algorithm. The decoder is expected to know the length of the runlength codes (I've set it to 7). 870 bits plus a end of file character
Jeremy Taylor Code that just sends the position of the 1s. 1442 bits plus an EOF character.

David MacKay
Last modified: Fri Mar 12 15:12:36 2004