Theoretical Limits of BWT-MTF-Huffman Architecture on enwik9

[Mm Mm]

Hello everyone,

I hope you are all doing well.

I have recently managed to compress the enwik9 dataset down to under 145 MB using a custom algorithm derived from the BWT ➔ MTF ➔ Huffman coding pipeline.

Currently, I am exploring a few additional ideas to integrate into my compressor to further optimize the ratio. However, before proceeding, I wanted to consult the experts here: Is there any known mathematical barrier or theoretical limit inherent to this specific family of algorithms that would prevent it from breaking the current world record?

I would highly appreciate your insights and technical feedback on this.

Thank you for your time!

---

[Matt Mahoney]

The best known BWT program compresses enwik9 to 157 MB. We don't know what the limit is. http://mattmahoney.net/dc/text.html

-- Matt Mahoney, mattma...@gmail.com

--
You received this message because you are subscribed to the Google Groups "Hutter Prize" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hutter-prize...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/hutter-prize/b0751da6-4ee1-425c-b31a-035d48b39771n%40googlegroups.com.