Assumptions for Compressor

[Android User]

I just wanted to check whether a particular scenario aligned with the "spirit of the prize":

Let's say I wrote a compressor that completely ignored the enwik9 file, and essentially built the decompressor without any other input. If the combined size of the compressor and decompressor were sufficiently small that they won the prize, would this be acceptable within the rules?

I'm trying to establish how much the compressor can 'know' about the input data  in advance.

Many thanks,

John

[Matt Mahoney]

Yes you can do that, but why would you? Comp.exe is allowed to read enwik9 and outputs archive9.exe. archive9.exe must run with no input and output enwik9. Your score is size(comp.exe) + size(archive9.exe). If comp.exe does not read enwik9 then it has to contain a compressed copy of it, just like archive9, which would double the total size.

-- 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/ba9e6254-ea1d-4f7d-9fe5-d0990db8299bn%40googlegroups.com.

[Android User]

You are of course correct - it wouldn't make sense to go to that extreme. I just wanted to check what could be included in the compressor (such as pre-computed tables etc.), and the logical conclusion of that would be if everything was pre-computed and included in the compressor.

Thank you for your answer; I find the whole challenge really interesting, and I'm looking forward to having a go at it.

Regards,

John