Re: A truely BIJECTIVE BWT is here!

biject <davvid_a_scott@xxxxxxxxx> writes:
I just created what I think is the first truely bijective
BWT through my Scottification process.
My code is a mod of Marks BWT and should
function in most cases as a drop in replacement.
I call it BWTS.

my bijective BWT at:

http://bijective.dogma.net/bwts.zip

it does what BWT should have done the transform with out
an index. N bytes to N bytes with only the order changed.
I don't think anyone else has a practical working bijective
BWT transform it took a lot of trail and error to make it
so that on the average it will lead to bijective BWT compression
with out the stupid INDEX and it will lead to better compression
on the average.


I suspect this is impossible. In order to achieve an
improvement in compression, BWT needs to map less-
compressible vectors onto more-compressible vectors.
However, there are fewer of the latter. It gets away
with this by mapping multiple source vectors onto the
same output vector, and then adding an auxiliary index
such that the process is reversible. BWT as such cannot
work without that index.

Now it's entirely possible that you've created a new
transform that encourages "better compression on the
average" from the back end, but the fact that it is
better is almost certainly unrelated to a lack of an
index, and is probably harmed by that lack.

Care to outline the algorithm here?

Phil
--
Dear aunt, let's set so double the killer delete select all.
-- Microsoft voice recognition live demonstration
.

Relevant Pages

  • Re: A truely BIJECTIVE BWT is here!
    ... If you look at the BWT ... rotation of a given input to the same output string. ... character sequences. ... Compression speed is also likely to be worse. ...
    (comp.compression)
  • Re: A truely BIJECTIVE BWT is here!
    ... But it wasn't fair since you didn't use the full wrap around BWT ... But when you add the indexing your see its not ... this is lossy compression since the index needed ... but the output is a pure BWT wrap around compression ...
    (comp.compression)
  • Re: A truely BIJECTIVE BWT is here!
    ... improvement in compression, ... If you look at the BWT ... The output vectors that we get are more compressible ONLY because the ... see how the transform can be reversible without additional ...
    (comp.compression)
  • Re: A truely BIJECTIVE BWT is here!
    ... If you look at the BWT ... see how the transform can be reversible without additional ... Marks compressed size ... My Compression code http://bijective.dogma.net/ ...
    (comp.compression)
  • Re: I know this wont work, I just want to know -why-.
    ... > justify to an "inventor" why their thoughts wont be practical. ... > for random-recursive compression, ... single pointer required for a full BWT. ...
    (comp.compression)