Re: Still on the Arithmetic coder

Thanks for the replies. Yes it will give me a good learning exercise. I just
have to work out the chaining routine.

Actually, the reason for the migration to greater than 64bit was that it
seemed that under certain circumstances the arithmetic coder became a linear
predictor. To me this seemed to suggest that frequencies ahead might be
seen(predictable) - not sure on that yet - or that as a linear predictor it
might be useful as a cryptography tool.

regards
"David A. Scott" <daVvid_a_scott@xxxxxxxxx> wrote in message
news:Xns9757EF7E75F06H110W296LC45WIN3030R@xxxxxxxxxxxxxxxxxx
> "Learning_boy" <Newbie@xxxxxxxxxxx> wrote in
> news:43d9a8e0$0$18640$5a62ac22@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx:
>
>> Am still working on learning about the arithmetic coder. Have tried
>> the 64 bit mod and am wondering what one would have to do to increase
>> the accurracy if trying 128 bit. Would I have to try setting up an
>> array to read in and perform calculations on the data?
>>
>> Thanks
>>
>>
>>
>>
>
> Actually using a full 64 bit arithmetic coder is about as good as
> it gets going to 128 bit would most likely buy you very little if any
> gain. What makes you think its not giving you good accurracy now.
> Maybe you need a better model for the type of compression you want
> to do but a proper arithmetic with a good model is as good as it gets.
>
> Actually when it comes to entropy encoders there is nothing that
> really beats an arithmetic. Yes before you state there are many
> flashes in the pan that can quote thousands of lines of texts or
> even modify arithmetic to attempt to make there point. But the fact
> is in real world practial applications its going to be along time
> before arithmetic compressors get replaced. There is nothing easier
> to mate with models for practical problems.
>
> In fact 32 bits is usually overkill but 64 bits is just a bit
> sweeter.
>
> But to directly anwser you question its not hard to write routines
> where a series of intergers are chained together to get any length
> of bits you want. The problem is the time it takes to do the add
> subtracts and multiplies. I would suggest you try using 4 16 integers
> in an arry and then write subroutines to try to exactly duplicate the
> 64 bit method. the subroutines then can be called every where you would
> have done some simple operation on the 64 bit numbers. When you get this
> done and it works go for broke and chain 4 64 bit numbers together to
> get 256 bits. Hay it would be a good learning exercise.
>
>
> David A. Scott
> --
> My Crypto code
> http://bijective.dogma.net/crypto/scott19u.zip
> http://www.jim.com/jamesd/Kong/scott19u.zip old version
> My Compression code http://bijective.dogma.net/
> **TO EMAIL ME drop the roman "five" **
> Disclaimer:I am in no way responsible for any of the statements
> made in the above text. For all I know I might be drugged.
> As a famous person once said "any cryptograhic
> system is only as strong as its weakest link"
>


.