RE: *** GMX Spamverdacht *** Re: paradoxes of randomness
> If the output is random,then it will have no > mathametical structure,so I shouldn't be able to > compress it at all. You could very well end up with all tails. That's a sequence that has the same probability of happening that any other sequence. A compressor will look for redundancy in the input you give it, not in the algorithm you used to generate that input (conceptually, a compressor could deduce the (determinist) algorithm from the output, but if you bring it true randomness, chances are it will not). Thus, a compressor will compress very well a sequence made of all tails, but badly another which exhibits no detectable redundancy. Once you have the sequence, you lost a lot of info about whatever algorithm was used to generate it. A sequence of all tails could have been generated by a simple algorithm which generates all tails. That's an emergement description of this one particular sequence, but one that would not apply to *all* sequences your algorithm can ever produce. That's lost information, and that's why it can be compressed. -- Vincent Penquerc'h
Re: *** GMX Spamverdacht *** Re: paradoxes of randomness
hi, Thank you-one more question. Will the information obtained from the 2^32 tests have a zero compression rate? If one of the occurance should yield all heads and one occurance yields all tails-there appears to be scope for compression. If the output is random,then it will have no mathametical structure,so I shouldn't be able to compress it at all. Regards Sarath. --- Dave Howe <[EMAIL PROTECTED]> wrote: > for a sufficiently large sample you *should* see > roughly equal numbers of > heads and tails in the average case - but : > for 32 coins in 2^32 tests you should see: > one occurance of all heads (and one of all tails) > 32 occurances of one tail, 31 heads (and 32 of one > head, 31 tails) > 496 occurances of two > and so forth up the chain > none of these are guaranteed - it *is* random after > all - but given a > sufficiently large number of tests, statistically > you should see the > above. > __ Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software http://sitebuilder.yahoo.com
