Big number calculations

Ok obviously REBOL has limitation with very big numbers. However, with the kind of calculations I want to make alot of software will have limitations. I'm looking to pick some brains for an alternative to scientific notation. I like scientific notation as it represents large numbers in a compact manner however; in my case, I have large numbers and want to represent a very large number in the smallest footprint possible. Scientific Notation doesnt acheive that with the type of numbers I will be calculating. Any mathmatics gurus have any ideas? References to where I can find out about the notations would be helpful also. Paul Tretter

What type of numbers are we talking about? Is there any type of commonality between these numbers? For instance if you are representing a binary progression, the number "9" could have and equivelent decimal system value of 512. Another way to think about this technique is degrees used to measure an angle. Accounting for the numbers range could also be benificial in a shorter representation. Also what degree of accuracy is desired? Using a high number base could make a considerable difference. REBOL supports base 64, and if you brew your own you probably could achieve around base 94 by using common ascii characters. If your trying to save disk or memory space, using raw binary representation (ie "A" = 65 "AA" = 16705) is sensible. Of course many number representations would yeild characters that are not necessarily visible. And of course there is always compression. --Ryan [ptretter--norcom2000--com] wrote:

I'm not sure what you're asking (or whether I can help). If by footprint, you mean printed representation, then using base 16, 32, 64, etc with ascii or unicode symbols will work. If you mean computer memory space, then you either have to give up on representing all numbers uniquely, which is what exponential notation does, or you have to just accept the memory hit and use bignums, as in LISP. Brent Meeker On Wed, 21 Feb 2001 [ptretter--norcom2000--com] wrote:

[ptretter--norcom2000--com] wrote:

> Ok obviously REBOL has limitation with very big numbers. However, > with the kind of calculations I want to make alot of software will > ideas? References to where I can find out about the notations would > be helpful also.