Den tisdag 12 juli 2016 kl. 21:40:36 UTC+2 skrev jonas.t...@gmail.com: > Den tisdag 12 juli 2016 kl. 20:20:52 UTC+2 skrev Michael Torrie: > > On 07/12/2016 11:46 AM, jonas.thornv...@gmail.com wrote: > > > Well the algorithm start with looking up a suitable folding structure > > > "close enough to the number", then it works down the folding > > > structure finding the fold where the difference or sum between the > > > folds closest to zero. > > > > > > You do the same prinicple with the remainder until zero is achieved. > > > > > > So our first fold can either be bigger or smaller, and it seek a > > > configuration for the fold that close in max on the actual random > > > number. The second fold could be a fold that depending upon our first > > > fold was bigger or smaller than number either will add or subtract > > > lower layers of the fold. > > > > > > There will come out a difference that need to be folded, the process > > > is repeated until there is nothing to fold. > > > > > > It is basicly a search algorithm looking for suitable folding > > > structures. > > > > Better patent it quickly then. And you will win a noble prize for math > > if you could do what you say you could. > > I must stress when i say number here i really mean +100000 decimal digit. So > i basicly search in on big numbers that i compress. So i divide the dataset > into suitable sizes for compression.
And the dataset chunks that comes out from the process can also be treated like a new datafile, so the compression is iterative down to a limit. -- https://mail.python.org/mailman/listinfo/python-list
