mersenneforum.org Ordered disorder
 Register FAQ Search Today's Posts Mark Forums Read

 2019-04-25, 05:59 #1 jasong     "Jason Goatcher" Mar 2005 1101101100112 Posts Ordered disorder Sorry if this is deemed to be deserving of Bloggorhea, but anyway... How do you define order so that you could make something as disordered as possible? For example, take the numbers 0-9, what order would be the most disordered, and can you explain it, or generate it, deterministically? Could prime numbers ultimately be explained as a disorder of a particular type? What abouts binary, what's the most disordered binary series you can come up with, and what the rules to generate it again? If you guys can recommend Kindle books or web pages about this, that would be awesome.
 2019-04-25, 06:13 #2 retina Undefined     "The unspeakable one" Jun 2006 My evil lair 2×23×137 Posts I'm unaware of any formal definition of disorder. But a reasonable proxy could be randomness. And from there you could measure compressibility (as in zip or 7z etc.). Generally the more randomness you have the less compressible is the input. So just run your data through 7z and examine the file size. A larger ratio of output size to input size could be a measure of disorder. For the digits 0-9 appearing only once, a compressor probably wouldn't be a good measure, you'd need more data to get a good approximation.
 2019-04-25, 07:55 #3 Nick     Dec 2012 The Netherlands 3×587 Posts There is a concept known as information entropy Maybe that is what you are looking for?
 2019-04-25, 09:14 #4 axn     Jun 2003 22·1,301 Posts Also its cousin Kolmogorov complexity
 2019-04-25, 15:27 #5 Dr Sardonicus     Feb 2017 Nowhere 22·32·11·13 Posts If it's a physical system, I'd go with entropy. If it's a sequence of numbers, say decimal or binary digits, I would think instead of (apparent) randomness. For binary digits, a "rule" like "flip a coin, 1 for heads, 0 for tails" might do. There are statistical tests for (apparent) randomness. Sequences of digits may be viewed as representing real numbers. There are what are called normal numbers, whose (e.g. decimal) expansions contain all digits and all blocks of 2, 3, or any finite number of digits in the correct proportions. AFAIK there is no good way to test known numbers (like pi) for being "normal" (to base ten). It is however very easy to show that "almost all" numbers are normal to every base. Another notion would seem to be related to the "Kolmogorov complexity" mentioned above. namely, can you compute the n-th digit of the number without computing all the previous digits? For the decimal expansion of pi, the answer is AFAIK not known, but in binary the answer is "yes." However, if memory serves, it makes no sense to say, "This number is random."
 2019-04-26, 07:15 #6 LaurV Romulan Interpreter     "name field" Jun 2011 Thailand 22×11×223 Posts Defining a "most disordered thing" (to keep the language) is a known paradox, and it is not possible. If you have some "disordered" thing to which you attach "the most", then it is not as disordered as other things (you already ordered it, put it in the "top" of some list with disordered things).

 Similar Threads Thread Thread Starter Forum Replies Last Post mackerel Hardware 38 2016-11-20 06:33 jinydu Abstract Algebra & Algebraic Number Theory 2 2013-11-25 09:08

All times are UTC. The time now is 14:17.

Wed Dec 8 14:17:55 UTC 2021 up 138 days, 8:46, 2 users, load averages: 1.47, 1.58, 1.57