Is it known whether most numbers are normal or not
In a comment over at mathematics.stackexchange.com, I just (well, now stuff has happened and it's 8 hours ago) claimed that for most numbers it is not known whether they are normal. And then I got to think whether that is actually true?
Is there some set of numbers that (in some way, whether my measure, cardinality or something I haven't thought of) encompasses "most numbers", for which we know none (or just a few - in some meaningful sense) are normal? I believe that e.g. most rational numbers (but neither by measure nor by cardinality can they be said to be "most numbers") are not normal.
1 answer
The measure of the set of all real numbers that are not normal is $0.$
Notice first that we can restrict attention to numbers between $0$ and $1.$ A number in that interval is normal precisely of that number plus any integer you choose is normal.
Take the probability that a number between $0$ and $1$ is in any interval $(a,b),$ where $a
Then the digits of a number are independent random variables, i.e. the conditional probability that the $k$th digit is (for example) $6,$ given the values of some of the other digits, is $1/10$ regardless of the values of those other digits.
It follows from the strong law of large numbers, a standard theorem in the theory of probability, that the probability that each of the digits $0$ through $9$ occurs $1/10$ of the time is $1.$ For sequences of digits, rather than individual digits, there are complications that can be overcome in fairly routine ways. Hence the probability that a such a random number is not normal is $0.$
That does not mean there are no non-normal numbers. It's like the fact that a dart thrown at the unit square has probability zero of landing exactly on the diagonal is zero, but that doesn't mean there are not plenty of points on the diagonal.
0 comment threads
1 comment thread