Does the string of prime numbers $$2357111317\ldots$$ contain every natural number as its sub-string?
-
4It's unlikely to contain π as a substring, because π contains a decimal point and this string doesn't. Do you mean "every integer"? – gnasher729 Aug 30 '14 at 23:26
-
2No minus signs, @gnasher729. So has to refer to non-negative integers. :) – Thomas Andrews Aug 30 '14 at 23:27
-
4See Copeland-Erdos constant. – Lucian Aug 30 '14 at 23:30
-
1@Buddha, you didn't mention natural numbers, so change your question. And extending it to reals would be stupid because your string doesn't contain a decimal point or a minus character, so it doesn't contain the numbers 1.5 and -2, for example. – gnasher729 Aug 30 '14 at 23:34
-
3Possible duplicate of Proof that there are infinitely many prime numbers starting with a given digit string – Watson Nov 28 '18 at 12:57
4 Answers
It follows from Dirichlet's Theorem.
If $d$ is the number we want to find, define $s=10d+1$. By definition, $\gcd(s,10)=1$ and $s$ contains the digits of $d$.
Dirichlet's Theorem's implies there's a prime of the form $p:=s+k \times 10^n$ where $10^n$ is chosen so that it has as many zeroes as digits of $s$. The digits of $d$ appear in the digits of $p$, and thus in the given string of primes.

- 26,845
-
1Very nice! This is actually exercise 4.16.b in Apostol's Introduction to Analytic Number Theory. – Greg Hurst Aug 31 '14 at 05:28
-
If you mean "every non-negative integer", then the answer is yes.
First, it contains the integer 0 as a substring because 101 is a prime.
Next, for every integer i > 0, there is a prime that starts with the integer i: Take the known results about gaps between prime numbers, for example that there is always a prime between $n^3$ and$(n+1)^3$ for large n. There is always a large n such that both $n^3$ and$(n+1)^3$ start with the digits of i, therefore the prime between $n^3$ and$(n+1)^3$ also starts with the digits of i.
(This is true because one of i, 10i and 100i is not the cube of an integer).

- 10,113
-
1I love that you have two solutions here: you show that every $k$ is a prefix of a prime written in the decimal system, the other answer shows that every $k$ is a suffix of it :) – yo' Aug 31 '14 at 19:56
-
3@tohecz Yes, it actually means there's no need to string the primes together, simply $2,3,5, \dots$ will contain all natural numbers somewhere. – Mark Hurd Sep 02 '14 at 01:54
Let $n$ be a natural number, and let $\mathcal N(n)$ be the set of natural numbers not containing $n$ as a substring when written in base 10. It is well-known that $\sum_{k\in \mathcal N(n)}\frac{1}{k}$ converges.
Let $\mathcal P$ be the set of prime numbers. It is also well-known that $\sum_{p \in\mathcal P}\frac{1}{p}$ diverges.
Therefore it cannot be the case that $ \mathcal P\subseteq\mathcal N(n)$, and $n$ is therefore a substring of some prime (of infinitely many primes, in fact).

- 14,881
It follows from the fact that the Copeland-Erdös constant is a normal number.
For any natural number $N$ with $k$ digits, $N$ appears in the Copeland-Erdos constant with a natural density of $10^{-k}$.

- 23,793

- 111