Given a number $N$ with some properties:

  • It is only constructed using these digits: $\{1,3,5,7,9\}$ (e.g. $135$ , $7713$. NOT: $1231$ which includes the digit $2$)
  • It has at least $3$ distinct odd digits. (for example these are not allowed: $7755$, $13$, $11$, $3$ etc..)
  • Its sum is not divisible by $3$ (and thus $3 \nmid N$)

I've written a (very clunky) Python program that takes a random number with these very properties, and checked:

If the number is divisible by $3$ - skip
Else - check each permutation of the number if it is a prime.

For example (for simplicity I've used $5131$ which has only $2$ unique odd digits):

$335511$ - skip, it is divisible by $3$.
$5131$ - check each permutation of this number if it is a prime:
$5131$ - not a prime. $5113$ - not a prime. etc... until we hit a prime: $5113$

The results were quite nice, I've noticed that if the numbers meet those constraints then:

  • At least one permutation of $N$ is a prime

These constraints look very 'harsh' at the beginning, but I don't think it is that trivial - that a number with $3$ or more unique odd digits has at least one permutation that is a prime (If it is not divisible by $3$).

Is there any reason why these numbers behave this way?


Now assume $N$ can be constructed only using $\{1,3,7,9\}$ (no $5$ allowed) and it has at least $3$ unique digits (so maybe it contains $1,3,7$ or $1,3,7,9$ or $3,7,9$ etc..)

Now the number is either divisible by $3$ or has at least one prime permutation.
I could not seem to find any counter-example to this, however it must exist as what @lulu said, it is a matter of probability, but after checking millions and million of numbers - I couldn't ... so such number does exist?

Thank you!

  • 4
    $\begingroup$ There are a lot of prime numbers. Just for example, if you stick with $8$ digits or fewer then there are $\pi(10^8)\approx 5.7\times 10^6$ primes. It follows that if you choose a random odd number with $8$ digits there's a greater than $\frac 19$ chance that it is prime. Since it is likely that your number has many more than $9$ permutations, this may just be chance. $\endgroup$
    – lulu
    Jul 16 '20 at 10:51
  • $\begingroup$ To build counterexamples, I'd take a long number with all but two digits equal to $5$ and then two other odd digits. That keeps the number of relevant permutations as low as possible (roughly $2L$ where $L$ is the length). $\endgroup$
    – lulu
    Jul 16 '20 at 11:27
  • $\begingroup$ @lulu Hey thank you for answering! please see my update question it has more constraints $\endgroup$
    – CSch of x
    Jul 16 '20 at 18:04
  • $\begingroup$ Just as a sidelight, you may want to investigate the topic of permutation primes: those primes for which every permutation of their digits produces a prime, e.g. 991, the largest such with three digits. There is a Wiki page on them if you are interested. $\endgroup$ Jul 16 '20 at 18:13
  • $\begingroup$ @ChrisLeary Thanks for the answer! but I am talking about at least one* prime permutation, not all $\endgroup$
    – CSch of x
    Jul 16 '20 at 18:17

No permutation of the digits $1,5,5,5,7$ creates a prime. Obviously it cannot end in a $5$, and the others are:


For 10 digits there is also $3555555557$ and $5555555579$.

I found these by computer. There are no other such combinations with 12 digits or fewer.

  • $\begingroup$ Nice catch! I've actually looked it up beforehand (not the same number but other number with 5 in it) and it seems that this what causes the problem, what about a number that is constructed from $\{1,3,7,9\}$ ? $\endgroup$
    – CSch of x
    Jul 16 '20 at 10:51
  • $\begingroup$ @Remember1312 As lulu points out in their comment, the probability is pretty low. It is also not surprising that in the ones I found, all but two of the digits are $5$, because then there are only $2n-2$ numbers that would have to be composite. WIthout a $5$, the number of permutations that would have to be composite is so much bigger that it is very very unlikely to happen, though not necessarily impossible, and I think with more digits it gets ever more unlikely. $\endgroup$ Jul 16 '20 at 11:21
  • $\begingroup$ Thank you sir! Please see my update, I am very curious to look for your answers, thank you again $\endgroup$
    – CSch of x
    Jul 16 '20 at 18:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.