Problems & Puzzles: Puzzles

Puzzle 584. Embedded but not overlapped prime factors

Claudio Meller posed the question 637 in his site:

22940075 = 52  229  4007

I have gotten the following other solutions:

735, 3792, 1341275, 13115375, 22940075, 29373375, 71624133, ...

Q1. Send only the largest case you can find.

Q2. Prove that there is not any integer containing concatenated (embedded but not overlapped) all its prime factors, counting repetitions, or send a counterexample.


Contributions came from Giovanni Resta, Jan van Delden, Hakan Summakoğlu, Emmanuel Vantieghem.


Giovanni wrote:

Q1. 347 digits

492 =
2^2 x 3^3 x


My claim is that
if n = P_1 P_2 P_3 ... P_k with the various primes non
necessarily distinct, then n cannot be a concatenation
(in any order) of P_1, P_2,..., P_k.

Let us call D(x) the number of digits of x.
We want D(n) = D(P_1)+D(P_2)+...+D(P_k).

if D(P_j) = D_j then 10^(D_j-1) < P_j < 10^D_j

Hence I can write P_j = 10^D_j*(1 - E_j) with 0 < E_j < 1.

Let me multiply these representations of the P_j, taking
apart the powers of 10.

n = 10^(D_1+D_2+...+D_k)*(1-E_1)(1-E_2)...(1-E_k) ,

The question is: could n start with the digits of one of the
Pj ? If this is the case, then we can write n as P_j*10^(Dn-D_j)+R_j,
where Dn = D_1+D_2+...+D_k is equal to the number of digits of n,
and R_j>=0.
So, it must be n >= P_j*10^(Dn-D_j). On the other side,
how large could be n= 10^(D_1+D_2+...+D_k)*(1-E_1)(1-E_2)...(1-E_k) ?

We want to see if it could be
n = 10^(D_1+D_2+...+D_k)*(1-E_1)(1-E_2)...(1-E_k) >= P_j*10^(Dn-D_j)
for at least one P_j.
Since P_j = 10^D_j (1 - E_j), and D_1+D_2+...+D_k=Dn we can write
10^(D_1+D_2+...+D_k)*(1-E_1)(1-E_2)...(1-E_k) >=
10^D_j (1 - E_j)*10^(Dn-D_j),
10^Dn*(1-E_1)(1-E_2)...(1-E_k) >= (1-E_j)*10^Dn, and,
taking out the 10^Dn factor and the (1-E_j) factor:
(1-E_1)(1-E_2)...(1-E_{j-1}) (1-E_{j+1})...(1-E_k) >= 1

but this is clearly impossible, since a product of quantities
smaller than 1 cannot be greater or equal to 1.

Hence n cannot start with any prime in its factorization, as it claimed.

Let me make an example.
Take for example n = 83 * 997 * 9973  = 825275723
and let me "prove" that 825275723 cannot start with
83, 997 or 9973.

First of all I check if the number of digits match:
2 + 3 + 4 = 9 OK.

As I wrote, I can write n as
n = 10^2*(1-0.17) * 10^3*(1-0.003) * 10^4*(1-0.0027)

since 10^2*(1-0.17) = 83, and so on.

Let us Assume I want that n "starts" with 83 (the other cases
are similar), so, 83 is my P_j.

n must be equal to 83 followed by a number of digits
equal to the digits of the primes except 83 itself,
so I can write it like
83*10^3*10^4 + something (R_j is my "something" in the proof).
In any case, I can say that n must be greater or equal to
83*10^3*10^4, so, using the expression for n above, I have

10^2*(1-0.17) * 10^3*(1-0.003) * 10^4*(1-0.0027) >= 83*10^3*10^4

Now, let me express 83 as 10^2*(1-0.17) on the right term above:

10^2*(1-0.17) * 10^3*(1-0.003) * 10^4*(1-0.0027) >= 10^2*(1-0.17)* 10^3*10^4

At this point I can delete all the powers of 10, since the sum of
the exponents on the left is the same as on the right.

(1-0.17) * (1-0.003) * (1-0.0027) >= (1-0.17)

Then I can delete the term 1-0.17 (1-E_j) and I'm left with
(1-0.003) * (1-0.0027) >= 1

which is impossible, since the product of two terms smaller than 1
cannot be greater of equal to 1.
So I "proved" that 825275723 does not start with "83".


Jan wrote:

If q=(26*10^(n-1)+1)/63 is prime (q={412698}n.4127 with n=6k+5) then the concatenation 13.q.5 has factors: 5^2*13*q and a total of 6k+7 digits.
It is prime for k=0,2,3,182,..? (I found this family by considering p.q.5=p*q*5^2 and observing that q should start with a 4 and end with a 7).
My largest value not of this type is 1019127375=3^2*5^3*7*127*1019.
Write N as A.B, the concatenation of the numbers A and B. [A,B prime not necessary].
Choose n to be the smallest value for which B<10^n.
We have: N=A*10^n+B. Can N be equal to A*B?
We have N=A*B<A*10^n but also N=A*10^n+B which is larger, so equality is not possible.
Suppose we have that N[k]=p[1].p[2]....p[k] and proven that this is larger than p[1]*p[2]*..p[k].
We know this is true if k=2 (Choose A=p[1] and B=p[2]).
Induction step: Consider N[k+1]=p[1].p[2]...p[k+1]
Split N[k+1] using the terms A=p[1].p[2]...p[k]=N[k] and B=p[k+1].
Therefore N[k+1] can not be written as a product of primes with multiplicity 1.

In fact it can be proven, that we need at least 3 different prime factors. (Omitted)


Hakan Summakoğlu wrote:

Q1: 319953792 =2^735379199.

I found a different integer -35331072 (35331072=2^103073153)


Emmanuel wrote:

Q1.  The 13 smallest numbers that are a concatenation of their prime factors is sequence A121342 at the OEIS.  There are no such numbers in the region  10^10 14 10^9.

Later he added:

I found a  788-digit solution to Q1 :
  the number  p = (3x10^787 + 5)/35  is prime (certified by PRIMO).  The concatenation of  3,p,5  factors as  3^2 x 5 x p. 

Q2.  The concatenation of two numbers  A  and  B  is always > AB.

Proof : Assume B  has  n  digits.  Then, the concatenation  N  of  A  and  B  is  A 10^n+B >= A 10^n > AB.

Now, if  N1  is the concatenation of  k  numbers  A1, A2,...  , Ak,  N2  the concatenation of   A2, A3,...  , Ak, ... , Ni  the concatenation of  Ai, Ai+1,...  , Ak , then we have :
      N1  > A1 N2 > A1 A2 N3 > ... > A1 A2 ... Ak.
Hence it is impossible that the concatenation of  k  numbers equals the product of these numbers.


Records   |  Conjectures  |  Problems  |  Puzzles