Maths Olympiad Prep

Library / /34 of 36

, 2023

Combinatorics Difficulty 9.0 Shortlist Prove it Baltic Way

For positive integers nn, kk let S(n,k)S(n, k) be the number of ways to partition a set with nn elements into kk non-empty parts. For example S(3,2)=3S(3, 2) = 3, because there are three ways to partition the set {1,2,3}\{1, 2, 3\} into two parts. For another example, S(4,2)=7S(4, 2) = 7.
Let pp be a prime number, and mm, nn positive integers such that p1mnp-1 \mid m-n. Prove that S(m,i)S(n,i)(modp)S(m, i) \equiv S(n, i) \pmod{p} for all i=1,2,,p1i = 1, 2, \dots, p-1.

Solution

Let Epi(n,i)\mathrm{Epi}(n, i) be the number of surjective functions from a set with nn elements to a set with ii elements. It is not hard to see that S(n,i)=Epi(n,i)i!S(n, i) = \frac{\mathrm{Epi}(n, i)}{i!}. Indeed, a surjective function from {1,,n}\{1, \dots, n\} to {1,,i}\{1, \dots, i\} is the same thing as a partition of {1,,n}\{1, \dots, n\} into ii parts, together with a numbering of the blocks of the partition, from 11 to ii. For every partition there exists i!i! numberings, and therefore S(n,i)i!=Epi(n,i)S(n, i) \cdot i! = \mathrm{Epi}(n, i).

When 1ip11 \le i \le p-1, gcd(p,i!)=1\gcd(p, i!) = 1. It follows that for 1ip11 \le i \le p-1, S(n,i)S(m,i)(modp)S(n, i) \equiv S(m, i) \pmod{p} if and only if Epi(n,i)Epi(m,i)(modp)\mathrm{Epi}(n, i) \equiv \mathrm{Epi}(m, i) \pmod{p}. So it is enough to prove that if p1mnp-1 \mid m-n then Epi(n,i)Epi(m,i)(modp)\mathrm{Epi}(n, i) \equiv \mathrm{Epi}(m, i) \pmod{p} for all 1ip11 \le i \le p-1.

There is a formula for Epi(n,i)\mathrm{Epi}(n, i):
Epi(n,i)=j=0i1(1)j(ij)n(ij). \mathrm{Epi}(n, i) = \sum_{j=0}^{i-1} (-1)^j (i-j)^n \binom{i}{j}.
This formula is based on nothing more than the inclusion-exclusion principle. The number of surjective functions is the number of all functions i-i times the number of functions that miss one particular element +(i2)+ \binom{i}{2} times the number of functions that miss two particular elements, etc. The number of functions that miss jj specified elements is (ij)n(i-j)^n.

Now suppose mm, nn are positive integers satisfying p1mnp-1 \mid m-n. Then aman(modp)a^m \equiv a^n \pmod{p} for all integers aa. When aa is not divisible by pp this follows from Fermat's little theorem. When aa is divisible by pp, both sides are zero mod pp. But we only need the cases when a=1,,p1a = 1, \dots, p-1 anyway.

It follows that if p1mnp-1 \mid m-n then Epi(n,i)Epi(m,i)(modp)\mathrm{Epi}(n,i) \equiv \mathrm{Epi}(m,i) \pmod{p} for all 1ip11 \le i \le p-1, and we are done.

Lemma:
For every i>1i > 1 and n>in > i:
S(n,i)=iS(n1,i)+S(n1,i1) S(n,i) = i \cdot S(n-1,i) + S(n-1,i-1)
Proof. Select an element xx. There are S(n1,i)S(n-1,i) ways to partition the remaining elements into ii different sets. By adding xx to any of them, we get a partition of the original set. All the partitions obtained this way are different and the only partitions that cannot be obtained this way are the ones where xx ends up alone. But in that case the n1n-1 other elements must form i1i-1 sets, yielding another S(n1,i1)S(n-1,i-1) partitions. \square

Lemma:
For every pp prime, p>i1p > i \ge 1 and λ0\lambda \ge 0:
S(i+λ(p1),i)=1(modp). S(i + \lambda(p-1), i) = 1 \pmod{p}.
Proof. As S(n,1)=1S(n,1) = 1 and S(i,i)=1S(i,i) = 1, this is obvious for i=1i = 1 or λ=0\lambda = 0. We proceed by induction over ii and λ\lambda. Using the induction hypothesis and the previous lemma:
S(i+λ(p1),i)=iS(i1+λ(p1),i)+S(i1+λ(p1),i1)=iS(i1+λ(p1),i)+1mod p \begin{aligned} S(i + \lambda(p-1), i) &= i \cdot S(i-1 + \lambda(p-1), i) + S(i-1 + \lambda(p-1), i-1) \\ &= i \cdot S(i-1 + \lambda(p-1), i) + 1 \quad \text{mod } p \end{aligned}
Otherwise applying this p1p-1 times gives
S(i+λ(p1),i)=ip1S(i+(λ1)(p1),i)+j=0p2ijmod p=ip11+ip11i1mod p=11+0mod p. \begin{aligned} S(i + \lambda(p-1), i) &= i^{p-1} \cdot S(i + (\lambda - 1)(p-1), i) + \sum_{j=0}^{p-2} i^j \quad \text{mod } p \\ &= i^{p-1} \cdot 1 + \frac{i^{p-1} - 1}{i-1} \quad \text{mod } p \\ &= 1 \cdot 1 + 0 \quad \text{mod } p. \end{aligned}
Where the last equality holds by Fermat's little theorem as p>i>1p > i > 1 implies i0,1(modp)i \neq 0, 1 \pmod{p}. \square

We can now proceed to proving the problem statement by induction: If i=1i=1 the statement is obvious. If m=im=i or n=in=i, we have the statement of the second lemma. So assume i>1i>1 and n>in>i, m>im>i and that the problem statement holds for all smaller values. Then
S(n,i)S(m,i)=i(S(n1,i)S(m1,i))++S(n1,i1)S(m1,i1)=i0+0mod p=0mod p. \begin{aligned} S(n,i) - S(m,i) &= i \cdot (S(n-1,i) - S(m-1,i)) + \\ &\quad + S(n-1,i-1) - S(m-1,i-1) \\ &= i \cdot 0 + 0 \quad \text{mod } p \\ &= 0 \quad \text{mod } p. \end{aligned}

Want a route through all this instead of an archive? The track puts 2,000 problems in a working order, from AMC 10 level to the IMO shortlist.

Source: MathNet, licensed CC-BY-4.0. Statement and solution reproduced as published; topic and difficulty added by this site.