Maths Olympiad Prep

Library / /96 of 106

Combinatorics Difficulty 8.9 Shortlist Prove it IMO

For any finite sets XX and YY of positive integers, denote by fX(k)f_{X}(k) the kextthk^{ ext{th}} smallest positive integer not in XX, and let
XY=X{fX(y):yY}. X * Y = X \cup \{ f_{X}(y) : y \in Y \} .
Let AA be a set of a>0a > 0 positive integers, and let BB be a set of b>0b > 0 positive integers. Prove that if AB=BAA * B = B * A, then
A(A(A(AA)))A appears b times =B(B(B(BB)))B appears a times . \begin{equation*} \underbrace{A *(A * \cdots *(A *(A * A)) \ldots)}_{A \text{ appears } b \text{ times }} = \underbrace{B *(B * \cdots *(B *(B * B)) \ldots)}_{B \text{ appears } a \text{ times }} . \tag{U.S.A.} \end{equation*}

Solutions — 2

Solution 1

For any function g:Z>0Z>0g: \mathbb{Z}_{>0} \rightarrow \mathbb{Z}_{>0} and any subset XZ>0X \subset \mathbb{Z}_{>0}, we define g(X)={g(x):xX}g(X) = \{ g(x) : x \in X \}. We have that the image of fXf_{X} is fX(Z>0)=Z>0Xf_{X}(\mathbb{Z}_{>0}) = \mathbb{Z}_{>0} \setminus X. We now show a general lemma about the operation *, with the goal of showing that * is associative.

Lemma 1. Let XX and YY be finite sets of positive integers. The functions fXYf_{X * Y} and fXfYf_{X} \circ f_{Y} are equal.

*Proof.* We have
fXY(Z>0)=Z>0(XY)=(Z>0X)fX(Y)=fX(Z>0)fX(Y)=fX(Z>0Y)=fX(fY(Z>0)) f_{X * Y}(\mathbb{Z}_{>0}) = \mathbb{Z}_{>0} \setminus (X * Y) = (\mathbb{Z}_{>0} \setminus X) \setminus f_{X}(Y) = f_{X}(\mathbb{Z}_{>0}) \setminus f_{X}(Y) = f_{X}(\mathbb{Z}_{>0} \setminus Y) = f_{X}(f_{Y}(\mathbb{Z}_{>0}))
Thus, the functions fXYf_{X * Y} and fXfYf_{X} \circ f_{Y} are strictly increasing functions with the same range. Because a strictly increasing function is uniquely defined by its range, we have fXY=fXfYf_{X * Y} = f_{X} \circ f_{Y}.

Lemma 1 implies that * is associative, in the sense that (AB)C=A(BC)(A * B) * C = A * (B * C) for any finite sets A,BA, B, and CC of positive integers. We prove the associativity by noting
Z>0((AB)C)=f(AB)C(Z>0)=fAB(fC(Z>0))=fA(fB(fC(Z>0)))=fA(fBC(Z>0))=fA(BC)(Z>0)=Z>0(A(BC)) \begin{gathered} \mathbb{Z}_{>0} \setminus ((A * B) * C) = f_{(A * B) * C}(\mathbb{Z}_{>0}) = f_{A * B}(f_{C}(\mathbb{Z}_{>0})) = f_{A}(f_{B}(f_{C}(\mathbb{Z}_{>0}))) \\ = f_{A}(f_{B * C}(\mathbb{Z}_{>0})) = f_{A * (B * C)}(\mathbb{Z}_{>0}) = \mathbb{Z}_{>0} \setminus (A * (B * C)) \end{gathered}
In light of the associativity of *, we may drop the parentheses when we write expressions like A(BC)A * (B * C). We also introduce the notation
Xk=X(X(X(XX)))X appears k times  X^{* k} = \underbrace{X *(X * \cdots *(X *(X * X)) \ldots)}_{X \text{ appears } k \text{ times }}
Our goal is then to show that AB=BAA * B = B * A implies Ab=BaA^{* b} = B^{* a}. We will do so via the following general lemma.

Lemma 2. Suppose that XX and YY are finite sets of positive integers satisfying XY=YXX * Y = Y * X and X=Y|X| = |Y|. Then, we must have X=YX = Y.

*Proof.* Assume that XX and YY are not equal. Let ss be the largest number in exactly one of XX and YY. Without loss of generality, say that sXYs \in X \setminus Y. The number fX(s)f_{X}(s) counts the sths^{th} number not in XX, which implies that
fX(s)=s+X{1,2,,fX(s)} \begin{equation*} f_{X}(s) = s + | X \cap \{ 1, 2, \ldots, f_{X}(s) \} | \tag{1} \end{equation*}
Since fX(s)sf_{X}(s) \geqslant s, we have that
{fX(s)+1,fX(s)+2,}X={fX(s)+1,fX(s)+2,}Y \{ f_{X}(s) + 1, f_{X}(s) + 2, \ldots \} \cap X = \{ f_{X}(s) + 1, f_{X}(s) + 2, \ldots \} \cap Y
which, together with the assumption that X=Y|X| = |Y|, gives
X{1,2,,fX(s)}=Y{1,2,,fX(s)} \begin{equation*} | X \cap \{ 1, 2, \ldots, f_{X}(s) \} | = | Y \cap \{ 1, 2, \ldots, f_{X}(s) \} | \tag{2} \end{equation*}
Now consider the equation
tY{1,2,,t}=s t - | Y \cap \{ 1, 2, \ldots, t \} | = s
This equation is satisfied only when t[fY(s),fY(s+1))t \in [ f_{Y}(s), f_{Y}(s+1) ), because the left hand side counts the number of elements up to tt that are not in YY. We have that the value t=fX(s)t = f_{X}(s) satisfies the above equation because of (1) and (2). Furthermore, since fX(s)Xf_{X}(s) \notin X and fX(s)sf_{X}(s) \geqslant s, we have that fX(s)Yf_{X}(s) \notin Y due to the maximality of ss. Thus, by the above discussion, we must have fX(s)=fY(s)f_{X}(s) = f_{Y}(s).

Finally, we arrive at a contradiction. The value fX(s)f_{X}(s) is neither in XX nor in fX(Y)f_{X}(Y), because ss is not in YY by assumption. Thus, fX(s)XYf_{X}(s) \notin X * Y. However, since sXs \in X, we have fY(s)YXf_{Y}(s) \in Y * X, a contradiction.

Solution 2

We will use Lemma 1 from Solution 1. Additionally, let XkX^{* k} be defined as in Solution 1. If XX and YY are finite sets, then
fX=fYfX(Z>0)=fY(Z>0)(Z>0X)=(Z>0Y)X=Y, \begin{equation*} f_{X} = f_{Y} \Longleftrightarrow f_{X}(\mathbb{Z}_{>0}) = f_{Y}(\mathbb{Z}_{>0}) \Longleftrightarrow (\mathbb{Z}_{>0} \setminus X) = (\mathbb{Z}_{>0} \setminus Y) \Longleftrightarrow X = Y, \tag{3} \end{equation*}
where the first equivalence is because fXf_{X} and fYf_{Y} are strictly increasing functions, and the second equivalence is because fX(Z>0)=Z>0Xf_{X}(\mathbb{Z}_{>0}) = \mathbb{Z}_{>0} \setminus X and fY(Z>0)=Z>0Yf_{Y}(\mathbb{Z}_{>0}) = \mathbb{Z}_{>0} \setminus Y.

Denote g=fAg = f_{A} and h=fBh = f_{B}. The given relation AB=BAA * B = B * A is equivalent to fAB=fBAf_{A * B} = f_{B * A} because of (3), and by Lemma 1 of the first solution, this is equivalent to gh=hgg \circ h = h \circ g. Similarly, the required relation Ab=BaA^{* b} = B^{* a} is equivalent to gb=hag^{b} = h^{a}. We will show that
gb(n)=ha(n) \begin{equation*} g^{b}(n) = h^{a}(n) \tag{4} \end{equation*}
for all nZ>0n \in \mathbb{Z}_{>0}, which suffices to solve the problem.

To start, we claim that (4) holds for all sufficiently large nn. Indeed, let pp and qq be the maximal elements of AA and BB, respectively; we may assume that pqp \geqslant q. Then, for every npn \geqslant p we have g(n)=n+ag(n) = n + a and h(n)=n+bh(n) = n + b, whence gb(n)=n+ab=ha(n)g^{b}(n) = n + a b = h^{a}(n), as was claimed.

In view of this claim, if (4) is not identically true, then there exists a maximal ss with gb(s)ha(s)g^{b}(s) \neq h^{a}(s). Without loss of generality, we may assume that g(s)sg(s) \neq s, for if we had g(s)=h(s)=sg(s) = h(s) = s, then ss would satisfy (4). As gg is increasing, we then have g(s)>sg(s) > s, so (4) holds for n=g(s)n = g(s). But then we have
g(gb(s))=gb+1(s)=gb(n)=ha(n)=ha(g(s))=g(ha(s)) g(g^{b}(s)) = g^{b+1}(s) = g^{b}(n) = h^{a}(n) = h^{a}(g(s)) = g(h^{a}(s))
where the last equality holds in view of gh=hgg \circ h = h \circ g. By the injectivity of gg, the above equality yields gb(s)=ha(s)g^{b}(s) = h^{a}(s), which contradicts the choice of ss. Thus, we have proved that (4) is identically true on Z>0\mathbb{Z}_{>0}, as desired.

Comment 2. We present another proof of Lemma 2 of the first solution.

Let x=X=Yx = |X| = |Y|. Say that uu is the smallest number in XX and vv is the smallest number in YY; assume without loss of generality that uvu \leqslant v.

Let TT be any finite set of positive integers, and define t=Tt = |T|. Enumerate the elements of XX as x1<x2<<xnx_{1} < x_{2} < \cdots < x_{n}. Define Sm=fTX(m1)(X)S_{m} = f_{T * X^{*(m-1)}}(X), and enumerate its elements sm,1<sm,2<<sm,ns_{m, 1} < s_{m, 2} < \cdots < s_{m, n}. Note that the SmS_{m} are pairwise disjoint; indeed, if we have m<mm < m', then
SmTXmTX(m1)andSm=(TXm)(TX(m1)) S_{m} \subset T * X^{* m} \subset T * X^{*(m' - 1)} \quad \text{and} \quad S_{m'} = (T * X^{* m'}) \setminus (T * X^{*(m' - 1)})
We claim the following statement, which essentially says that the SmS_{m} are eventually linear translates of each other:

Claim. For every ii, there exists some mim_{i} and cic_{i} such that for all m>mim > m_{i}, we have that sm,i=t+mncis_{m, i} = t + m n - c_{i}. Furthermore, the cic_{i} do not depend on the choice of TT.

First, we show that this claim implies Lemma 2. We may choose T=XT = X and T=YT = Y. Then, there is some mm' such that for all mmm \geqslant m', we have
fXm(X)=fYX(m1)(X) \begin{equation*} f_{X^{* m}}(X) = f_{Y * X^{*(m-1)}}(X) \tag{5} \end{equation*}
Because uu is the minimum element of XX, vv is the minimum element of YY, and uvu \leqslant v, we have that
(m=mfXm(X))Xm=(m=mfYX(m1)(X))(YX(m1))={u,u+1,} \left( \bigcup_{m = m'}^{\infty} f_{X * m}(X) \right) \cup X^{* m'} = \left( \bigcup_{m = m'}^{\infty} f_{Y * X^{*(m-1)}}(X) \right) \cup (Y * X^{*(m' - 1)}) = \{ u, u + 1, \ldots \}
and in both the first and second expressions, the unions are of pairwise distinct sets. By (5), we obtain Xm=YX(m1)X^{* m'} = Y * X^{*(m' - 1)}. Now, because XX and YY commute, we get Xm=X(m1)YX^{* m'} = X^{*(m' - 1)} * Y, and so X=YX = Y.

We now prove the claim.

*Proof of the claim.* We induct downwards on ii, first proving the statement for i=ni = n, and so on.

Assume that mm is chosen so that all elements of SmS_{m} are greater than all elements of TT (which is possible because TT is finite). For i=ni = n, we have that sm,n>sk,ns_{m, n} > s_{k, n} for every k<mk < m. Thus, all (m1)n(m - 1) n numbers of the form sk,us_{k, u} for k<mk < m and 1un1 \leqslant u \leqslant n are less than sm,ns_{m, n}. We then have that sm,ns_{m, n} is the ((m1)n+xn)th((m - 1) n + x_{n})^{\text{th}} number not in TT, which is equal to t+(m1)n+xnt + (m - 1) n + x_{n}. So we may choose cn=xnnc_{n} = x_{n} - n, which does not depend on TT, which proves the base case for the induction.

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.