BWM Runde 2 Mock Paper 1 · IMOolympiad.com · Original practice

4 written-solution problems · Take-home proof practice; no fixed examination timer

This is independent preparation for a take-home competition. For actual entries, follow the organiser’s rules on independent work and permitted collaboration; our hints and solutions are for these original practice tasks only.

Original independent practice in English. Not an official paper or predicted selection test. Keep hints and solutions closed during your attempt.

How to review your proof

Check that you have used every condition, justified the main idea, covered all cases and stated the conclusion. A different complete proof can also be correct. This is a self-review checklist; written proofs are not automatically marked.

Question 1

Let kk be a fixed positive integer. Determine all functions f:Z>0→Z>0f:\mathbb Z_{>0}\to\mathbb Z_{>0} satisfying

f(f(n))=n+kfor every positive integer n.\begin{gathered}f(f(n))=n+k\\\text{for every positive integer }n.\end{gathered}

Your description should show for which kk such functions exist and give their exact number.

Hint 1

Evaluate f(f(f(n)))f(f(f(n))) in two ways to obtain f(n+k)=f(n)+kf(n+k)=f(n)+k. Thus the first kk values determine the rest.

Hint 2

Look at the residue classes 1,…,k1,\ldots,k. They must be paired, with no class paired to itself. For a pair i,ji,j, write f(i)=j+tkf(i)=j+tk, f(j)=i+skf(j)=i+sk.

Worked solution 1

Reduce to the first block. Applying the equation at f(n)f(n) and also applying ff to the equation at nn gives

f(f(f(n)))=f(n)+k=f(n+k).f(f(f(n)))=f(n)+k=f(n+k).

Consequently f(i+qk)=f(i)+qkf(i+qk)=f(i)+qk for 1≤i≤k1\le i\le k and every nonnegative integer qq. Every positive integer has a unique such form.

Let π(i)\pi(i) be the representative in {1,…,k}\{1,\ldots,k\} of the residue of f(i)f(i) modulo kk. Positivity permits writing f(i)=π(i)+tikf(i)=\pi(i)+t_i k, with ti≥0t_i\ge0. The composition condition modulo kk says π(π(i))=i\pi(\pi(i))=i. Thus π\pi is a permutation consisting of fixed points or pairs.

A fixed point is impossible: if f(i)=i+tkf(i)=i+tk, the shift rule gives f(f(i))=i+2tkf(f(i))=i+2tk, which cannot equal i+ki+k for integer tt. Therefore the kk residue classes must split into disjoint pairs, so kk must be even.

Describe every allowed pair. For a pair i,ji,j, write f(i)=j+tkf(i)=j+tk and f(j)=i+skf(j)=i+sk, with s,t≥0s,t\ge0. Then f(f(i))=i+(s+t)kf(f(i))=i+(s+t)k, so s+t=1s+t=1. Exactly two choices remain:

(f(i),f(j))=(j,i+k)or(j+k,i).(f(i),f(j))=(j,i+k)\quad\text{or}\quad(j+k,i).

Conversely, choose any pairing, choose either of these two orientations for each pair, and extend by f(i+qk)=f(i)+qkf(i+qk)=f(i)+qk. All values are positive, and direct substitution gives f(f(i+qk))=i+(q+1)kf(f(i+qk))=i+(q+1)k. Hence every construction works.

Count. For even k=2mk=2m, there are k!/(2mm!)k!/(2^m m!) pairings: arrange the labels, group consecutive labels in pairs, then divide by the two orders within each pair and by the m!m! orders of the pairs. Each pair has two independent orientations. Thus the total is k!/m!k!/m!. Different choices change the first kk values, so they give different functions. For odd kk, there are none.

Conclusion: For odd kk, none. For even k=2mk=2m, pair the residue classes and orient each pair as described; there are k!/m!k!/m! functions.

Question 2

Let a,b≥2a,b\ge2 be coprime integers. Prove that every integer N≥(a−1)(b−1)N\ge(a-1)(b-1) can be written as

N=ax+by(x,y nonnegative integers),\begin{gathered}N=ax+by\\(x,y\text{ nonnegative integers}),\end{gathered}

but that ab−a−bab-a-b cannot be written in this form.

Hint 1

The bb numbers 0,a,2a,…,(b−1)a0,a,2a,\ldots,(b-1)a have all possible remainders modulo bb.

Hint 2

Choose xx in 0,…,b−10,\ldots,b-1 so that b∣N−axb\mid N-ax. A divisible number greater than −b-b is nonnegative.

Worked solution 2

Because a,ba,b are coprime, the numbers axax, for x=0,…,b−1x=0,\ldots,b-1, give distinct remainders modulo bb. Equal remainders would imply b∣a(x1−x2)b\mid a(x_1-x_2), hence b∣x1−x2b\mid x_1-x_2, forcing equality in this range. Thus some xx in that range has ax≡N(modb)ax\equiv N\pmod b.

For that xx, the integer N−axN-ax is a multiple of bb, and

N−ax≥(a−1)(b−1)−a(b−1)=1−b>−b.N-ax\ge(a-1)(b-1)-a(b-1)=1-b>-b.

The negative multiple of bb closest to zero is −b-b; every other negative multiple is even smaller. Hence N−ax≥0N-ax\ge0, so y=(N−ax)/by=(N-ax)/b is a nonnegative integer. This constructs the required representation.

Now suppose ab−a−b=ax+byab-a-b=ax+by with x,y≥0x,y\ge0. Rearranging gives

a(x+1)+b(y+1)=ab.a(x+1)+b(y+1)=ab.

Reduction modulo bb and coprimality imply b∣x+1b\mid x+1. Since x+1x+1 is positive, x+1≥bx+1\ge b, so the first term on the left is at least abab. The second term is strictly positive, contradicting the equality. Thus the stated bound cannot be lowered by 1.

Conclusion: Every N≥ab−a−b+1N\ge ab-a-b+1 is representable; ab−a−bab-a-b is not.

Question 3

A rectangular array has mm rows and nn columns, where m,n≥1m,n\ge1, and arbitrary real entries. A move changes the sign of every entry in one row or one column.

Prove that a sequence of moves can make every row sum and every column sum nonnegative. More strongly, prove that any process which always flips a row or column whose current sum is negative must stop after at most 2m+n−12^{m+n}-1 moves.

Hint 1

Use the sum of all entries to measure progress. How much does it change when a negative-sum line is flipped?

Hint 2

Only the parity of the number of flips of each row and column matters, so there are finitely many possible resulting arrays.

Worked solution 3

Let SS be the sum of all entries in the current array. If a selected row or column has sum t<0t<0, flipping it changes that line’s contribution from tt to −t-t, while all other entries stay unchanged. Hence the new total is S−2t>SS-2t>S. Every permitted move strictly increases the total.

Despite the entries being arbitrary real numbers, there are only finitely many configurations. An entry is multiplied by −1-1 each time its row or its column is flipped. Its final sign therefore depends only on whether the number of flips of that row and column is even or odd. There are at most 2m+n2^{m+n} choices of these parities, so at most that many resulting arrays.

A process with strictly increasing total can never revisit an array. Including the starting array, it visits at most 2m+n2^{m+n} different arrays, so it makes at most 2m+n−12^{m+n}-1 moves. When it stops, no negative-sum row or column is left, since otherwise another permitted move would exist. The bound works regardless of which negative line is chosen at each step, and it proves the requested existence as well.

Conclusion: Flipping any negative-sum line always terminates within 2m+n−12^{m+n}-1 moves, with all line sums nonnegative.

Question 4

Let ABCDABCD be a convex quadrilateral, with vertices in counterclockwise order. Construct a square externally on each side AB,BC,CD,DAAB,BC,CD,DA, and call their centres P,Q,R,SP,Q,R,S, respectively. Suppose P≠RP\ne R. Prove that

PR=QSPR⊥QS.\begin{gathered}PR=QS\\ PR\perp QS.\end{gathered}Four external squares on a convex quadrilateral, with their centresABCDPQRSOriginal construction. The proof does not rely on the drawing.

Hint 1

The centre of an external square lies half a side length to the right of its side’s midpoint.

Hint 2

Use coordinate pairs. The operation J(u,v)=(−v,u)J(u,v)=(-v,u) rotates an arrow through 90∘90^\circ. Show that the arrow from QQ to SS is JJ applied to the arrow from PP to RR.

Worked solution 4

We introduce coordinate arrows only to keep track of the four quarter-turns. Write each point as its coordinate pair. Pairs are added and subtracted component by component; for example, B−AB-A is the arrow from AA to BB. Define

J(u,v)=(−v,u).J(u,v)=(-v,u).

This turns an arrow 90∘90^\circ counterclockwise. It preserves its length, and J(J(u,v))=−(u,v)J(J(u,v))=-(u,v). Also J(U+V)=J(U)+J(V)J(U+V)=J(U)+J(V), by checking the two coordinates.

For a counterclockwise quadrilateral, the outside of each directed side is to its right. The centre of the square on ABAB is its midpoint plus half the clockwise-turned side arrow. Therefore

2P=A+B−J(B−A).2P=A+B-J(B-A).

The same rule gives

2Q=B+C−J(C−B),2R=C+D−J(D−C),2S=D+A−J(A−D).\begin{gathered}2Q=B+C-J(C-B),\\ 2R=C+D-J(D-C),\\ 2S=D+A-J(A-D).\end{gathered}

Set U=C−AU=C-A and V=D−BV=D-B. Subtracting the formulas and using the componentwise rules yields

2(R−P)=U+V+J(U)−J(V),2(R-P)=U+V+J(U)-J(V),2(S−Q)=V−U+J(U)+J(V).2(S-Q)=V-U+J(U)+J(V).

Applying JJ to the first right-hand side gives J(U)+J(V)−U+VJ(U)+J(V)-U+V, exactly the second right-hand side. Thus S−Q=J(R−P)S-Q=J(R-P): the second segment’s arrow is a quarter-turn of the first. Since P≠RP\ne R, both arrows are nonzero. A quarter-turn preserves their lengths and makes them perpendicular, proving both claims.

Conclusion: PR=QSPR=QS and the two nonzero segments are perpendicular.

After this paper

Choose one gap in your proof to repair, study the linked idea, and write a complete solution again before the next mock.

Choose another paper · Check your German selection route

Format reference: official organiser information. Questions and explanations are independent practice material.