🇮🇳 INMO · 2021

INMO 2021

Indian National Mathematical Olympiad 2021 — 6 problems with solutions

INMO 2021

Indian National Mathematical Olympiad 2021. 6 problems. Problems sourced from MathNet (CC BY 4.0). Solutions are expert-authored.


Problem 1

Algebra
Problem

Find all real functions ff from RR\mathbb{R} \rightarrow \mathbb{R} satisfying the relation

f(x2+yf(x))=xf(x+y)f\left(x^{2}+y f(x)\right)=x f(x+y)
View Solution

Put x=0x=0 and we get f(yf(0))=0f(y f(0))=0. If f(0)0f(0) \neq 0, then yf(0)y f(0) takes all real values when yy varies over real line. We get f(x)0f(x) \equiv 0. Suppose f(0)=0f(0)=0. Taking y=xy=-x, we get f(x2xf(x))=0f\left(x^{2}-x f(x)\right)=0 for all real xx.

Suppose there exists x00x_{0} \neq 0 in R\mathbb{R} such that f(x0)=0f\left(x_{0}\right)=0. Putting x=x0x=x_{0} in the given relation we get

f(x02)=x0f(x0+y)f\left(x_{0}^{2}\right)=x_{0} f\left(x_{0}+y\right)

for all yRy \in \mathbb{R}. Now the left side is a constant and hence it follows that ff is a constant function. But the only constant function which satisfies the equation is identically zero function, which is already obtained. Hence we may consider the case where f(x)0f(x) \neq 0 for all x0x \neq 0.

Since f(x2xf(x))=0f\left(x^{2}-x f(x)\right)=0, we conclude that x2xf(x)=0x^{2}-x f(x)=0 for all x0x \neq 0. This implies that f(x)=xf(x)=x for all x0x \neq 0. Since f(0)=0f(0)=0, we conclude that f(x)=xf(x)=x for all xRx \in \mathbb{R}. Thus we have two functions: f(x)0f(x) \equiv 0 and f(x)=xf(x)=x for all xRx \in \mathbb{R}.


Problem 2

Combinatorics
Problem

There are four basket-ball players AA, BB, CC, DD. Initially, the ball is with AA. The ball is always passed from one person to a different person. In how many ways can the ball come back to AA after seven passes? (For example ACBDABCAA \rightarrow C \rightarrow B \rightarrow D \rightarrow A \rightarrow B \rightarrow C \rightarrow A and ADADCABAA \rightarrow D \rightarrow A \rightarrow D \rightarrow C \rightarrow A \rightarrow B \rightarrow A are two ways in which the ball can come back to AA after seven passes.)

View Solution

Let xnx_{n} be the number of ways in which AA can get back the ball after nn passes. Let yny_{n} be the number of ways in which the ball goes back to a fixed person other than AA after nn passes. Then

xn=3yn1,x_{n}=3 y_{n-1},

and

yn=xn1+2yn1y_{n}=x_{n-1}+2 y_{n-1}

We also have x1=0x_{1}=0, x2=3x_{2}=3, y1=1y_{1}=1 and y2=2y_{2}=2.

Eliminating yny_{n} and yn1y_{n-1}, we get xn+1=3xn1+2xnx_{n+1}=3 x_{n-1}+2 x_{n}. Thus

x3=3x1+2x2=2×3=6x4=3x2+2x3=(3×3)+(2×6)=9+12=21x5=3x3+2x4=(3×6)+(2×21)=18+42=60x6=3x4+2x5=(3×21)+(2×60)=63+120=183x7=3x5+2x6=(3×60)+(2×183)=180+366=546\begin{aligned} & x_{3}=3 x_{1}+2 x_{2}=2 \times 3=6 \\ & x_{4}=3 x_{2}+2 x_{3}=(3 \times 3)+(2 \times 6)=9+12=21 \\ & x_{5}=3 x_{3}+2 x_{4}=(3 \times 6)+(2 \times 21)=18+42=60 \\ & x_{6}=3 x_{4}+2 x_{5}=(3 \times 21)+(2 \times 60)=63+120=183 \\ & x_{7}=3 x_{5}+2 x_{6}=(3 \times 60)+(2 \times 183)=180+366=546 \end{aligned}

Alternate solution: Since the ball goes back to one of the other 3 persons, we have

xn+3yn=3nx_{n}+3 y_{n}=3^{n}

since there are 3n3^{n} ways of passing the ball in nn passes. Using xn=3yn1x_{n}=3 y_{n-1}, we obtain

xn1+xn=3n1x_{n-1}+x_{n}=3^{n-1}

with x1=0x_{1}=0. Thus

x7=36x6=3635+x5=3635+34x4=3635+3433+x3=3635+3433+32x2=3635+3433+323=(2×35)+(2×33)+(2×3)=486+54+6=546\begin{array}{r} x_{7}=3^{6}-x_{6}=3^{6}-3^{5}+x_{5}=3^{6}-3^{5}+3^{4}-x_{4}=3^{6}-3^{5}+3^{4}-3^{3}+x_{3} \\ =3^{6}-3^{5}+3^{4}-3^{3}+3^{2}-x_{2}=3^{6}-3^{5}+3^{4}-3^{3}+3^{2}-3 \\ =\left(2 \times 3^{5}\right)+\left(2 \times 3^{3}\right)+(2 \times 3)=486+54+6=546 \end{array}

Problem 3

Geometry
Problem

Let ABCDABCD be a convex quadrilateral. Let the diagonals ACAC and BDBD intersect in PP. Let PEPE, PFPF, PGPG and PHPH be the altitudes from PP onto the sides ABAB, BCBC, CDCD and DADA respectively. Show that ABCDABCD has an incircle if and only if

1PE+1PG=1PF+1PH\frac{1}{PE} + \frac{1}{PG} = \frac{1}{PF} + \frac{1}{PH}
Note

This problem originally includes a geometry diagram. The diagram is omitted in this version.

View Solution

Let AP=pAP = p, BP=qBP = q, CP=rCP = r, DP=sDP = s; AB=aAB = a, BC=bBC = b, CD=cCD = c and DA=dDA = d. Let APB=CPD=θ\angle APB = \angle CPD = \theta. Then BPC=DPA=πθ\angle BPC = \angle DPA = \pi - \theta. Let us also write PE=h1PE = h_1, PF=h2PF = h_2, PG=h3PG = h_3 and PH=h4PH = h_4.

Observe that

h1a=pqsinθ,h2b=qrsinθ,h3c=rssinθ,h4d=spsinθh_1 a = p q \sin \theta, \quad h_2 b = q r \sin \theta, \quad h_3 c = r s \sin \theta, \quad h_4 d = s p \sin \theta

Hence

1h1+1h3=1h2+1h4\frac{1}{h_1} + \frac{1}{h_3} = \frac{1}{h_2} + \frac{1}{h_4}

is equivalent to

apq+crs=bqr+dsp\frac{a}{p q} + \frac{c}{r s} = \frac{b}{q r} + \frac{d}{s p}

This is the same as

ars+cpq=bsp+dqra r s + c p q = b s p + d q r

Thus we have to prove that a+c=b+da + c = b + d if and only if ars+cpq=bsp+dqra r s + c p q = b s p + d q r. Now we can write a+c=b+da + c = b + d as

a2+c2+2ac=b2+d2+2bda^2 + c^2 + 2 a c = b^2 + d^2 + 2 b d

But we know that

a2=p2+q22pqcosθ,c2=r2+s22rscosθb2=q2+r2+2qrcosθ,d2=p2+s2+2pscosθ\begin{aligned} & a^2 = p^2 + q^2 - 2 p q \cos \theta, \quad c^2 = r^2 + s^2 - 2 r s \cos \theta \\ & b^2 = q^2 + r^2 + 2 q r \cos \theta, \quad d^2 = p^2 + s^2 + 2 p s \cos \theta \end{aligned}

Hence a+c=b+da + c = b + d is equivalent to

pqcosθrscosθ+ac=pscosθ+qrcosθ+bd- p q \cos \theta - r s \cos \theta + a c = p s \cos \theta + q r \cos \theta + b d

Similarly, by squaring ars+cpq=bsp+dqra r s + c p q = b s p + d q r we can show that it is equivalent to

pqcosθrscosθ+ac=pscosθ+qrcosθ+bd- p q \cos \theta - r s \cos \theta + a c = p s \cos \theta + q r \cos \theta + b d

We conclude that a+c=b+da + c = b + d is equivalent to cpq+ars=bps+dqrc p q + a r s = b p s + d q r. Hence ABCDABCD has an incircle if and only if

1h1+1h3=1h2+1h4\frac{1}{h_1} + \frac{1}{h_3} = \frac{1}{h_2} + \frac{1}{h_4}

Problem 4

Number Theory Combinatorics
Problem

From a set of 11 square integers, show that one can choose 6 numbers a2,b2,c2,d2,e2,f2a^{2}, b^{2}, c^{2}, d^{2}, e^{2}, f^{2} such that

a2+b2+c2d2+e2+f2(mod12)a^{2}+b^{2}+c^{2} \equiv d^{2}+e^{2}+f^{2} \quad(\bmod 12)
View Solution

The first observation is that we can find 5 pairs of squares such that the two numbers in a pair have the same parity. We can see this as follows:

Odd numbersEven numbersOdd pairsEven pairsTotal pairs
011055
110055
29145
38145
47235
56235
65325
74325
83415
92415
101505
110505

Let us take such 5 pairs: say (x12,y12),(x22,y22),,(x52,y52)\left(x_{1}^{2}, y_{1}^{2}\right),\left(x_{2}^{2}, y_{2}^{2}\right), \ldots,\left(x_{5}^{2}, y_{5}^{2}\right). Then xj2yj2x_{j}^{2}-y_{j}^{2} is divisible by 4 for 1j51 \leq j \leq 5. Let rjr_{j} be the remainder when xj2yj2x_{j}^{2}-y_{j}^{2} is divisible by 3,1j53, 1 \leq j \leq 5. We have 5 remainders r1,r2,r3,r4,r5r_{1}, r_{2}, r_{3}, r_{4}, r_{5}. But these can be 0, 1 or 2. Hence either one of the remainders occurs 3 times or each of the remainders occurs once. If, for example, r1=r2=r3r_{1}=r_{2}=r_{3}, then 3 divides r1+r2+r3r_{1}+r_{2}+r_{3}; if r1=0,r2=1r_{1}=0, r_{2}=1 and r3=2r_{3}=2, then again 3 divides r1+r2+r3r_{1}+r_{2}+r_{3}. Thus we can always find three remainders whose sum is divisible by 3. This means we can find 3 pairs, say, (x12,y12),(x22,y22),(x32,y32)\left(x_{1}^{2}, y_{1}^{2}\right),\left(x_{2}^{2}, y_{2}^{2}\right),\left(x_{3}^{2}, y_{3}^{2}\right) such that 3 divides (x12y12)+(x22y22)+(x32y32)\left(x_{1}^{2}-y_{1}^{2}\right)+\left(x_{2}^{2}-y_{2}^{2}\right)+\left(x_{3}^{2}-y_{3}^{2}\right). Since each difference is divisible by 4, we conclude that we can find 6 numbers a2,b2,c2,d2,e2,f2a^{2}, b^{2}, c^{2}, d^{2}, e^{2}, f^{2} such that

a2+b2+c2d2+e2+f2(mod12)a^{2}+b^{2}+c^{2} \equiv d^{2}+e^{2}+f^{2} \quad(\bmod 12)

Problem 5

Geometry
Problem

Let Γ1\Gamma_{1} and Γ2\Gamma_{2} be two circles of unequal radii, with centres O1O_{1} and O2O_{2} respectively, in the plane intersecting in two distinct points AA and BB. Assume that the centre of each of the circles Γ1\Gamma_{1} and Γ2\Gamma_{2} is outside the other. The tangent to Γ1\Gamma_{1} at BB intersects Γ2\Gamma_{2} again in CC, different from BB; the tangent to Γ2\Gamma_{2} at BB intersects Γ1\Gamma_{1} again in DD, different from BB. The bisectors of DAB\angle D A B and CAB\angle C A B meet Γ1\Gamma_{1} and Γ2\Gamma_{2} again in XX and YY, respectively, different from AA. Let PP and QQ be the circumcentres of triangles ACDA C D and XAYX A Y, respectively. Prove that PQP Q is the perpendicular bisector of the line segment O1O2O_{1} O_{2}.

Note

This problem originally includes a geometry diagram. The diagram is omitted in this version.

View Solution

Let CBA=α\angle C B A=\alpha and DBA=β\angle D B A=\beta. Then BDA=α\angle B D A=\alpha and BCA=β\angle B C A=\beta. We also observe that AO1O2=(AO1B/2)=α\angle A O_{1} O_{2}=\left(\angle A O_{1} B / 2\right)=\alpha and, simiarly, AO2O1=β\angle A O_{2} O_{1}=\beta. Hence

O1AO2=180(α+β)\angle O_{1} A O_{2}=180^{\circ}-(\alpha+\beta)

We also have

PO1A=DO1A2=2DBA2=DBA=β\angle P O_{1} A=\frac{\angle D O_{1} A}{2}=\frac{2 \angle D B A}{2}=\angle D B A=\beta

Hence PO1O2=PO1A+AO1O2=β+α\angle P O_{1} O_{2}=\angle P O_{1} A+\angle A O_{1} O_{2}=\beta+\alpha. Similarly, we can get PO2O1=α+β\angle P O_{2} O_{1}=\alpha+\beta. It follows that PP lies on the perpendicular bisector of O1O2O_{1} O_{2}.

Now we observe that

XQY=3602XAY=3602(180αβ)=2(α+β)\angle X Q Y=360^{\circ}-2 \angle X A Y=360^{\circ}-2\left(180^{\circ}-\alpha-\beta\right)=2(\alpha+\beta)

This gives

O1QO2=12(XQA+YQA)=XQY2=α+β\angle O_{1} Q O_{2}=\frac{1}{2}(\angle X Q A+\angle Y Q A)=\frac{\angle X Q Y}{2}=\alpha+\beta

This shows that A,O1,O2,QA, O_{1}, O_{2}, Q are concyclic. We also have

ABX=ABD+DBX=β+DAX=β+DAB2ABY=ABC+CBY=α+CAY=α+BAC2\begin{aligned} & \angle A B X=\angle A B D+\angle D B X=\beta+\angle D A X=\beta+\frac{\angle D A B}{2} \\ & \angle A B Y=\angle A B C+\angle C B Y=\alpha+\angle C A Y=\alpha+\frac{\angle B A C}{2} \end{aligned}

Adding we obtain

ABX+ABY=α+β+12(DAB+BAC)=α+β+(180αβ)=180\angle A B X+\angle A B Y=\alpha+\beta+\frac{1}{2}(\angle D A B+\angle B A C)=\alpha+\beta+\left(180^{\circ}-\alpha-\beta\right)=180^{\circ}

Hence X,B,YX, B, Y are collinear. Now

QAX=12(180AQX)=90βXAO1=12(180XO1A)=9012(3602ABX)=ABX90\begin{gathered} \angle Q A X=\frac{1}{2}\left(180^{\circ}-\angle A Q X\right)=90^{\circ}-\beta \\ \angle X A O_{1}=\frac{1}{2}\left(180^{\circ}-\angle X O_{1} A\right)=90^{\circ}-\frac{1}{2}\left(360^{\circ}-2 \angle A B X\right)=\angle A B X-90^{\circ} \end{gathered}

Hence

QAO1=90β+ABX90=ABXβ=DAB2=O1AO22\angle Q A O_{1}=90^{\circ}-\beta+\angle A B X-90^{\circ}=\angle A B X-\beta=\frac{\angle D A B}{2}=\frac{\angle O_{1} A O_{2}}{2}

This shows that AQA Q bisects O1AO2\angle O_{1} A O_{2} and therefore the chords QO1Q O_{1} and QO2Q O_{2} subtend equal angles on the circumference of the circle passing through QO2AO1Q O_{2} A O_{1}. Hence QO2=QO1Q O_{2}=Q O_{1}. This means QQ lies on the perpendicular bisector of O1O2O_{1} O_{2}.

Combining, we get that PQP Q is the perpendicular bisector of O1O2O_{1} O_{2}.


Problem 6

Algebra
Problem

Suppose P(x)P(x) is a polynomial with real coefficients satisfying the condition P(cosθ+sinθ)=P(cosθsinθ)P(\cos \theta + \sin \theta) = P(\cos \theta - \sin \theta), for every real θ\theta. Prove that P(x)P(x) can be expressed in the form

P(x)=a0+a1(1x2)2+a2(1x2)4++an(1x2)2nP(x) = a_{0} + a_{1} (1 - x^{2})^{2} + a_{2} (1 - x^{2})^{4} + \cdots + a_{n} (1 - x^{2})^{2n}

for some real numbers a0,a1,a2,,ana_{0}, a_{1}, a_{2}, \ldots, a_{n} and nonnegative integer nn.

View Solution

Changing θ\theta to θπ/2\theta - \pi / 2, we see that

P(sinθ+cosθ)=P(sinθcosθ)P(\sin \theta + \cos \theta) = P(\sin \theta - \cos \theta)

This shows that P(x)=P(x)P(x) = P(-x) for all x[2,2]x \in [-\sqrt{2}, \sqrt{2}] and as PP is a polynomial, in fact,

P(x)=P(x)P(x) = P(-x)

for all xRx \in \mathbb{R}. Hence P(x)P(x) is an even polynomial; P(x)=Q(x2)P(x) = Q(x^{2}) for some polynomial Q(x)Q(x). This gives

Q(1+sin(2θ))=P(cosθ+sinθ)=P(cosθsinθ)=Q(1sin(2θ))Q(1 + \sin(2\theta)) = P(\cos \theta + \sin \theta) = P(\cos \theta - \sin \theta) = Q(1 - \sin(2\theta))

Taking t=sin(2θ)t = \sin(2\theta), we see that Q(1+t)=Q(1t)Q(1 + t) = Q(1 - t). Hence Q(0)=Q(2)Q(0) = Q(2).

Consider Q(t)Q(0)Q(t) - Q(0). This vanishes both at t=0t = 0 and t=2t = 2. Hence t(2t)t(2 - t) is a factor of Q(t)Q(0)Q(t) - Q(0). We obtain

Q(t)Q(0)=t(2t)h(t)Q(t) - Q(0) = t(2 - t) h(t)

for some polynomial h(t)h(t). Using Q(1+t)=Q(1t)Q(1 + t) = Q(1 - t), it follows that h(1+t)=h(1t)h(1 + t) = h(1 - t). Hence by induction we get

Q(t)=k=0nbktk(2t)kQ(t) = \sum_{k=0}^{n} b_{k} t^{k} (2 - t)^{k}

Hence

P(x)=Q(x2)=k=0nbk(x2(2x2))k=k=0nbk(1(1x2)2)kP(x) = Q(x^{2}) = \sum_{k=0}^{n} b_{k} (x^{2}(2 - x^{2}))^{k} = \sum_{k=0}^{n} b_{k} (1 - (1 - x^{2})^{2})^{k}

Using binomial theorem, we can write this as

P(x)=k=0nak(1x2)2kP(x) = \sum_{k=0}^{n} a_{k} (1 - x^{2})^{2k}

for some coefficients aka_{k}, 0kn0 \leq k \leq n.