🇮🇳 INMO · 2015

INMO 2015

Indian National Mathematical Olympiad 2015 — 6 problems with solutions

INMO 2015

Indian National Mathematical Olympiad 2015. 6 problems. Problems sourced from MathNet (CC BY 4.0). Solutions are expert-authored.


Problem 1

Combinatorics Algebra
Problem

Prove that for every positive integer nn there exists a unique ordered pair (a,b)(a, b) of positive integers such that

n=12(a+b1)(a+b2)+an = \frac{1}{2}(a + b - 1)(a + b - 2) + a
Note

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

View Solution

We have to prove that f:N×NNf: \mathbb{N} \times \mathbb{N} \rightarrow \mathbb{N} defined by

f(a,b)=12(a+b1)(a+b2)+a,a,bNf(a, b) = \frac{1}{2}(a + b - 1)(a + b - 2) + a, \quad \forall a, b \in \mathbb{N}

is a bijection. (Note that the right side is a natural number.) To this end define

T(n)=n(n+1)2,nN{0}T(n) = \frac{n(n+1)}{2}, \quad n \in \mathbb{N} \cup \{0\}

An idea of the proof can be obtained by looking at the following table of values of f(a,b)f(a, b) for some small values of a,ba, b. We observe that the nn-th diagonal runs from (1,n)(1, n)-th position to (n,1)(n, 1)-th position and the entries are nn consecutive integers; the first entry in the nn-th diagonal is one more than the last entry of the (n1)(n-1)-th diagonal. For example the first entry in 5-th diagonal is 11 which is one more than the last entry of 4-th diagonal which is 10. Observe that 5-th diagonal starts from 11 and ends with 15 which accounts for 5 consecutive natural numbers. Thus we see that f(n1,1)+1=f(1,n)f(n-1, 1) + 1 = f(1, n). We also observe that the first nn diagonals exhaust all the natural numbers from 1 to T(n)T(n). (Thus a kind of visual bijection is already there. We formally prove the property.)

We first observe that

f(a,b)T(a+b2)=a>0f(a, b) - T(a + b - 2) = a > 0

and

T(a+b1)f(a,b)=(a+b1)(a+b)2(a+b1)(a+b2)2a=b10T(a + b - 1) - f(a, b) = \frac{(a + b - 1)(a + b)}{2} - \frac{(a + b - 1)(a + b - 2)}{2} - a = b - 1 \geq 0

Thus we have

T(a+b2)<f(a,b)=(a+b1)(a+b2)2+aT(a+b1)T(a + b - 2) < f(a, b) = \frac{(a + b - 1)(a + b - 2)}{2} + a \leq T(a + b - 1)

Suppose f(a1,b1)=f(a2,b2)f(a_1, b_1) = f(a_2, b_2). Then the previous observation shows that

T(a1+b12)<f(a1,b1)T(a1+b11)T(a2+b22)<f(a2,b2)T(a2+b21)\begin{aligned} & T(a_1 + b_1 - 2) < f(a_1, b_1) \leq T(a_1 + b_1 - 1) \\ & T(a_2 + b_2 - 2) < f(a_2, b_2) \leq T(a_2 + b_2 - 1) \end{aligned}

Since the sequence T(n)n=0\langle T(n) \rangle_{n=0}^{\infty} is strictly increasing, it follows that a1+b1=a2+b2a_1 + b_1 = a_2 + b_2. But then the relation f(a1,b1)=f(a2,b2)f(a_1, b_1) = f(a_2, b_2) implies that a1=a2a_1 = a_2 and b1=b2b_1 = b_2. Hence ff is one-one.

Let nn be any natural number. Since the sequence T(n)n=0\langle T(n) \rangle_{n=0}^{\infty} is strictly increasing, we can find a natural number kk such that

T(k1)<nT(k)T(k-1) < n \leq T(k)

Equivalently,

(k1)k2<nk(k+1)2\frac{(k-1)k}{2} < n \leq \frac{k(k+1)}{2}

Now set a=nk(k1)2a = n - \frac{k(k-1)}{2} and b=ka+1b = k - a + 1. Observe that a>0a > 0. Now (1) shows that

a=nk(k1)2k(k+1)2k(k1)2=ka = n - \frac{k(k-1)}{2} \leq \frac{k(k+1)}{2} - \frac{k(k-1)}{2} = k

Hence b=ka+11b = k - a + 1 \geq 1. Thus aa and bb are both positive integers and

f(a,b)=12(a+b1)(a+b2)+a=k(k1)2+a=nf(a, b) = \frac{1}{2}(a + b - 1)(a + b - 2) + a = \frac{k(k-1)}{2} + a = n

This shows that every natural number is in the range of ff. Thus ff is also onto. We conclude that ff is a bijection.


Problem 2

Algebra
Problem

Let XX denote the set of all triples (a,b,c)(a, b, c) of integers. Define a function f:XXf: X \rightarrow X by

f(a,b,c)=(a+b+c,ab+bc+ca,abc)f(a, b, c) = (a + b + c, ab + bc + ca, abc)

Find all triples (a,b,c)(a, b, c) in XX such that f(f(a,b,c))=(a,b,c)f(f(a, b, c)) = (a, b, c).

View Solution

We show that the solution set consists of {(t,0,0);tZ}{(1,1,1)}\{(t, 0, 0) ; t \in \mathbb{Z}\} \cup \{(-1, -1, 1)\}. Let us put a+b+c=da + b + c = d, ab+bc+ca=eab + bc + ca = e and abc=fabc = f. The given condition f(f(a,b,c))=(a,b,c)f(f(a, b, c)) = (a, b, c) implies that

d+e+f=a,de+ef+fd=b,def=cd + e + f = a, \quad de + ef + fd = b, \quad def = c

Thus abcdef=fcabcdef = fc and hence either cf=0cf = 0 or abde=1abde = 1.

Case I: Suppose cf=0cf = 0. Then either c=0c = 0 or f=0f = 0. However c=0c = 0 implies f=0f = 0 and vice-versa. Thus we obtain a+b=da + b = d, d+e=ad + e = a, ab=eab = e and de=bde = b. The first two relations give b=eb = -e. Thus e=ab=aee = ab = -ae and de=b=ede = b = -e. We get either e=0e = 0 or a=d=1a = d = -1. If e=0e = 0, then b=0b = 0 and a=d=ta = d = t, say. We get the triple (a,b,c)=(t,0,0)(a, b, c) = (t, 0, 0), where tZt \in \mathbb{Z}. If e0e \neq 0, then a=d=1a = d = -1. But then d+e+f=ad + e + f = a implies that 1+e+0=1-1 + e + 0 = -1 forcing e=0e = 0. Thus we get the solution family (a,b,c)=(t,0,0)(a, b, c) = (t, 0, 0), where tZt \in \mathbb{Z}.

Case II: Suppose cf0cf \neq 0. In this case abde=1abde = 1. Hence either all are equal to 11; or two equal to 11 and the other two equal to 1-1; or all equal to 1-1. Suppose a=b=d=e=1a = b = d = e = 1. Then a+b+c=da + b + c = d shows that c=1c = -1. Similarly f=1f = -1. Hence e=ab+bc+ca=111=1e = ab + bc + ca = 1 - 1 - 1 = -1 contradicting e=1e = 1. Suppose a=b=1a = b = 1 and d=e=1d = e = -1. Then a+b+c=da + b + c = d gives c=3c = -3 and d+e+f=ad + e + f = a gives f=3f = 3. But then f=abc=11(3)=3f = abc = 1 \cdot 1 \cdot (-3) = -3, a contradiction. Similarly a=b=1a = b = -1 and d=e=1d = e = 1 is not possible. If a=1,b=1,d=1,e=1a = 1, b = -1, d = 1, e = -1, then a+b+c=da + b + c = d gives c=1c = 1. Similarly f=1f = 1. But then f=abc=11(1)=1f = abc = 1 \cdot 1 \cdot (-1) = -1 a contradiction. If a=1,b=1,d=1,e=1a = 1, b = -1, d = -1, e = 1, then c=1c = -1 and e=ab+bc+ca=1+11=1e = ab + bc + ca = -1 + 1 - 1 = -1 and a contradiction to e=1e = 1. The symmetry between (a,b,c)(a, b, c) and (d,e,f)(d, e, f) shows that a=1,b=1,d=1,e=1a = -1, b = 1, d = 1, e = -1 is not possible. Finally if a=1,b=1,d=1a = -1, b = 1, d = -1 and e=1e = 1, then c=1c = -1 and f=1f = -1. But then f=abcf = abc is not satisfied. The only case left is that of a,b,d,ea, b, d, e being all equal to 1-1. Then c=1c = 1 and f=1f = 1. It is easy to check that (1,1,1)(-1, -1, 1) is indeed a solution.

Alternatively cf0cf \neq 0 implies that c1|c| \geq 1 and f1|f| \geq 1. Observe that

d22e=a2+b2+c2,a22b=d2+e2+f2d^{2} - 2e = a^{2} + b^{2} + c^{2}, \quad a^{2} - 2b = d^{2} + e^{2} + f^{2}

Adding these two, we get 2(b+e)=b2+c2+e2+f2-2(b + e) = b^{2} + c^{2} + e^{2} + f^{2}. This may be written in the form

(b+1)2+(e+1)2+c2+f22=0(b + 1)^{2} + (e + 1)^{2} + c^{2} + f^{2} - 2 = 0

We conclude that c2+f22c^{2} + f^{2} \leq 2. Using c1|c| \geq 1 and f1|f| \geq 1, we obtain c=1|c| = 1 and f=1|f| = 1, b+1=0b + 1 = 0 and e+1=0e + 1 = 0. Thus b=e=1b = e = -1. Now a+d=d+e+f+a+b+ca + d = d + e + f + a + b + c and this gives b+c+e+f=0b + c + e + f = 0. It follows that c=f=1c = f = 1 and finally a=d=1a = d = -1.


Problem 3

Geometry
Problem

In a cyclic quadrilateral ABCDABCD, AB=aAB = a, BC=bBC = b, CD=cCD = c, ABC=120\angle ABC = 120^\circ, and ABD=30\angle ABD = 30^\circ. Prove that

i) ca+bc \geq a + b;

ii) c+ac+b=cab|\sqrt{c+a} - \sqrt{c+b}| = \sqrt{c-a-b}.

Note

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

View Solution

Applying cosine rule to triangle ABCABC, we get

AC2=a2+b22abcos120=a2+b2+abAC^2 = a^2 + b^2 - 2ab \cos 120^\circ = a^2 + b^2 + ab

Observe that DAC=DBC=12030=90\angle DAC = \angle DBC = 120^\circ - 30^\circ = 90^\circ. Thus we get

c2=AC2cos230=43(a2+b2+ab)c^2 = \frac{AC^2}{\cos^2 30^\circ} = \frac{4}{3}(a^2 + b^2 + ab)

So

c2(a+b)2=43(a2+b2+ab)(a2+b2+2ab)=(ab)230c^2 - (a+b)^2 = \frac{4}{3}(a^2 + b^2 + ab) - (a^2 + b^2 + 2ab) = \frac{(a-b)^2}{3} \geq 0

This proves ca+bc \geq a + b and thus (i) is true.

For proving (ii), consider the product

Q=(α+β+γ)(αβγ)(α+βγ)(αβ+γ)Q = (\alpha + \beta + \gamma)(\alpha - \beta - \gamma)(\alpha + \beta - \gamma)(\alpha - \beta + \gamma)

where α=c+a\alpha = \sqrt{c+a}, β=c+b\beta = \sqrt{c+b} and γ=cab\gamma = \sqrt{c-a-b}. Expanding the product, we get

Q=(c+a)2+(c+b)2+(cab)22(c+a)(c+b)2(c+a)(cab)2(c+b)(cab)=3c2+4a2+4b2+4ab=0\begin{aligned} Q &= (c+a)^2 + (c+b)^2 + (c-a-b)^2 - 2(c+a)(c+b) - 2(c+a)(c-a-b) - 2(c+b)(c-a-b) \\ &= -3c^2 + 4a^2 + 4b^2 + 4ab \\ &= 0 \end{aligned}

Thus at least one of the factors must be equal to 00. Since α+β+γ>0\alpha + \beta + \gamma > 0 and α+βγ>0\alpha + \beta - \gamma > 0, it follows that the product of the remaining two factors is 00. This gives

c+ac+b=caborc+ac+b=cab\sqrt{c+a} - \sqrt{c+b} = \sqrt{c-a-b} \quad \text{or} \quad \sqrt{c+a} - \sqrt{c+b} = -\sqrt{c-a-b}

We conclude that

c+ac+b=cab|\sqrt{c+a} - \sqrt{c+b}| = \sqrt{c-a-b}

Problem 4

Algebra
Problem

a. Prove that if nn is a positive integer such that n40112n \geq 4011^{2}, then there exists an integer ll such that n<l2<(1+12005)nn < l^{2} < \left(1 + \frac{1}{2005}\right) n.

b. Find the smallest positive integer MM for which whenever an integer nn is such that nMn \geq M, there exists an integer ll, such that n<l2<(1+12005)nn < l^{2} < \left(1 + \frac{1}{2005}\right) n.

View Solution

a. Let n40112n \geq 4011^{2} and mNm \in \mathbb{N} be such that m2n<(m+1)2m^{2} \leq n < (m+1)^{2}. Then

(1+12005)n(m+1)2(1+12005)m2(m+1)2=m220052m1=12005(m24010m2005)=12005((m2005)2200522005)12005((40112005)2200522005)=12005(20062200522005)=12005(40112005)=20062005>0\begin{aligned} \left(1+\frac{1}{2005}\right) n - (m+1)^{2} & \geq \left(1+\frac{1}{2005}\right) m^{2} - (m+1)^{2} \\ & = \frac{m^{2}}{2005} - 2m - 1 \\ & = \frac{1}{2005}\left(m^{2} - 4010m - 2005\right) \\ & = \frac{1}{2005}\left((m-2005)^{2} - 2005^{2} - 2005\right) \\ & \geq \frac{1}{2005}\left((4011-2005)^{2} - 2005^{2} - 2005\right) \\ & = \frac{1}{2005}\left(2006^{2} - 2005^{2} - 2005\right) \\ & = \frac{1}{2005}(4011 - 2005) = \frac{2006}{2005} > 0 \end{aligned}

Thus we get

n<(m+1)2<(1+12005)nn < (m+1)^{2} < \left(1 + \frac{1}{2005}\right) n

and l2=(m+1)2l^{2} = (m+1)^{2} is the desired square.

b. We show that M=40102+1M = 4010^{2} + 1 is the required least number. Suppose nMn \geq M. Write n=40102+kn = 4010^{2} + k, where kk is a positive integer. Note that we may assume n<40112n < 4011^{2} by part (a). Now

(1+12005)n40112=(1+12005)(40102+k)40112=40102+24010+k+k200540112=(4010+1)2+(k1)+k200540112=(k1)+k2005>0\begin{aligned} \left(1 + \frac{1}{2005}\right) n - 4011^{2} & = \left(1 + \frac{1}{2005}\right)\left(4010^{2} + k\right) - 4011^{2} \\ & = 4010^{2} + 2 \cdot 4010 + k + \frac{k}{2005} - 4011^{2} \\ & = (4010 + 1)^{2} + (k - 1) + \frac{k}{2005} - 4011^{2} \\ & = (k - 1) + \frac{k}{2005} > 0 \end{aligned}

Thus we obtain

40102<n<40112<(1+12005)n4010^{2} < n < 4011^{2} < \left(1 + \frac{1}{2005}\right) n

We check that M=40102M = 4010^{2} will not work. For suppose n=40102n = 4010^{2}. Then

(1+12005)40102=40102+24010=401121<40112\left(1 + \frac{1}{2005}\right) 4010^{2} = 4010^{2} + 2 \cdot 4010 = 4011^{2} - 1 < 4011^{2}

Thus there is no square integer between nn and (1+12005)n\left(1 + \frac{1}{2005}\right) n. This proves (b).


Problem 5

Geometry
Problem

In a triangle ABCABC right-angled at CC, the median through BB bisects the angle between BABA and the bisector of B\angle B. Prove that

52<ABBC<3\frac{5}{2}<\frac{AB}{BC}<3
Note

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

View Solution

Solution 1.

Since EE is the mid-point of ACAC, we have AE=EC=b/2AE = EC = b/2. Since BDBD bisects ABC\angle ABC, we also know that CD=ab/(a+c)CD = ab/(a+c). Since BEBE bisects ABD\angle ABD, we also have

BD2BA2=DE2EA2\frac{BD^{2}}{BA^{2}} = \frac{DE^{2}}{EA^{2}}

However,

BD2=BC2+CD2=a2+a2b2(a+c)2DE2=(b2aba+c)2\begin{aligned} BD^{2} & = BC^{2} + CD^{2} = a^{2} + \frac{a^{2}b^{2}}{(a+c)^{2}} \\ DE^{2} & = \left(\frac{b}{2} - \frac{ab}{a+c}\right)^{2} \end{aligned}

Using these in the above expression and simplifying, we get

a2{(a+c)2+b2}=c2(ca)2a^{2}\left\{(a+c)^{2}+b^{2}\right\}=c^{2}(c-a)^{2}

Using c2=a2+b2c^{2}=a^{2}+b^{2} and eliminating bb, we obtain

c32ac2a2c2a3=0c^{3}-2ac^{2}-a^{2}c-2a^{3}=0

Introducing t=c/at=c/a, this reduces to a cubic equation;

t32t2t2=0t^{3}-2t^{2}-t-2=0

Consider the function f(t)=t32t2t2f(t)=t^{3}-2t^{2}-t-2 for t>0t>0 (as c/ac/a is positive). For 0<t20<t \leq 2, we see that f(t)=t2(t2)t2<0f(t)=t^{2}(t-2)-t-2<0. We also observe that f(t)=(t2)(t21)4f(t)=(t-2)(t^{2}-1)-4 is strictly increasing on (2,)(2, \infty). It is easy to compute

f(5/2)=118<0,andf(3)=4>0f(5/2)=-\frac{11}{8}<0, \quad \text{and} \quad f(3)=4>0

Hence there is a unique value of tt in the interval (5/2,3)(5/2,3) such that f(t)=0f(t)=0. We conclude that

52<ca<3\frac{5}{2}<\frac{c}{a}<3
View Solution

Solution 2.

Let us take B/4=θ\angle B/4=\theta. Then EBC=DBE=θ\angle EBC=\angle DBE=\theta and CBD=2θ\angle CBD=2\theta. Using sine rule in triangles BEABEA and BECBEC, we get

BEsinA=AEsinθBEsin90=CEsin3θ\begin{aligned} \frac{BE}{\sin A} & = \frac{AE}{\sin \theta} \\ \frac{BE}{\sin 90^{\circ}} & = \frac{CE}{\sin 3\theta} \end{aligned}

Since AE=CEAE=CE, we obtain sin3θsinA=sinθ\sin 3\theta \sin A=\sin \theta. However A=904θA=90^{\circ}-4\theta. Thus we get sin3θcos4θ=sinθ\sin 3\theta \cos 4\theta=\sin \theta. Note that

ca=1cos4θ=sin3θsinθ=34sin2θ\frac{c}{a}=\frac{1}{\cos 4\theta}=\frac{\sin 3\theta}{\sin \theta}=3-4\sin^{2}\theta

This shows that c/a<3c/a<3. Using c/a=34sin2θc/a=3-4\sin^{2}\theta, it is easy to compute cos2θ=((c/a)1)/2\cos 2\theta=((c/a)-1)/2. Hence

ac=cos4θ=12(ca1)21\frac{a}{c}=\cos 4\theta=\frac{1}{2}\left(\frac{c}{a}-1\right)^{2}-1

Suppose c/a5/2c/a \leq 5/2. Then ((c/a)1)29/4((c/a)-1)^{2} \leq 9/4 and a/c2/5a/c \geq 2/5. Thus

25ac=12(ca1)21981=18\frac{2}{5} \leq \frac{a}{c}=\frac{1}{2}\left(\frac{c}{a}-1\right)^{2}-1 \leq \frac{9}{8}-1=\frac{1}{8}

which is absurd. We conclude that c/a>5/2c/a>5/2.


Problem 6

Number Theory Algebra
Problem

Let nn be a natural number such that n=a2+b2+c2n = a^{2} + b^{2} + c^{2}, for some natural numbers a,b,ca, b, c. Prove that

9n=(p1a+q1b+r1c)2+(p2a+q2b+r2c)2+(p3a+q3b+r3c)29 n = (p_{1} a + q_{1} b + r_{1} c)^{2} + (p_{2} a + q_{2} b + r_{2} c)^{2} + (p_{3} a + q_{3} b + r_{3} c)^{2}

where pjp_{j}‘s, qjq_{j}‘s, rjr_{j}‘s are all nonzero integers. Further, if 33 does not divide at least one of a,b,ca, b, c, prove that 9n9 n can be expressed in the form x2+y2+z2x^{2} + y^{2} + z^{2}, where x,y,zx, y, z are natural numbers none of which is divisible by 33.

View Solution

It can be easily seen that

9n=(2b+2ca)2+(2c+2ab)2+(2a+2bc)29 n = (2 b + 2 c - a)^{2} + (2 c + 2 a - b)^{2} + (2 a + 2 b - c)^{2}

Thus we can take p1=p2=p3=2p_{1} = p_{2} = p_{3} = 2, q1=q2=q3=2q_{1} = q_{2} = q_{3} = 2 and r1=r2=r3=1r_{1} = r_{2} = r_{3} = -1. Suppose 33 does not divide gcd(a,b,c)\gcd(a, b, c). Then 33 does divide at least one of a,b,ca, b, c; say 33 does not divide aa. Note that each of 2b+2ca2 b + 2 c - a, 2c+2ab2 c + 2 a - b and 2a+2bc2 a + 2 b - c is either divisible by 33 or none of them is divisible by 33, as the difference of any two sums is always divisible by 33. If 33 does not divide 2b+2ca2 b + 2 c - a, then we have the required representation. If 33 divides 2b+2ca2 b + 2 c - a, then 33 does not divide 2b+2c+a2 b + 2 c + a. On the other hand, we also note that

9n=(2b+2c+a)2+(2c2ab)2+(2a+2bc)2=x2+y2+z29 n = (2 b + 2 c + a)^{2} + (2 c - 2 a - b)^{2} + (-2 a + 2 b - c)^{2} = x^{2} + y^{2} + z^{2}

where x=2b+2c+ax = 2 b + 2 c + a, y=2c2aby = 2 c - 2 a - b and z=2a+2bcz = -2 a + 2 b - c. Since xy=3(b+a)x - y = 3(b + a) and 33 does not divide xx, it follows that 33 does not divide yy as well. Similarly, we conclude that 33 does not divide zz.