Library · Between the Challenge and the Olympiad · Chapter 5

Polynomials and Equations

Revised Report an error
On this page
  1. Problems
  2. Solutions
  3. Solution: PRMO 2012, Q1
  4. Solution: PRMO 2013, Q3
  5. Solution: PRMO 2015 Part A, Q1
  6. Solution: PRMO 2015 Part A, Q15
  7. Solution: PRMO 2017, Q3
  8. Solution: IOQM 2025 Part SEP, Q7
  9. Solution: PRMO 2013, Q11
  10. Solution: PRMO 2014, Q5
  11. Solution: PRMO 2014, Q6
  12. Solution: PRMO 2014, Q7
  13. Solution: PRMO 2014, Q8
  14. Solution: PRMO 2015 Part A, Q3
  15. Solution: PRMO 2017, Q4
  16. Solution: PRMO 2017, Q19
  17. Solution: PRMO 2018, Q1
  18. Solution: PRMO 2019, Q2
  19. Solution: IOQM 2023, Q2
  20. Solution: IOQM 2024, Q5
  21. Solution: IOQM 2024, Q16
  22. Solution: IOQM 2025 Part SEP, Q28
  23. Solution: PRMO 2012, Q12
  24. Solution: PRMO 2012, Q13
  25. Solution: PRMO 2012, Q17
  26. Solution: PRMO 2014, Q9
  27. Solution: PRMO 2015 Part A, Q12
  28. Solution: PRMO 2015 Part A, Q18
  29. Solution: PRMO 2017, Q2
  30. Solution: PRMO 2018, Q9
  31. Solution: PRMO 2019, Q8
  32. Solution: IOQM 2023, Q12
  33. Solution: IOQM 2024, Q10
  34. Solution: IOQM 2026, Q11
  35. Solution: PRMO 2012, Q18
  36. Solution: PRMO 2013, Q16
  37. Solution: IOQM 2020, Q11
  38. Solution: IOQM 2020, Q21
  39. Solution: IOQM 2021 Part A, Q10
  40. Solution: IOQM 2022, Q14
  41. Solution: IOQM 2022, Q16
  42. Solution: IOQM 2026, Q21

Problems

Problem 1

Rachel was asked by her teacher to subtract 3 from a certain number and then divide the result by 9. Instead, she subtracted 9 and then divided the result by 3. She got 43 as the answer. What would have been her answer if she had solved the problem correctly?

Problem 2

It is given that the equation x2+ax+20=0x^2 + ax + 20 = 0 has integer roots. What is the sum of all possible values of aa?

Problem 3

A man walks a certain distance and rides back in 3343\frac{3}{4} hours; he could ride both ways in 2122\frac{1}{2} hours. How many hours would it take him to walk both ways?

Problem 4

If 3x+2y=9853^x + 2^y = 985 and 3x−2y=4733^x - 2^y = 473, what is the value of xyxy?

Problem 5

A contractor has two teams of workers: team AA and team BB. Team AA can complete a job in 12 days and team BB can do the same job in 36 days. Team AA starts working on the job and team BB joins team AA after four days. The team AA withdraws after two more days. For how many more days should team BB work to complete the job?

Problem 6

If 60%60\% of a number xx is 4040, then what is x%x\% of 6060?

Problem 7

Three real numbers xx, yy, zz are such that x2+6y=−17x^2 + 6y = -17, y2+4z=1y^2 + 4z = 1 and z2+2x=2z^2 + 2x = 2. What is the value of x2+y2+z2x^2 + y^2 + z^2?

Problem 8

If real numbers aa, bb, cc, dd, ee satisfy a+1=b+2=c+3=d+4=e+5=a+b+c+d+e+3,a + 1 = b + 2 = c + 3 = d + 4 = e + 5 = a + b + c + d + e + 3, what is the value of a2+b2+c2+d2+e2a^2 + b^2 + c^2 + d^2 + e^2?

Problem 9

What is the smallest possible natural number nn for which the equation x2−nx+2014=0x^2 - nx + 2014 = 0 has integer roots?

Problem 10

If x(x4)=4x^{(x^4)} = 4, what is the value of x(x2)+x(x8)x^{(x^2)} + x^{(x^8)}?

Problem 11

Let SS be a set of real numbers with mean MM. If the means of the sets S∪{15}S \cup \{15\} and S∪{15,1}S \cup \{15, 1\} are M+2M + 2 and M+1M + 1, respectively, then how many elements does SS have?

Problem 12

The equations x2−4x+k=0x^2 - 4x + k = 0 and x2+kx−4=0x^2 + kx - 4 = 0, where kk is a real number, have exactly one common root. What is the value of kk?

Problem 13

Let a,ba, b be integers such that all the roots of the equation (x2+ax+20)(x2+17x+b)=0(x^2 + ax + 20)(x^2 + 17x + b) = 0 are negative integers. What is the smallest possible value of a+ba + b ?

Problem 14

Suppose 1,2,31, 2, 3 are the roots of the equation x4+ax2+bx=cx^4 + ax^2 + bx = c. Find the value of cc.

Problem 15

A book is published in three volumes, the pages being numbered from 1 onwards. The page numbers are continued from the first volume to the second volume to the third. The number of pages in the second volume is 50 more than that in the first volume, and the number of pages in the third volume is one and a half times that in the second. The sum of the page numbers on the first pages of the three volumes is 1709. If nn is the last page number, what is the largest prime factor of nn?

Problem 16

Let f(x)=x2+ax+bf(x) = x^2 + ax + b. If for all nonzero real xx f(x+1x)=f(x)+f(1x)f\left(x + \frac{1}{x}\right) = f(x) + f\left(\frac{1}{x}\right) and the roots of f(x)=0f(x) = 0 are integers, what is the value of a2+b2a^2 + b^2?

Problem 17

Find the number of elements in the set {(a,b)∈N:2≤a,b≤2023,log⁡a(b)+6log⁡b(a)=5}\{(a,b) \in \mathbb{N} : 2 \leq a, b \leq 2023, \log_a(b) + 6\log_b(a) = 5\}

Problem 18

Let a=xy+yz+zxa = \dfrac{x}{y} + \dfrac{y}{z} + \dfrac{z}{x}, let b=xz+yx+zyb = \dfrac{x}{z} + \dfrac{y}{x} + \dfrac{z}{y} and let c=(xy+yz)(yz+zx)(zx+xy)c = \left(\dfrac{x}{y} + \dfrac{y}{z}\right)\left(\dfrac{y}{z} + \dfrac{z}{x}\right)\left(\dfrac{z}{x} + \dfrac{x}{y}\right). The value of ∣ab−c∣|ab - c| is:

Problem 19

Let f:R→Rf : \mathbb{R} \to \mathbb{R} be a function satisfying the relation 4f(3−x)+3f(x)=x24f(3-x) + 3f(x) = x^2 for any real xx. Find the value of f(27)−f(25)f(27) - f(25) to the nearest integer. (Here R\mathbb{R} denotes the set of real numbers.)

Problem 20

A function is defined on the set of positive integers such that if nn is an odd integer, f(n)=n−1f(n) = n - 1 and if nn is an even integer, f(n)=n2−1f(n) = n^2 - 1. Determine the sum of all possible values of nn such that f(f(n))=99f(f(n)) = 99.

Problem 21

If 12011+20112−1=m−n\dfrac{1}{\sqrt{2011 + \sqrt{2011^2 - 1}}} = \sqrt{m} - \sqrt{n}, where mm and nn are positive integers, what is the value of m+nm + n?

Problem 22

If a=b−ca = b - c, b=c−db = c - d, c=d−ac = d - a and abcd≠0abcd \neq 0 then what is the value of ab+bc+cd+da\dfrac{a}{b} + \dfrac{b}{c} + \dfrac{c}{d} + \dfrac{d}{a}?

Problem 23

Let x1,x2,x3x_1, x_2, x_3 be the roots of the equation x3+3x+5=0x^3 + 3x + 5 = 0. What is the value of the expression (x1+1x1)(x2+1x2)(x3+1x3)?\left(x_1 + \frac{1}{x_1}\right)\left(x_2 + \frac{1}{x_2}\right)\left(x_3 + \frac{1}{x_3}\right)?

Problem 24

Natural numbers k,l,pk, l, p and qq are such that if aa and bb are roots of x2−kx+l=0x^2 - kx + l = 0 then a+1ba + \frac{1}{b} and b+1ab + \frac{1}{a} are the roots of x2−px+q=0x^2 - px + q = 0. What is the sum of all possible values of qq?

Problem 25

Let aa, bb, and cc be real numbers such that a−7b+8c=4a - 7b + 8c = 4 and 8a+4b−c=78a + 4b - c = 7. What is the value of a2−b2+c2a^2 - b^2 + c^2?

Problem 26

Let aa, bb and cc be such that a+b+c=0a + b + c = 0 and P=a22a2+bc+b22b2+ca+c22c2+abP = \frac{a^2}{2a^2 + bc} + \frac{b^2}{2b^2 + ca} + \frac{c^2}{2c^2 + ab} is defined. What is the value of PP?

Problem 27

Suppose a,ba, b are positive real numbers such that aa+bb=183a\sqrt{a} + b\sqrt{b} = 183, ab+ba=182\quad a\sqrt{b} + b\sqrt{a} = 182. Find 95(a+b)\frac{9}{5}(a+b).

Problem 28

Suppose a,ba, b are integers and a+ba + b is a root of x2+ax+b=0x^2 + ax + b = 0. What is the maximum possible value of b2b^2?

Problem 29

How many positive integers nn are there such that 3≤n≤1003 \leq n \leq 100 and x2n+x+1x^{2^n} + x + 1 is divisible by x2+x+1x^2 + x + 1?

Problem 30

Let P(x)=x3+ax2+bx+cP(x) = x^3 + ax^2 + bx + c be a polynomial where a,b,ca, b, c are integers and cc is odd. Let pip_i be the value of P(x)P(x) at x=ix = i. Given that p13+p23+p33=3p1p2p3p_1^3 + p_2^3 + p_3^3 = 3p_1p_2p_3, find the value of p2+2p1−3p0p_2 + 2p_1 - 3p_0.

Problem 31

Determine the number of positive integral values of pp for which there exists a triangle with sides a,ba, b, and cc which satisfy a2+(p2+9)b2+9c2−6ab−6pbc=0.a^2 + (p^2 + 9)b^2 + 9c^2 - 6ab - 6pbc = 0.

Problem 32

A function ff is defined on the set of integers such that for any two integers mm and nn, f(mn+1)=f(m)f(n)−f(n)−m+2f(mn+1) = f(m)f(n) - f(n) - m + 2 holds and f(0)=1f(0) = 1. Determine the largest positive integer NN such that ∑k=1Nf(k)<100\sum_{k=1}^{N} f(k) < 100.

Problem 33

Let f(x)f(x) and g(x)g(x) be two polynomials of degree 2 such that f(−2)g(−2)=f(3)g(3)=4.\frac{f(-2)}{g(-2)} = \frac{f(3)}{g(3)} = 4. If g(5)=2,f(7)=12,g(7)=−6g(5) = 2, f(7) = 12, g(7) = -6, what is the value of f(5)f(5)?

Problem 34

Find the number of non-constant polynomials P(x)P(x), with real coefficients, such that P(x2)=P(P(x))P(x^2) = P(P(x))

Problem 35

What is the sum of the squares of the roots of the equation x2−7[x]+5=0x^2 - 7[x] + 5 = 0? (Here [x][x] denotes the greatest integer less than or equal to xx. For example [3.4]=3[3.4] = 3 and [−2.3]=−3[-2.3] = -3.)

Problem 36

Let f(x)=x3−3x+bf(x) = x^3 - 3x + b and g(x)=x2+bx−3g(x) = x^2 + bx - 3, where bb is a real number. What is the sum of all possible values of bb for which the equations f(x)=0f(x) = 0 and g(x)=0g(x) = 0 have a common root?

Problem 37

Let X={−5,−4,−3,−2,−1,0,1,2,3,4,5}X = \{-5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5\} and S={(a,b)∈X×X:x2+ax+b and x3+bx+aS = \{(a,b) \in X \times X : x^2 + ax + b \text{ and } x^3 + bx + a have at least a common real zero}.\text{have at least a common real zero}\}. How many elements are there in SS?

Problem 38

A total fixed amount of NN thousand pounds is given to three persons A,B,CA, B, C, every year, each being given an amount proportional to her age. In the first year, AA got half the total amount. When the sixth payment was made, AA got six-seventh of the amount that she had in the first year; BB got £1000 less than that she had in the first year; and CC got twice of that she had in the first year. Find NN.

Problem 39

Suppose that PP is the polynomial of least degree with integer coefficients such that P(7+5)=2(7−5)P\left(\sqrt{7} + \sqrt{5}\right) = 2\left(\sqrt{7} - \sqrt{5}\right). Find P(2)P(2).

Problem 40

Let x,y,zx, y, z be complex numbers such that xy+z+yz+x+zx+y=9\frac{x}{y+z} + \frac{y}{z+x} + \frac{z}{x+y} = 9 x2y+z+y2z+x+z2x+y=64\frac{x^2}{y+z} + \frac{y^2}{z+x} + \frac{z^2}{x+y} = 64 x3y+z+y3z+x+z3x+y=488\frac{x^3}{y+z} + \frac{y^3}{z+x} + \frac{z^3}{x+y} = 488 If xyz+yzx+zxy=mn\dfrac{x}{yz} + \dfrac{y}{zx} + \dfrac{z}{xy} = \dfrac{m}{n} where m,nm, n are positive integers with GCD(m,n)=1\mathrm{GCD}(m,n) = 1, find m+nm + n.

Problem 41

Let a,b,ca, b, c be reals satisfying 3ab+2=6b,3bc+2=5c,3ca+2=4a.3ab + 2 = 6b, \qquad 3bc + 2 = 5c, \qquad 3ca + 2 = 4a. Let Q\mathbb{Q} denote the set of all rational numbers. Given that the product abcabc can take two values rs∈Q\dfrac{r}{s} \in \mathbb{Q} and tu∈Q\dfrac{t}{u} \in \mathbb{Q}, in lowest form, find r+s+t+ur + s + t + u.

Problem 42

For some real numbers m,nm, n and a positive integer aa, the list (a+1)n2,m2,a(n+1)2(a+1)n^2, m^2, a(n+1)^2 consists of three consecutive integers written in increasing order. What is the largest possible value of m2m^2?

Problem 43

Let P(x)=x2025,Q(x)=x4+x3+2x2+x+1P(x) = x^{2025}, Q(x) = x^4 + x^3 + 2x^2 + x + 1. Let R(x)R(x) be the polynomial remainder when the polynomial P(x)P(x) is divided by the polynomial Q(x)Q(x). Find R(3)R(3).

Problem 44

Complex numbers x,y,zx, y, z satisfy the following system of equations: x2+y2+z=xyx+y2+z2=yzx2+y+z2=xz\begin{aligned} x^2 + y^2 + z &= xy\\ x + y^2 + z^2 &= yz\\ x^2 + y + z^2 &= xz \end{aligned} Determine the sum of all distinct possible values of ∣(x2−y)(y2−z)(z2−x)∣|(x^2 - y)(y^2 - z)(z^2 - x)|.

Solutions

Solution: PRMO 2012, Q1

Key idea

Undo Rachel’s operations to recover the number, then do the intended ones.

Let the number be NN. Rachel computed (N−9)/3=43(N - 9)/3 = 43, so N−9=129N - 9 = 129 and N=138N = 138.

The intended calculation was 138−39=1359=15.\frac{138 - 3}{9} = \frac{135}{9} = 15.

Answer 15

Solution: PRMO 2013, Q3

Key idea

Every factorisation of 2020 has a negated twin, and the two contribute opposite values of aa.

If the roots are integers rr and ss then rs=20rs = 20 and a=−(r+s)a = -(r+s). The factor pairs of 2020 over the integers come in two families, positive and negative: (1,20),(2,10),(4,5)and(−1,−20),(−2,−10),(−4,−5),(1,20), (2,10), (4,5) \quad \text{and} \quad (-1,-20), (-2,-10), (-4,-5), giving a=−21,−12,−9a = -21, -12, -9 and a=21,12,9a = 21, 12, 9 respectively.

Every value is matched by its negative, because negating both roots leaves the product unchanged and reverses the sum. So the six values cancel in pairs and their total is 0.0.

Answer 0

Solution: PRMO 2015 Part A, Q1

Key idea

The question gives a total for one walk plus one ride and a total for two rides, so the single ride is the quantity to isolate first, and everything else follows by subtraction.

Let ww be the time for one walk of that distance and rr the time for one ride. The two facts we are given translate directly: w+r=334=154,2r=212=52.w + r = 3\tfrac34 = \tfrac{15}{4}, \qquad 2r = 2\tfrac12 = \tfrac52. The second gives r=54r = \tfrac54 at once, and then the first gives w=154−54=104=52.w = \tfrac{15}{4} - \tfrac54 = \tfrac{10}{4} = \tfrac52. Walking both ways therefore takes 2w=52w = 5 hours.

The reason the arithmetic is so short is that the second condition was handed to us already solved for a single ride. A question phrased as three separate journey times often hides one equation that needs no work at all, and it pays to look for it before setting up anything larger.

Answer 5

Solution: PRMO 2015 Part A, Q15

Key idea

Adding and subtracting the two equations separates the powers of 33 from the powers of 22 immediately.

Adding the two equations gives 2⋅3x=14582 \cdot 3^x = 1458, so 3x=729=36,x=6.3^x = 729 = 3^6, \qquad x = 6. Subtracting them gives 2⋅2y=5122 \cdot 2^y = 512, so 2y=256=28,y=8.2^y = 256 = 2^8, \qquad y = 8. Therefore xy=48xy = 48.

The only thing to be careful about is recognising 729729 and 256256 as exact powers rather than reaching for logarithms, which the competition never intends. Building up the small powers of 22 and 33 from memory is worth the five minutes it takes.

Answer 48

Solution: PRMO 2017, Q3

Key idea

Work in fractions of the job per day and add up what each stretch of time completes.

Team AA does 112\tfrac1{12} of the job each day and team BB does 136\tfrac1{36}.

In the first four days, AA alone completes 412=13\tfrac{4}{12} = \tfrac13. In the next two days both teams work, completing 2(112+136)=2⋅3+136=836=29.2\left(\frac1{12} + \frac1{36}\right) = 2 \cdot \frac{3 + 1}{36} = \frac{8}{36} = \frac29. So after six days the fraction done is 13+29=59\tfrac13 + \tfrac29 = \tfrac59, leaving 49\tfrac49.

Team BB alone needs 4/91/36=49×36=16\frac{4/9}{1/36} = \frac49 \times 36 = 16 more days.

Answer 16

Solution: IOQM 2025 Part SEP, Q7

Key idea

"pp percent of qq" and "qq percent of pp" are the same number, pq/100pq/100, so the answer is handed over without finding xx at all.

Both quantities in the question are 60x100\dfrac{60x}{100}: the first reads "6060 percent of xx" and the second "xx percent of 6060". Since we are told the first equals 4040, so does the second.

If you would rather see the number, 60%60\% of xx is 4040 gives x=2003x = \tfrac{200}{3}, and then x%x\% of 6060 is 2003⋅60100=40\tfrac{200}{3} \cdot \tfrac{60}{100} = 40, as promised.

Answer 40

Solution: PRMO 2013, Q11

Key idea

Adding the three equations completes three squares at once, and a sum of squares equal to zero fixes every variable.

Add the three equations: x2+y2+z2+2x+6y+4z=−17+1+2=−14.x^2 + y^2 + z^2 + 2x + 6y + 4z = -17 + 1 + 2 = -14. Move everything to one side and complete each square: (x+1)2+(y+3)2+(z+2)2=−14+1+9+4=0.(x+1)^2 + (y+3)^2 + (z+2)^2 = -14 + 1 + 9 + 4 = 0.

A sum of three real squares vanishes only when each does, so x=−1,y=−3,z=−2,x = -1, \qquad y = -3, \qquad z = -2, and these do satisfy all three original equations. Hence x2+y2+z2=1+9+4=14.x^2 + y^2 + z^2 = 1 + 9 + 4 = 14.

Answer 14

Solution: PRMO 2014, Q5

Key idea

Every variable is the common value minus a known constant, so the sum of all five is 5t−155t - 15 and the last equation pins tt down.

Call the common value tt. Reading the chain from the left, a=t−1,b=t−2,c=t−3,d=t−4,e=t−5,a = t - 1, \quad b = t - 2, \quad c = t - 3, \quad d = t - 4, \quad e = t - 5, so their sum is 5t−155t - 15. The final expression in the chain says t=(a+b+c+d+e)+3=5t−15+3=5t−12,t = (a+b+c+d+e) + 3 = 5t - 15 + 3 = 5t - 12, hence 4t=124t = 12 and t=3t = 3.

The five numbers are therefore 2,1,0,−1,−22, 1, 0, -1, -2, and a2+b2+c2+d2+e2=4+1+0+1+4=10.a^2 + b^2 + c^2 + d^2 + e^2 = 4 + 1 + 0 + 1 + 4 = 10.

Answer 10

Solution: PRMO 2014, Q6

Key idea

The roots multiply to 2014=2×19×532014 = 2 \times 19 \times 53 and add to nn, so the smallest nn comes from the most balanced factor pair, 38×5338 \times 53.

If the roots are integers rr and ss then rs=2014rs = 2014 and r+s=nr + s = n. Since 2014>02014 > 0 and nn is a natural number, both roots are positive.

Factorising, 2014=2×19×532014 = 2 \times 19 \times 53, so the factor pairs and their sums are (1,2014)→2015,(2,1007)→1009,(1, 2014) \to 2015, \qquad (2, 1007) \to 1009, (19,106)→125,(38,53)→91.(19, 106) \to 125, \qquad (38, 53) \to 91. The sum is smallest for the most balanced pair, which is the general rule for a fixed product, so n=91.n = 91.

Answer 91

Solution: PRMO 2014, Q7

Key idea

x=2x = \sqrt2 solves x(x4)=4x^{(x^4)} = 4, and then both requested powers are powers of 22.

Work with the size of xx rather than its sign, which settles both cases at once. Write t=∣x∣t = |x|. Whatever xx is, x4=t4x^4 = t^4, and if the power x(x4)x^{(x^4)} is a real number then its size is ∣x∣|x| raised to that same exponent. So the equation forces t t4=4.t^{\,t^4} = 4.

That has exactly one positive solution. For 0<t≤10 < t \le 1 the left side is at most 11, so t>1t > 1; and for t>1t > 1 both tt and t4t^4 increase with tt, so t t4t^{\,t^4} is strictly increasing and meets the value 44 once. Since (2)4=4\left(\sqrt2\right)^4 = 4, that once is at t=2.t = \sqrt2.

So x=2x = \sqrt2 or x=−2x = -\sqrt2, and both really do work. The first is immediate. For the second, the exponent x4x^4 is 44, a whole number, so the power is defined without any question about what a negative base raised to a fractional power should mean, and (−2)4=4\left(-\sqrt2\right)^4 = 4 as well.

Both signs give the same answer, because the two exponents in question are also even. With x=±21/2x = \pm 2^{1/2}, x(x2)=212×2=2,x(x8)=212×16=28=256.x^{(x^2)} = 2^{\frac12 \times 2} = 2, \qquad x^{(x^8)} = 2^{\frac12 \times 16} = 2^8 = 256. Their sum is 2+256=258.2 + 256 = 258.

The exponents x2x^2, x4x^4, x8x^8 are successively squared, so the three values are 212^1, 222^2 and 282^8: the answer grows alarmingly fast, which is the joke in the question.

Answer 258

Solution: PRMO 2014, Q8

Key idea

Each added element changes the mean by a stated amount, giving two linear equations in the size of SS and its mean.

Let SS have nn elements and total nMnM.

First note that neither 1515 nor 11 already lies in SS, which is what makes the two unions bigger. If 1515 were in SS then S∪{15}S \cup \{15\} would be SS itself, with mean MM rather than M+2M+2. And if 11 were in SS then S∪{15,1}S \cup \{15, 1\} would be S∪{15}S \cup \{15\}, with mean M+2M+2 rather than M+1M+1. So the two unions have n+1n+1 and n+2n+2 elements.

Adding 1515 gives a set of n+1n+1 elements with mean M+2M + 2: nM+15=(n+1)(M+2)=nM+M+2n+2,nM + 15 = (n+1)(M+2) = nM + M + 2n + 2, so M+2n=13M + 2n = 13.

Adding 1515 and 11 gives n+2n + 2 elements with mean M+1M + 1: nM+16=(n+2)(M+1)=nM+2M+n+2,nM + 16 = (n+2)(M+1) = nM + 2M + n + 2, so 2M+n=142M + n = 14.

Solving the pair, M=13−2nM = 13 - 2n and 2(13−2n)+n=142(13 - 2n) + n = 14, so 3n=123n = 12 and n=4,M=5.n = 4, \qquad M = 5. A check: a four-element set of mean 55 has total 2020; adding 1515 gives 3535 over five elements, mean 7=M+27 = M+2; adding 11 as well gives 3636 over six, mean 6=M+16 = M+1.

Answer 4

Solution: PRMO 2015 Part A, Q3

Key idea

Subtracting one equation from the other kills the x2x^2 term, and what is left is linear, so the common root is pinned down before kk is.

Suppose rr satisfies both x2−4x+k=0x^2 - 4x + k = 0 and x2+kx−4=0x^2 + kx - 4 = 0. Subtracting the second from the first removes the quadratic term: (−4−k)r+(k+4)=0,that is(k+4)(1−r)=0.(-4 - k)r + (k + 4) = 0, \qquad \text{that is} \qquad (k+4)(1 - r) = 0. So either k=−4k = -4 or r=1r = 1, and the two cases are worth separating carefully.

If k=−4k = -4 the two equations both read x2−4x−4=0x^2 - 4x - 4 = 0. They are then the same equation and share both roots, which the problem has ruled out by insisting on exactly one common root.

That leaves r=1r = 1. Putting x=1x = 1 into the first equation gives 1−4+k=01 - 4 + k = 0, so k=3k = 3. It is worth confirming that this really does produce exactly one shared root and not two: with k=3k = 3 the equations factorise as (x−1)(x−3)=0(x-1)(x-3) = 0 and (x−1)(x+4)=0(x-1)(x+4) = 0, whose roots are {1,3}\{1,3\} and {1,−4}\{1,-4\}. They meet in 11 alone, as required.

Answer 3

Solution: PRMO 2017, Q4

Key idea

The two quadratics are independent, so minimise each separately: aa is a sum of two factors of 2020 and bb is a product of two numbers adding to 1717.

If all roots are negative integers, write the roots of the first quadratic as −p-p and −q-q with p,qp, q positive integers, and those of the second as −r-r and −s-s. Comparing coefficients, pq=20,a=p+q,r+s=17,b=rs.pq = 20, \quad a = p + q, \qquad r + s = 17, \quad b = rs. The two quadratics share no constraint, so aa and bb can be minimised separately.

For aa: the factor pairs of 2020 are (1,20)(1,20), (2,10)(2,10) and (4,5)(4,5), giving a=21,12,9a = 21, 12, 9. The smallest is a=9a = 9, which is the general rule that a product splits most evenly nearest its square root.

For bb, use positivity of the integers r,sr,s and their sum 1717: rs−16=rs−(r+s)+1=(r−1)(s−1)≥0.rs-16=rs-(r+s)+1=(r-1)(s-1)\ge0. Equality is attained at (r,s)=(1,16)(r,s)=(1,16), so the least bb is 1616.

Hence the smallest value of a+ba + b is 9+16=25.9 + 16 = 25.

Note that the two halves pull in opposite directions: a fixed product wants a balanced split, a fixed sum wants a lopsided one. Both are the same fact about (r−s)2≥0(r-s)^2 \ge 0 seen from different sides.

Answer 25

Solution: PRMO 2017, Q19

Key idea

The quartic has no cubic term, so its four roots sum to zero and the fourth root must be −6-6; the product of the roots is then the constant term.

Write the equation as x4+0⋅x3+ax2+bx−c=0x^4 + 0 \cdot x^3 + ax^2 + bx - c = 0, a monic quartic with four roots, three of which are 11, 22 and 33. Let the fourth be rr: the factor theorem gives the three factors x−1,x−2,x−3x-1,x-2,x-3, and the remaining monic factor is linear, x−rx-r. In (x−1)(x−2)(x−3)(x−r),(x-1)(x-2)(x-3)(x-r), the x3x^3 term takes a constant from one factor and xx from the other three, so its coefficient is −(1+2+3+r)-(1+2+3+r). The constant term takes a constant from every factor, giving 1⋅2⋅3⋅r1\cdot2\cdot3\cdot r.

The coefficient of x3x^3 is zero, and for a monic quartic that coefficient is minus the sum of the roots. So 1+2+3+r=0,r=−6.1 + 2 + 3 + r = 0, \qquad r = -6.

The constant term of a monic quartic is the product of its roots. Here the constant term is −c-c, so −c=1⋅2⋅3⋅(−6)=−36,c=36.-c = 1 \cdot 2 \cdot 3 \cdot (-6) = -36, \qquad c = 36.

Neither aa nor bb was needed, though they follow at once: the polynomial is (x−1)(x−2)(x−3)(x+6)(x-1)(x-2)(x-3)(x+6).

Answer 36

Solution: PRMO 2018, Q1

Key idea

The three first-page numbers are 11, one more than the first volume, and one more than the first two together, so their sum is a single linear equation in the length of the first volume.

Let the first volume have xx pages. Then the second has x+50x + 50 and the third has 32(x+50)\tfrac32(x+50).

The first page of volume one is numbered 11. The first page of volume two follows the whole of volume one, so it is x+1x + 1. The first page of volume three follows volumes one and two, so it is x+(x+50)+1=2x+51x + (x+50) + 1 = 2x + 51. Adding, 1+(x+1)+(2x+51)=3x+53=1709,1 + (x+1) + (2x+51) = 3x + 53 = 1709, so 3x=16563x = 1656 and x=552x = 552. The three volumes then have 552552, 602602 and 903903 pages, and the last page number is n=552+602+903=2057.n = 552 + 602 + 903 = 2057.

Factorising, 2057=11×187=11×11×172057 = 11 \times 187 = 11 \times 11 \times 17, so the largest prime factor is 1717.

The condition that the third volume is one and a half times the second is what forces x+50x + 50 to be even, and 602602 duly is. A problem that hands you a fractional multiplier is usually checking that you notice the parity it demands.

Answer 17

Solution: PRMO 2019, Q2

Key idea

Expanding both sides shows every term matches except the constants, so the functional equation says nothing more than b=2b = 2, and the integer roots then leave two possibilities for aa with the same square.

Write f(x)=x2+ax+bf(x) = x^2 + ax + b and expand the two sides for nonzero xx. On the left, f ⁣(x+1x)=(x+1x)2+a(x+1x)+b,f\!\left(x + \tfrac1x\right) = \left(x + \tfrac1x\right)^2 + a\left(x + \tfrac1x\right) + b, which expands to x2+1x2+2+a(x+1x)+bx^2 + \tfrac{1}{x^2} + 2 + a\left(x + \tfrac1x\right) + b. On the right, f(x)+f ⁣(1x)=x2+ax+b+1x2+ax+b=x2+1x2+a(x+1x)+2b.f(x) + f\!\left(\tfrac1x\right) = x^2 + ax + b + \tfrac{1}{x^2} + \tfrac{a}{x} + b = x^2 + \tfrac{1}{x^2} + a\left(x + \tfrac1x\right) + 2b. Everything cancels except the constants, leaving 2+b=2b2 + b = 2b, that is b=2b = 2. Notice that aa is untouched: the functional equation constrains only the constant term.

Now use the second condition. The roots of x2+ax+2=0x^2 + ax + 2 = 0 are integers with product 22, so they are 11 and 22, or −1-1 and −2-2. Their sum is −a-a, giving a=−3a = -3 or a=3a = 3. Either way a2=9a^2 = 9, and a2+b2=9+4=13.a^2 + b^2 = 9 + 4 = 13.

Answer 13

Solution: IOQM 2023, Q2

Key idea

Setting t=log⁡abt = \log_a b turns the condition into t+6/t=5t + 6/t = 5, a quadratic whose two roots say that bb is either a2a^2 or a3a^3.

The two logarithms in the equation are reciprocals of one another, since log⁡ba=1/log⁡ab\log_b a = 1/\log_a b. Writing t=log⁡abt = \log_a b, which is legitimate because a,b≥2a, b \ge 2, the condition becomes t+6t=5,that ist2−5t+6=0,t + \frac{6}{t} = 5, \qquad \text{that is} \qquad t^2 - 5t + 6 = 0, so t=2t = 2 or t=3t = 3. In other words b=a2b = a^2 or b=a3b = a^3, and nothing else can occur.

Counting is now a matter of respecting the bound b≤2023b \le 2023.

For b=a2b = a^2 we need a2≤2023a^2 \le 2023, and since 442=193644^2 = 1936 while 452=202545^2 = 2025, the values a=2,3,…,44a = 2, 3, \ldots, 44 all work. That is 4343 pairs.

For b=a3b = a^3 we need a3≤2023a^3 \le 2023, and since 123=172812^3 = 1728 while 133=219713^3 = 2197, the values a=2,3,…,12a = 2, 3, \ldots, 12 work, giving 1111 pairs.

No pair is counted twice, because a2=a3a^2 = a^3 only for a=1a = 1, which is excluded. The total is 43+11=5443 + 11 = 54.

Answer 54

Solution: IOQM 2024, Q5

Key idea

The three cyclic ratios multiply to 11. Naming them p,q,rp,q,r makes aa a sum, bb a sum of pairwise products and cc a product of pairwise sums, which one expansion connects.

Set p=xy,q=yz,r=zx,so that pqr=1.p = \frac xy, \qquad q = \frac yz, \qquad r = \frac zx, \qquad \text{so that } pqr = 1. Then a=p+q+ra = p + q + r straight away. For bb, note that xz=1r\dfrac xz = \dfrac1r and yx=1p\dfrac yx = \dfrac1p and zy=1q\dfrac zy = \dfrac1q, so b=1p+1q+1r=qr+rp+pqpqr=pq+qr+rp,b = \frac1p + \frac1q + \frac1r = \frac{qr + rp + pq}{pqr} = pq + qr + rp, using pqr=1pqr = 1. So aa and bb are the first two elementary symmetric functions of p,q,rp, q, r, and the third is pqr=1pqr = 1.

Now c=(p+q)(q+r)(r+p)c = (p+q)(q+r)(r+p), and there is a standard identity for that product: (p+q)(q+r)(r+p)=(p+q+r)(pq+qr+rp)−pqr.(p+q)(q+r)(r+p) = (p+q+r)(pq+qr+rp) - pqr. It is worth seeing why rather than taking it on trust. Write out the two expansions: (p+q)(q+r)(r+p)=p2q+p2r+pq2+pr2+q2r+qr2+2pqr,(p+q+r)(pq+qr+rp)=p2q+p2r+pq2+pr2+q2r+qr2+3pqr.\begin{aligned} (p+q)(q+r)(r+p)&=p^2q+p^2r+pq^2+pr^2+q^2r+qr^2+2pqr,\\ (p+q+r)(pq+qr+rp)&=p^2q+p^2r+pq^2+pr^2+q^2r+qr^2+3pqr. \end{aligned} The six mixed-square terms match; the second line has one extra pqrpqr.

Hence c=ab−1c = ab - 1 and ∣ab−c∣=∣ab−(ab−1)∣=1.|ab - c| = |ab - (ab-1)| = 1.

Answer 1

Solution: IOQM 2024, Q16

Key idea

Replacing xx by 3−x3-x gives a second equation in the same two unknowns f(x)f(x) and f(3−x)f(3-x), and eliminating the second one produces ff outright.

The relation ties f(x)f(x) to f(3−x)f(3-x), so apply it at 3−x3-x as well, noting that 3−(3−x)=x3 - (3-x) = x: 4f(3−x)+3f(x)=x2,4f(x)+3f(3−x)=(3−x)2.4f(3-x) + 3f(x) = x^2, \qquad 4f(x) + 3f(3-x) = (3-x)^2. Treat f(x)f(x) and f(3−x)f(3-x) as two unknowns. Multiplying the first by 33, the second by 44, and subtracting removes f(3−x)f(3-x): 7f(x)=4(3−x)2−3x2=36−24x+x2,7f(x) = 4(3-x)^2 - 3x^2 = 36 - 24x + x^2, so f(x)=x2−24x+367.f(x) = \frac{x^2 - 24x + 36}{7}.

Now the required difference is exact, with no rounding needed: f(27)−f(25)=(729−648+36)−(625−600+36)7=567=8.f(27) - f(25) = \frac{(729 - 648 + 36) - (625 - 600 + 36)}{7} = \frac{56}{7} = 8.

Answer 8

Solution: IOQM 2025 Part SEP, Q28

Key idea

Applying ff twice changes the parity as it goes, so there are only two chains to follow and one of them ends in a non-square.

First exclude n=1n=1: it gives f(1)=0f(1)=0, outside the domain of ff, so f(f(1))f(f(1)) is undefined. For n≥2n\ge2, take the two parity cases.

If nn is odd then f(n)=n−1f(n) = n-1, which is even, so f(f(n))=(n−1)2−1=99,(n−1)2=100,n=11,f(f(n)) = (n-1)^2 - 1 = 99, \qquad (n-1)^2 = 100, \qquad n = 11, taking the positive root since nn is a positive integer. And 1111 is indeed odd, so this is a genuine solution.

If nn is even then f(n)=n2−1f(n) = n^2-1, which is odd, so f(f(n))=(n2−1)−1=n2−2=99,n2=101,f(f(n)) = (n^2-1) - 1 = n^2 - 2 = 99, \qquad n^2 = 101, which is not a perfect square. No solution here.

So the only value is n=11n = 11, and the sum of all possible values is 1111.

Answer 11

Solution: PRMO 2012, Q12

Key idea

a+a2−1=a+1+a−12\sqrt{a + \sqrt{a^2-1}} = \dfrac{\sqrt{a+1} + \sqrt{a-1}}{\sqrt2}, which turns the reciprocal into a difference of two square roots.

To remove the nested root, try a sum u+v\sqrt u+\sqrt v. Its square is u+v+2uvu+v+2\sqrt{uv}, so we want u+v=2011,4uv=20112−1.u+v=2011,\qquad4uv=2011^2-1. Then (u−v)2=(u+v)2−4uv=1(u-v)^2=(u+v)^2-4uv=1, giving u=1006u=1006 and v=1005v=1005. These numbers suggest the following identity, which also checks the calculation: (a+1+a−12)2=2a+2a2−12=a+a2−1.\left(\frac{\sqrt{a+1} + \sqrt{a-1}}{\sqrt2}\right)^2 = \frac{2a + 2\sqrt{a^2-1}}{2} = a + \sqrt{a^2 - 1}.

With a=2011a = 2011 that gives 12011+20112−1=22012+2010.\frac{1}{\sqrt{2011 + \sqrt{2011^2-1}}} = \frac{\sqrt2}{\sqrt{2012} + \sqrt{2010}}. Rationalising the denominator, whose conjugate difference is 2012−2010=22012 - 2010 = 2, 2(2012−2010)2=4024−40202=1006−1005.\frac{\sqrt2\left(\sqrt{2012} - \sqrt{2010}\right)}{2} = \frac{\sqrt{4024} - \sqrt{4020}}{2} = \sqrt{1006} - \sqrt{1005}.

So m=1006m = 1006 and n=1005n = 1005 is one answer, and m+n=2011m + n = 2011.

Why those two integers are forced

Nothing so far rules out a second pair, and the question asks for m+nm+n as though there were only one, so it is worth checking. Suppose the same number equals m−n\sqrt m - \sqrt n for positive integers mm and nn, and put k=m−nk = m - n, which is positive since m>n\sqrt m > \sqrt n. Because (1006−1005)(1006+1005)=1,\left(\sqrt{1006} - \sqrt{1005}\right)\left(\sqrt{1006} + \sqrt{1005}\right) = 1, the number is the reciprocal of 1006+1005\sqrt{1006} + \sqrt{1005}, and likewise (m−n)(m+n)=k\left(\sqrt m - \sqrt n\right)\left(\sqrt m + \sqrt n\right) = k gives m+n=k(1006+1005)\sqrt m + \sqrt n = k\left(\sqrt{1006} + \sqrt{1005}\right). Adding this to m−n=1006−1005\sqrt m - \sqrt n = \sqrt{1006} - \sqrt{1005}, 2m=(k+1)1006+(k−1)1005.2\sqrt m = (k+1)\sqrt{1006} + (k-1)\sqrt{1005}. Squaring, the cross term is 2(k+1)(k−1)1006⋅10052(k+1)(k-1)\sqrt{1006 \cdot 1005}, and 1006⋅1005=10110301006 \cdot 1005 = 1011030 lies strictly between 100521005^2 and 100621006^2, so it is not a perfect square and its root is irrational. Everything else in the equation is a whole number, so (k+1)(k−1)=0(k+1)(k-1) = 0, and with kk positive that means k=1k = 1. Then 2m=210062\sqrt m = 2\sqrt{1006}, so m=1006m = 1006 and n=1005n = 1005 after all, and the answer stands.

The answer being the year is not a coincidence: for any aa the same computation gives a+12−a−12\sqrt{\tfrac{a+1}{2}} - \sqrt{\tfrac{a-1}{2}}, whose two arguments add to aa.

Answer 2011

Solution: PRMO 2012, Q13

Key idea

Adding the three equations gives 2a+c=02a + c = 0, and everything else is a multiple of aa.

Add the three relations: a+b+c=(b−c)+(c−d)+(d−a)=b−a,a + b + c = (b - c) + (c - d) + (d - a) = b - a, so 2a+c=02a + c = 0, that is c=−2ac = -2a. Then c=d−ac = d - a gives d=c+a=−ad = c + a = -a, and b=c−db = c - d gives b=−2a+a=−ab = -2a + a = -a. The first relation is then satisfied automatically: b−c=−a+2a=ab - c = -a + 2a = a.

Since abcd≠0abcd \ne 0 we have a≠0a \ne 0, and ab+bc+cd+da=a−a+−a−2a+−2a−a+−aa=−1+12+2−1=12.\frac ab + \frac bc + \frac cd + \frac da = \frac{a}{-a} + \frac{-a}{-2a} + \frac{-2a}{-a} + \frac{-a}{a} = -1 + \frac12 + 2 - 1 = \frac12.

Answer 1/21/2

Solution: PRMO 2012, Q17

Key idea

The product is ∏(xi2+1)\prod(x_i^2+1) over ∏xi\prod x_i, and multiplying the cubic by its reflection in x↦−xx \mapsto -x produces a cubic whose roots are the xi2x_i^2.

Write P(x)=x3+3x+5=(x−x1)(x−x2)(x−x3)P(x) = x^3 + 3x + 5 = (x - x_1)(x - x_2)(x - x_3). Since xi+1/xi=(xi2+1)/xix_i + 1/x_i = \left(x_i^2+1\right)/x_i, ∏i=13(xi+1xi)=∏i(xi2+1)∏ixi.\prod_{i=1}^{3}\left(x_i + \frac{1}{x_i}\right) = \frac{\prod_i \left(x_i^2 + 1\right)}{\prod_i x_i}.

The denominator is Vieta on x3+0x2+3x+5x^3 + 0x^2 + 3x + 5, which gives x1x2x3=−5x_1x_2x_3 = -5.

For the numerator, reflect. Because PP has odd degree, P(−x)=∏i(−x−xi)=−∏i(x+xi),P(-x) = \prod_i \left(-x - x_i\right) = -\prod_i \left(x + x_i\right), so ∏i(x+xi)=−P(−x)\prod_i \left(x + x_i\right) = -P(-x), and multiplying the two products pairs each root with its negative: ∏i(x2−xi2)=P(x)[−P(−x)]=(x3+3x+5)(x3+3x−5),\prod_i \left(x^2 - x_i^2\right) = P(x)\left[-P(-x)\right] = \left(x^3 + 3x + 5\right)\left(x^3 + 3x - 5\right), a difference of two squares. So ∏i(x2−xi2)=(x3+3x)2−25=x2(x2+3)2−25.\prod_i \left(x^2 - x_i^2\right) = \left(x^3 + 3x\right)^2 - 25 = x^2\left(x^2+3\right)^2 - 25. Both sides are polynomials in x2x^2 that agree for every real xx, so writing y=x2y = x^2 they agree for every y≥0y \ge 0, and two polynomials that agree at infinitely many points are the same polynomial: ∏i(y−xi2)=y(y+3)2−25.\prod_i \left(y - x_i^2\right) = y(y+3)^2 - 25.

Now put y=−1y = -1, which is legitimate because that is an identity between polynomials rather than a statement about a square. The left side becomes ∏i(−1−xi2)=−∏i(xi2+1)\prod_i\left(-1 - x_i^2\right) = -\prod_i\left(x_i^2+1\right), and the right side is (−1)(−1+3)2−25=−29(-1)(-1+3)^2 - 25 = -29. So the numerator is 2929, and ∏i=13(xi+1xi)=29−5=−295.\prod_{i=1}^{3}\left(x_i + \frac{1}{x_i}\right) = \frac{29}{-5} = -\frac{29}{5}.

Answer −29/5-29/5

Solution: PRMO 2014, Q9

Key idea

The new product is l+2+1ll + 2 + \tfrac1l, and for that to be a natural number ll must divide 11.

Let aa and bb be the roots of the first quadratic, so a+b=ka + b = k and ab=lab = l, both natural numbers. In particular ab=l≥1ab = l \ge 1, so neither root is zero and the reciprocals below are defined. The new quadratic has roots a+1ba + \tfrac1b and b+1ab + \tfrac1a, so its coefficients are

p=a+b+1a+1b=k+a+bab=k+kl,p = a + b + \frac1a + \frac1b = k + \frac{a+b}{ab} = k + \frac{k}{l}, q=(a+1b)(b+1a)=ab+1+1+1ab=l+2+1l.q = \left(a + \frac1b\right)\left(b + \frac1a\right) = ab + 1 + 1 + \frac{1}{ab} = l + 2 + \frac1l.

Now qq is required to be a natural number, and ll is one, so 1l\tfrac1l must be an integer: l=1l = 1. Then q=1+2+1=4,q = 1 + 2 + 1 = 4, and p=2kp = 2k, which is a natural number for every natural kk, so such k,l,p,qk, l, p, q really do exist.

The value of qq is therefore forced, and the sum of all possible values is 44.

Answer 4

Solution: PRMO 2015 Part A, Q12

Key idea

Eliminating one variable leaves 13a=12−5b13a = 12 - 5b and 13c=5+12b13c = 5 + 12b, and (12,−5)(12,-5) and (5,12)(5,12) are two perpendicular vectors of the same length 1313, so squaring and adding collapses everything.

image

Two equations in three unknowns cannot determine aa, bb and cc separately, so the quantity asked for must be constant along the whole line of solutions. Let us find that line by solving for aa and cc in terms of bb.

Multiply the second equation 8a+4b−c=78a + 4b - c = 7 by 88 and add the first, a−7b+8c=4a - 7b + 8c = 4: 65a+25b=60,so13a+5b=12.65a + 25b = 60, \qquad \text{so} \qquad 13a + 5b = 12. Substituting a=12−5b13a = \tfrac{12 - 5b}{13} back into the first equation and clearing denominators gives 8c=4−a+7b=40+96b138c = 4 - a + 7b = \tfrac{40 + 96b}{13}, hence 13a=12−5b,13c=5+12b.13a = 12 - 5b, \qquad 13c = 5 + 12b.

Now square and add. The cross terms are −120b-120b and +120b+120b, which cancel: 169(a2+c2)=(12−5b)2+(5+12b)2=169+169b2=169(1+b2).169(a^2 + c^2) = (12 - 5b)^2 + (5 + 12b)^2 = 169 + 169b^2 = 169(1 + b^2). Dividing by 169169 gives a2+c2=1+b2a^2 + c^2 = 1 + b^2, and therefore a2−b2+c2=1.a^2 - b^2 + c^2 = 1.

The cancellation was not luck. The coefficient pairs (12,−5)(12,-5) and (5,12)(5,12) are perpendicular and both have length 1313, which is exactly the condition under which the sum of the two squares loses its cross term and keeps a clean multiple of 1+b21 + b^2. Ugly-looking coefficients in a problem like this are usually built to do something of that kind.

Answer 1

Solution: PRMO 2015 Part A, Q18

Key idea

With a+b+c=0a+b+c = 0 the denominator 2a2+bc2a^2 + bc factorises as (a−b)(a−c)(a-b)(a-c). The sum that remains is then the leading coefficient of the one quadratic taking the values a2a^2, b2b^2, c2c^2 at aa, bb, cc, and that quadratic is x2x^2.

Start with the denominators, which look unpromising until the condition is used. Since c=−a−bc = -a-b, 2a2+bc=2a2+b(−a−b)=2a2−ab−b2=(2a+b)(a−b),2a^2 + bc = 2a^2 + b(-a-b) = 2a^2 - ab - b^2 = (2a+b)(a-b), and 2a+b=a+(a+b)=a−c2a + b = a + (a+b) = a - c. So 2a2+bc=(a−b)(a−c),2a^2 + bc = (a-b)(a-c), and by the same computation with the letters rotated, 2b2+ca=(b−c)(b−a)2b^2 + ca = (b-c)(b-a) and 2c2+ab=(c−a)(c−b)2c^2 + ab = (c-a)(c-b). That PP is defined tells us these are all non-zero, so aa, bb, cc are three distinct numbers. What we must evaluate is P=a2(a−b)(a−c)+b2(b−a)(b−c)+c2(c−a)(c−b).P = \frac{a^2}{(a-b)(a-c)} + \frac{b^2}{(b-a)(b-c)} + \frac{c^2}{(c-a)(c-b)}.

At this point the condition a+b+c=0a + b + c = 0 has done its job and is no longer needed, because the sum above equals 11 for any three distinct numbers. Here is the reason, and the one fact behind it is that a quadratic is completely determined by its values at three different points: two quadratics that agree at three points have a difference of degree at most 22 with three roots, so that difference is the zero polynomial.

To make one term select the value at aa, it should vanish at bb and cc, so give it the factors (x−b)(x−c)(x-b)(x-c); divide by (a−b)(a−c)(a-b)(a-c) to make its value at aa equal 11. For a concrete example at the points 0,1,20,1,2, the polynomial −x(x−2)-x(x-2) is 11 at 11 and 00 at the other two. Apply this same recipe at a,b,ca,b,c and consider L(x)=a2(x−b)(x−c)(a−b)(a−c)+b2(x−a)(x−c)(b−a)(b−c)+c2(x−a)(x−b)(c−a)(c−b),L(x) = a^2 \frac{(x-b)(x-c)}{(a-b)(a-c)} + b^2 \frac{(x-a)(x-c)}{(b-a)(b-c)} + c^2 \frac{(x-a)(x-b)}{(c-a)(c-b)}, which has degree at most 22. Each of the three fractions is designed to equal 11 at one of the three points and 00 at the other two, so substituting x=ax = a makes the last two terms vanish and the first collapse to a2a^2, and similarly at x=bx = b and x=cx = c. So LL agrees with x2x^2 at aa, bb and cc, and by the fact just stated L(x)=x2L(x) = x^2 identically.

Comparing the coefficients of x2x^2 on both sides gives exactly P=1P = 1, since the coefficient of x2x^2 in LL is the sum PP and the coefficient of x2x^2 in x2x^2 is 11.

That way of assembling a polynomial from its values is called Lagrange interpolation. Here it explains why the sum is the coefficient of x2x^2, rather than merely confirming its value.

Answer 1

Solution: PRMO 2017, Q2

Key idea

In terms of x=ax = \sqrt a and y=by = \sqrt b the two given quantities are x3+y3x^3 + y^3 and xy(x+y)xy(x+y), and those are exactly the pieces of (x+y)3(x+y)^3.

Put x=ax = \sqrt a and y=by = \sqrt b, both positive. The two conditions become x3+y3=183,x2y+xy2=xy(x+y)=182.x^3 + y^3 = 183, \qquad x^2 y + x y^2 = xy(x+y) = 182.

Now recall the expansion (x+y)3=x3+y3+3xy(x+y)(x+y)^3 = x^3 + y^3 + 3xy(x+y), which is built from precisely those two quantities: (x+y)3=183+3×182=183+546=729,sox+y=9.(x+y)^3 = 183 + 3 \times 182 = 183 + 546 = 729, \qquad \text{so} \qquad x + y = 9.

From the second condition, 9xy=1829xy = 182, so xy=1829xy = \tfrac{182}{9}. Hence a+b=x2+y2=(x+y)2−2xy=81−3649=729−3649=3659,a + b = x^2 + y^2 = (x+y)^2 - 2xy = 81 - \frac{364}{9} = \frac{729 - 364}{9} = \frac{365}{9}, and 95(a+b)=95⋅3659=3655=73.\frac95 (a+b) = \frac95 \cdot \frac{365}{9} = \frac{365}{5} = 73.

The factor 95\tfrac95 in the question is a courtesy: it turns an ugly fraction into a whole number, and its presence is a hint that a+ba+b will have a 99 underneath.

Answer 73

Solution: PRMO 2018, Q9

Key idea

Substituting x=a+bx = a+b and eliminating aa leaves a=−(s+2)−2/(s−1)a = -(s+2) - 2/(s-1), so s−1s - 1 must divide 22 and there are only four candidates.

Write s=a+bs = a + b, so b=s−ab = s - a, and substitute x=sx = s into the equation: s2+as+b=0,that iss2+as+s−a=0.s^2 + as + b = 0, \qquad \text{that is} \qquad s^2 + as + s - a = 0. Collecting the terms in aa gives a(s−1)=−(s2+s)a(s - 1) = -(s^2 + s), and s=1s = 1 is impossible because it would force s2+s=2s^2 + s = 2 to vanish. So a=−s2+ss−1=−(s+2)−2s−1,a = -\frac{s^2 + s}{s - 1} = -(s+2) - \frac{2}{s-1}, using s2+s=(s−1)(s+2)+2s^2 + s = (s-1)(s+2) + 2.

For aa to be an integer, s−1s - 1 must divide 22, so s∈{−1,0,2,3}s \in \{-1, 0, 2, 3\}. Working out aa and then b=s−ab = s - a in each case: sabb2−10−1100002−68643−6981\begin{array}{c|rrr} s & a & b & b^2 \\ \hline -1 & 0 & -1 & 1 \\ 0 & 0 & 0 & 0 \\ 2 & -6 & 8 & 64 \\ 3 & -6 & 9 & 81 \end{array} Each really works: for instance s=3s = 3 gives x2−6x+9=(x−3)2x^2 - 6x + 9 = (x-3)^2, whose root 33 is indeed a+ba + b. So the maximum of b2b^2 is 8181.

Answer 81

Solution: PRMO 2019, Q8

Key idea

Dividing by x2+x+1x^2+x+1 leaves x3x^3 with remainder 11, so x2nx^{2^n} leaves the same remainder as xrx^r, where rr is what 2n2^n leaves modulo 33, and only r=2r=2 works.

Everything turns on one factorisation: x3−1=(x−1)(x2+x+1)x^3 - 1 = (x-1)\left(x^2+x+1\right), so x2+x+1x^2+x+1 divides x3−1x^3 - 1. It then divides x3m−1x^{3m}-1 for every m≥1m \ge 1 as well, since x3m−1=(x3−1)(x3m−3+x3m−6+⋯+1).x^{3m} - 1 = \left(x^3 - 1\right)\left(x^{3m-3} + x^{3m-6} + \cdots + 1\right).

Write 2n=3m+r2^n = 3m + r with remainder rr, so 0≤r≤20 \le r \le 2. Then x2n−xr=xr(x3m−1)x^{2^n} - x^{r} = x^{r}\left(x^{3m} - 1\right) is divisible by x2+x+1x^2+x+1, which says that x2n+x+1x^{2^n} + x + 1 and xr+x+1x^{r} + x + 1 leave the same remainder on division by x2+x+1x^2+x+1. So one is divisible by it exactly when the other is, and since 2n2^n is never a multiple of 33 there are only two cases to try: r=1:x+x+1=2x+1,r=2:x2+x+1.r = 1: \quad x + x + 1 = 2x+1, \qquad\qquad r = 2: \quad x^2 + x + 1. A polynomial of degree 11 is not divisible by one of degree 22, so r=1r=1 fails; r=2r=2 gives the divisor itself, so it succeeds. The condition is exactly 2n≡2(mod3).2^n \equiv 2 \pmod 3. Now 2≡−1(mod3)2 \equiv -1 \pmod 3, so 2n≡(−1)n2^n \equiv (-1)^n, which is 2≡−12 \equiv -1 exactly when nn is odd.

The odd nn with 3≤n≤1003 \le n \le 100 are 3,5,7,…,993, 5, 7, \ldots, 99, and there are 99−32+1=49\tfrac{99-3}{2} + 1 = 49 of them.

Answer 49

Solution: IOQM 2023, Q12

Key idea

The identity behind p13+p23+p33=3p1p2p3p_1^3+p_2^3+p_3^3 = 3p_1p_2p_3 allows only p1+p2+p3=0p_1+p_2+p_3 = 0 or p1=p2=p3p_1=p_2=p_3, and the parity of cc rules the first out.

The relevant identity is x3+y3+z3−3xyz=(x+y+z)⋅12[(x−y)2+(y−z)2+(z−x)2],x^3+y^3+z^3-3xyz = (x+y+z)\cdot\tfrac12\bigl[(x-y)^2+(y-z)^2+(z-x)^2\bigr], so the given condition holds exactly when p1+p2+p3=0p_1+p_2+p_3 = 0 or p1=p2=p3p_1 = p_2 = p_3.

Look at the first possibility. Adding the three values, p1+p2+p3=(1+8+27)+(1+4+9)a+(1+2+3)b+3c,p_1+p_2+p_3 = (1+8+27) + (1+4+9)a + (1+2+3)b + 3c, that is, 36+14a+6b+3c36 + 14a + 6b + 3c. Every term except 3c3c is even, and cc is odd, so 3c3c is odd and the whole sum is odd. An odd number is not 00. So this possibility cannot occur, and the parity condition on cc, which looks decorative, is exactly what removes it.

That leaves p1=p2=p3p_1 = p_2 = p_3, say all equal to kk. Then P(x)−kP(x) - k is a monic cubic vanishing at 11, 22 and 33, so P(x)=(x−1)(x−2)(x−3)+k.P(x) = (x-1)(x-2)(x-3) + k. Its constant term is c=P(0)=−6+kc = P(0) = -6 + k, which is odd precisely when kk is odd, so such polynomials do exist. Reading off the values we need, p0=k−6,p1=p2=k,p_0 = k - 6, \qquad p_1 = p_2 = k, and therefore p2+2p1−3p0=k+2k−3(k−6)=18.p_2 + 2p_1 - 3p_0 = k + 2k - 3(k-6) = 18.

The answer does not depend on kk, which is the sign that the question was built around the factorisation rather than around any particular polynomial.

Answer 18

Solution: IOQM 2024, Q10

Key idea

The whole expression is a sum of two squares, (a−3b)2+(pb−3c)2(a - 3b)^2 + (pb - 3c)^2, so the equation pins the sides down to a=3ba = 3b and c=pb/3c = pb/3, and only the triangle inequality is left.

image

Group the terms so that the squares appear: a2−6ab+9b2=(a−3b)2,p2b2−6pbc+9c2=(pb−3c)2,a^2 - 6ab + 9b^2 = (a - 3b)^2, \qquad p^2b^2 - 6pbc + 9c^2 = (pb - 3c)^2, and adding these accounts for every term of the given expression, since 9b2+p2b2=(p2+9)b29b^2 + p^2b^2 = (p^2+9)b^2. So the equation reads (a−3b)2+(pb−3c)2=0.(a-3b)^2 + (pb - 3c)^2 = 0. A sum of two real squares vanishes only when both vanish, so a=3b,c=pb3.a = 3b, \qquad c = \frac{pb}{3}.

Now impose the triangle inequalities on the sides 3b3b, bb, pb3\tfrac{pb}{3}, and divide throughout by the positive number bb: 1+p3>3 ⇒ p>6,3+1>p3 ⇒ p<12,1 + \frac p3 > 3 \ \Rightarrow\ p > 6, \qquad 3 + 1 > \frac p3 \ \Rightarrow\ p < 12, while the third inequality 3+p3>13 + \tfrac p3 > 1 holds automatically.

So p∈{7,8,9,10,11}p \in \{7, 8, 9, 10, 11\}, and every one of these does produce a genuine triangle: take b=3b = 3, so that the sides are a=9,b=3,c=p,a = 9, \qquad b = 3, \qquad c = p, and the two binding inequalities read 3+p>93 + p > 9 and 9+3>p9 + 3 > p, both of which hold for each of the five values. There are 55 of them.

Answer 5

Solution: IOQM 2025 Part SEP, Q7

Key idea

Putting n=0n = 0 collapses the functional equation into f(1)=f(m)−m+1f(1) = f(m) - m + 1, which says that ff is linear.

Set n=0n = 0 in the relation. Since mn+1=1mn + 1 = 1, f(1)=f(m)f(0)−f(0)−m+2=f(m)−m+1,f(1) = f(m)f(0) - f(0) - m + 2 = f(m) - m + 1, using f(0)=1f(0) = 1. Rearranged, f(m)=f(1)+m−1f(m) = f(1) + m - 1 for every integer mm, so ff is linear.

Now find the constant. Taking m=0m = 0 in that formula gives f(0)=f(1)−1f(0) = f(1) - 1, and f(0)=1f(0) = 1, so f(1)=2f(1) = 2 and f(m)=m+1.f(m) = m + 1. It is worth confirming this really satisfies the original relation, since we only used one special case of it: f(mn+1)=mn+2,f(mn+1) = mn + 2, f(m)f(n)−f(n)−m+2=(m+1)(n+1)−(n+1)−m+2=mn+2. ✓f(m)f(n) - f(n) - m + 2 = (m+1)(n+1) - (n+1) - m + 2 = mn + 2. \ \checkmark

Finally, ∑k=1Nf(k)=∑k=1N(k+1)=N(N+1)2+N=N(N+3)2.\sum_{k=1}^{N} f(k) = \sum_{k=1}^{N}(k+1) = \frac{N(N+1)}{2} + N = \frac{N(N+3)}{2}. For N=12N = 12 this is 12⋅152=90<100\tfrac{12 \cdot 15}{2} = 90 < 100, while N=13N = 13 gives 13⋅162=104\tfrac{13 \cdot 16}{2} = 104. So N=12N = 12.

Answer 12

Solution: IOQM 2025 Part SEP, Q7

Key idea

The combination f(x)−4g(x)f(x) - 4g(x) vanishes at −2-2 and at 33, so it is a known multiple of (x+2)(x−3)(x+2)(x-3), and the data at x=7x = 7 fixes the multiple.

Set h(x)=f(x)−4g(x)h(x) = f(x) - 4g(x). It has degree at most 22, and the two given ratios say h(−2)=h(3)=0h(-2) = h(3) = 0. A polynomial of degree at most 22 with those two roots is h(x)=k(x+2)(x−3)h(x) = k(x+2)(x-3) for some constant kk.

Evaluate at x=7x = 7, where all three values are given: h(7)=f(7)−4g(7)=12−4(−6)=36,k⋅9⋅4=36,k=1.h(7) = f(7) - 4g(7) = 12 - 4(-6) = 36, \qquad k \cdot 9 \cdot 4 = 36, \qquad k = 1.

Now evaluate at x=5x = 5: f(5)=4g(5)+h(5)=4⋅2+7⋅2=8+14=22.f(5) = 4g(5) + h(5) = 4 \cdot 2 + 7 \cdot 2 = 8 + 14 = 22.

Neither ff nor gg was ever determined, and neither needed to be. When two polynomials are tied together only through their ratio, the difference that the ratio kills is usually the object to work with.

Answer 22

Solution: IOQM 2026, Q11

Key idea

Comparing degrees forces PP to be quadratic. The two sides then differ by a multiple of x2−P(x)x^2 - P(x), so either P(x)=x2P(x) = x^2 or the other factor vanishes, and that happens only for P(x)=−x2P(x) = -x^2.

If PP has degree n≥1n \ge 1, then P(x2)P(x^2) has degree 2n2n and P(P(x))P(P(x)) has degree n⋅nn \cdot n. So n2=2nn^2 = 2n, and n=2n = 2. Write P(x)=ax2+bx+cP(x) = ax^2 + bx + c with a≠0a \ne 0.

Now subtract, keeping P(x)P(x) whole wherever it appears: P(x2)−P(P(x))=a(x4−P(x)2)+b(x2−P(x))=(x2−P(x))(a(x2+P(x))+b).\begin{aligned} P(x^2) - P(P(x)) &= a\bigl(x^4 - P(x)^2\bigr) + b\bigl(x^2 - P(x)\bigr) \\ &= \bigl(x^2 - P(x)\bigr)\bigl(a(x^2 + P(x)) + b\bigr). \end{aligned} This has to be the zero polynomial. A product of two non-zero polynomials is never the zero polynomial, since its leading term is the product of their leading terms, so one factor is zero.

Either P(x)=x2P(x) = x^2, or a(x2+P(x))+b=0a(x^2 + P(x)) + b = 0 for every xx. The second reads (a2+a)x2+abx+(ac+b)=0,(a^2 + a)x^2 + abx + (ac + b) = 0, so a2+a=0a^2 + a = 0, which with a≠0a \ne 0 gives a=−1a = -1; then ab=0ab = 0 gives b=0b = 0, and ac+b=0ac + b = 0 gives c=0c = 0. That is P(x)=−x2P(x) = -x^2.

Both work: for P(x)=x2P(x) = x^2 each side is x4x^4, and for P(x)=−x2P(x) = -x^2 each side is −x4-x^4. There are 22 such polynomials.

Answer 2

Solution: PRMO 2012, Q18

Key idea

Writing n=[x]n = [x] makes x2=7n−5x^2 = 7n - 5, so each candidate nn gives one positive xx, and only four of them have the right integer part.

Put n=[x]n = [x], so x2=7n−5x^2 = 7n - 5. The right-hand side must be non-negative, and a negative xx is impossible: it would force n≤−1n \le -1 and then 7n−5<07n - 5 < 0. So x=7n−5x = \sqrt{7n-5} with n≥1n \ge 1, and the condition to check is n≤7n−5<n+1n \le \sqrt{7n-5} < n+1.

Squaring, that condition is n2≤7n−5<(n+1)2n^2 \le 7n - 5 < (n+1)^2, which is a test on whole numbers and needs no square roots at all: n7n−5n2(n+1)21214✓2949×316916×4231625✓5302536✓6373649✓7444964×\begin{array}{r|c|c|c|l} n & 7n-5 & n^2 & (n+1)^2 & \\ \hline 1 & 2 & 1 & 4 & \checkmark \\ 2 & 9 & 4 & 9 & \times \\ 3 & 16 & 9 & 16 & \times \\ 4 & 23 & 16 & 25 & \checkmark \\ 5 & 30 & 25 & 36 & \checkmark \\ 6 & 37 & 36 & 49 & \checkmark \\ 7 & 44 & 49 & 64 & \times \end{array} The two failures at n=2n = 2 and n=3n = 3 are the cases where 7n−57n-5 reaches (n+1)2(n+1)^2 exactly, and beyond n=7n = 7 the root falls short of nn for good, since n2−7n+5>0n^2 - 7n + 5 > 0 from there on.

The four roots are 2\sqrt2, 23\sqrt{23}, 30\sqrt{30}, 37\sqrt{37}, and the sum of their squares is 2+23+30+37=92.2 + 23 + 30 + 37 = 92.

Answer 92

Solution: PRMO 2013, Q16

Key idea

Solving the quadratic for bb and substituting turns the pair into r4−4r2+3=0r^4 - 4r^2 + 3 = 0, whose four roots give three values of bb that cancel.

Let rr be a common root. From g(r)=0g(r) = 0 we get r2+br−3=0r^2 + br - 3 = 0, and r=0r = 0 is impossible since it would give −3=0-3 = 0. So b=3−r2r.b = \frac{3 - r^2}{r}.

Substituting into f(r)=r3−3r+b=0f(r) = r^3 - 3r + b = 0 and multiplying by rr, r4−3r2+3−r2=0,that isr4−4r2+3=0,r^4 - 3r^2 + 3 - r^2 = 0, \qquad \text{that is} \qquad r^4 - 4r^2 + 3 = 0, which factors as (r2−1)(r2−3)=0\left(r^2 - 1\right)\left(r^2 - 3\right) = 0. So r=±1r = \pm 1 or r=±3r = \pm\sqrt3, and the corresponding values of bb are r=1: b=2,r=−1: b=−2,r=±3: b=0.r = 1: \ b = 2, \qquad r = -1: \ b = -2, \qquad r = \pm\sqrt3: \ b = 0.

Each really occurs: at b=2b = 2 the two polynomials share the root 11, at b=−2b = -2 they share −1-1, and at b=0b = 0 they share 3\sqrt3, since then f(x)=x(x2−3)f(x) = x(x^2-3) and g(x)=x2−3g(x) = x^2 - 3.

The distinct values are 22, −2-2 and 00, and their sum is 0.0.

Answer 0

Solution: IOQM 2020, Q11

Key idea

Eliminating bb between the two polynomials collapses everything to a(1−r2)=0a(1-r^2)=0, so either aa vanishes or the shared root is ±1\pm 1, leaving three small families to count.

image

Suppose some real number rr is a zero of both polynomials, so that r2+ar+b=0andr3+br+a=0.r^2 + ar + b = 0 \qquad \text{and} \qquad r^3 + br + a = 0. Two equations in the unknowns aa and bb invite elimination, and the first is already solved for bb, namely b=−r2−arb = -r^2 - ar. Substituting that into the second and expanding, r3+(−r2−ar)r+a=r3−r3−ar2+a=a(1−r2),r^3 + (-r^2 - ar)r + a = r^3 - r^3 - ar^2 + a = a(1 - r^2), so the condition for a common root is a(1−r2)=0a(1 - r^2) = 0. That one line does most of the work: every solution falls into one of three families: a=0a = 0, or r=1r = 1, or r=−1r = -1. We count each in turn and then reconcile the overlaps.

The family a=0a = 0

The quadratic becomes x2+bx^2 + b, which has a real zero exactly when b≤0b \leq 0, the zeros being ±−b\pm\sqrt{-b}. The cubic becomes x3+bx=x(x2+b)x^3 + bx = x(x^2 + b), which shares those zeros. So every bb in {−5,−4,−3,−2,−1,0}\{-5, -4, -3, -2, -1, 0\} succeeds, giving 66 pairs.

The family r=1r = 1

Substituting r=1r = 1 into either equation gives the same condition 1+a+b=01 + a + b = 0, so b=−1−ab = -1-a. Requiring bb to lie in XX restricts aa to {−5,−4,…,4}\{-5, -4, \ldots, 4\}, which is 1010 pairs.

The family r=−1r = -1

Substituting r=−1r = -1 gives 1−a+b=01 - a + b = 0 in both, so b=a−1b = a - 1, and requiring b∈Xb \in X restricts aa to {−4,−3,…,5}\{-4, -3, \ldots, 5\}, another 1010 pairs.

These families are not disjoint, and it turns out they meet in a single point. Setting −1−a=a−1-1-a = a-1 forces a=0a = 0 and b=−1b = -1, and that same pair (0,−1)(0,-1) also satisfies the condition of the first family. So (0,−1)(0,-1) lies in all three families and in each of the three pairwise intersections, and inclusion-exclusion gives

∣S∣=6+10+10−1−1−1+1=24.|S| = 6 + 10 + 10 - 1 - 1 - 1 + 1 = 24.

Answer 24

Solution: IOQM 2020, Q21

Key idea

Each year’s share is an age divided by the total of the ages, and five years later every age has risen by five while the total has risen by fifteen. Three such comparisons pin down the ages, and the one absolute amount then fixes NN.

Write aa, bb, cc for the three ages in the first year and S=a+b+cS = a+b+c for their total, so that in the first year AA receives Na/SNa/S, and so on. Five years later, at the sixth payment, the ages are a+5a+5, b+5b+5, c+5c+5 and their total is S+15S+15.

The first clue is that AA received half the total in the first year, so Na/S=N/2Na/S = N/2, giving a=S/2a = S/2 immediately. The second clue compares AA’s sixth payment with her first: N(a+5)S+15=67⋅NaS=67⋅N2=3N7,\frac{N(a+5)}{S+15} = \frac67 \cdot \frac{Na}{S} = \frac67 \cdot \frac{N}{2} = \frac{3N}{7}, so a+5S+15=37\dfrac{a+5}{S+15} = \dfrac37. Substituting a=S/2a = S/2 and clearing denominators, 7(S2+5)=3(S+15)7\left(\tfrac{S}{2}+5\right) = 3(S+15), which simplifies to S/2=10S/2 = 10 and hence S=20S = 20. So a=10a = 10 and b+c=10b + c = 10.

The third clue, that CC received twice her first-year amount, isolates cc: N(c+5)35=2⋅Nc20=Nc10,so10(c+5)=35c,\frac{N(c+5)}{35} = 2 \cdot \frac{Nc}{20} = \frac{Nc}{10}, \qquad \text{so} \qquad 10(c+5) = 35c, giving 25c=5025c = 50 and c=2c = 2, and therefore b=8b = 8. Notice that all three of these clues were ratios, so NN has cancelled every time and the ages were determined without it.

Only now do we use the one clue that mentions an actual sum of money. Working in thousands of rupees, BB received one thousand less at the sixth payment than at the first: 13N35=8N20−1=14N35−1.\frac{13N}{35} = \frac{8N}{20} - 1 = \frac{14N}{35} - 1. Hence 1=N/351 = N/35 and N=35N = 35.

Answer 35

Solution: IOQM 2021 Part A, Q10

Key idea

The two surds multiply to something tiny: (7+5)(7−5)=2(\sqrt7+\sqrt5)(\sqrt7-\sqrt5) = 2. So the required value is 4/α4/\alpha, and the degree-four equation α\alpha satisfies rearranges to express 4/α4/\alpha as a polynomial in α\alpha.

Building the polynomial

Write α=7+5\alpha = \sqrt7 + \sqrt5. The first thing to notice is that (7+5)(7−5)=7−5=2,so7−5=2α.(\sqrt7 + \sqrt5)(\sqrt7 - \sqrt5) = 7 - 5 = 2, \qquad \text{so} \qquad \sqrt7 - \sqrt5 = \frac{2}{\alpha}. The condition P(α)=2(7−5)P(\alpha) = 2(\sqrt7 - \sqrt5) therefore reads P(α)=4αP(\alpha) = \dfrac{4}{\alpha}. That is not yet a polynomial condition, but it is close to one: multiplying by α\alpha gives αP(α)=4\alpha P(\alpha) = 4.

Next find what α\alpha satisfies. Squaring, α2=12+235\alpha^2 = 12 + 2\sqrt{35}, so α2−12=235\alpha^2 - 12 = 2\sqrt{35} and squaring again gives α4−24α2+144=140\alpha^4 - 24\alpha^2 + 144 = 140, that is α4−24α2+4=0.\alpha^4 - 24\alpha^2 + 4 = 0.

Now read that backwards as a statement about the number 44: 4=24α2−α4=α(24α−α3).4 = 24\alpha^2 - \alpha^4 = \alpha\left(24\alpha - \alpha^3\right). Comparing with αP(α)=4\alpha P(\alpha) = 4 and cancelling α\alpha, which is certainly not zero, P(x)=24x−x3.P(x) = 24x - x^3. This has integer coefficients and degree three, and P(2)=48−8=40P(2) = 48 - 8 = 40.

Why degree three is really necessary

The question asks for the polynomial of least degree, so the construction alone does not finish it. Everything below is about ruling out degrees one and two, and the degree-zero case as well.

The whole difficulty is one fact: α\alpha satisfies no equation with rational coefficients of degree below four. That comes down to showing that x4−24x2+4x^4 - 24x^2 + 4 does not factor over the rationals. We will justify that implication after checking the factorisations.

It has no rational root. To see why only integer candidates are needed, write a rational root as h/kh/k in lowest terms with k>0k>0. Clearing denominators gives h4−24h2k2+4k4=0h^4-24h^2k^2+4k^4=0, so kk divides h4h^4. Coprimality forces k=1k=1. Then hh divides 44 by the same equation, and x=1,2,4x = 1, 2, 4 give −19-19, −76-76 and −124-124, with the negative values the same because only even powers appear. So the only possible factorisation is into two quadratics, and being monic with no cubic term they must take the form (x2+ux+v)(x2−ux+w)=x4+(v+w−u2)x2+u(w−v)x+vw.\left(x^2 + ux + v\right)\left(x^2 - ux + w\right) = x^4 + (v + w - u^2)x^2 + u(w-v)x + vw. Matching coefficients gives u(w−v)=0u(w-v) = 0, v+w−u2=−24v + w - u^2 = -24 and vw=4vw = 4, all rational.

  • If u=0u = 0, then v+w=−24v + w = -24 and vw=4vw = 4, so vv and ww are roots of t2+24t+4t^2 + 24t + 4, whose discriminant 576−16=560576 - 16 = 560 is not the square of a rational.

  • If w=vw = v, then v2=4v^2 = 4 gives v=±2v = \pm 2, and 2v−u2=−242v - u^2 = -24 makes u2u^2 equal to 2828 or 2020, neither the square of a rational.

So F(x)=x4−24x2+4F(x)=x^4-24x^2+4 does not factor over the rationals. Suppose a non-zero rational polynomial QQ of degree below 44 also vanished at α\alpha. Repeated polynomial division, just as in the Euclidean algorithm for integers, reduces degrees and finds a common divisor of F,QF,Q. That divisor must be constant: any non-constant divisor of the unfactorable FF has degree 44, too large to divide QQ. Running the divisions backwards therefore gives rational polynomials U,VU,V with UF+VQ=1UF+VQ=1. Substituting x=αx=\alpha gives 0=10=1, a contradiction. Thus no non-zero rational polynomial of degree below 44 vanishes at α\alpha.

Now suppose some PP of degree at most two worked. Then xP(x)−4xP(x) - 4 would be a non-zero polynomial of degree at most three vanishing at α\alpha, which we have just ruled out. So degree three is least, and P(x)=24x−x3P(x) = 24x - x^3 is the polynomial the question asks for, with P(2)=40P(2) = 40.

Answer 40

Solution: IOQM 2022, Q14

Key idea

Let s=x+y+zs = x+y+z. Then y+z=s−xy + z = s - x, and the identity xks−x=sks−x−(a polynomial)\frac{x^k}{s-x} = \frac{s^k}{s-x} - (\text{a polynomial}) turns all three given sums into linear statements about the single quantity ∑1s−x\sum \frac{1}{s-x}.

Write s=x+y+zs = x + y + z, so each denominator is ss minus the corresponding numerator’s variable, and set T=∑1s−xT = \sum \dfrac{1}{s-x}, the sum running cyclically.

Take the three givens in turn. Since xs−x=ss−x−1\dfrac{x}{s-x} = \dfrac{s}{s-x} - 1, the first says sT−3=9,sosT=12.sT - 3 = 9, \qquad \text{so} \qquad sT = 12. For the second, note x2=s2−(s−x)(s+x)x^2 = s^2 - (s-x)(s+x), so x2s−x=s2s−x−(s+x)\dfrac{x^2}{s-x} = \dfrac{s^2}{s-x} - (s+x) and summing gives s2T−(3s+s)=64s^2 T - (3s + s) = 64, that is s2T−4s=64s^2T - 4s = 64. Using sT=12sT = 12 this reads 12s−4s=6412s - 4s = 64, so s=8,T=128=32.s = 8, \qquad T = \tfrac{12}{8} = \tfrac32. For the third, x3=s3−(s−x)(s2+sx+x2)x^3 = s^3 - (s-x)(s^2+sx+x^2) gives s3T−(3s2+s2+∑x2)=488s^3T - \left(3s^2 + s^2 + \sum x^2\right) = 488. With s=8s = 8 and T=32T = \tfrac32 this is 768−256−∑x2=488768 - 256 - \sum x^2 = 488, so ∑x2=24\sum x^2 = 24.

Two of the three elementary symmetric functions follow at once. From ∑x2=s2−2∑xy\sum x^2 = s^2 - 2\sum xy we get ∑xy=64−242=20\sum xy = \tfrac{64-24}{2} = 20. For the third, evaluate TT directly: putting everything over the common denominator ∏(s−x)\prod (s-x), T=∑(s−y)(s−z)∏(s−x)=s2+∑xys3−s2⋅s+s∑xy−xyz=64+20160−xyz.T = \frac{\sum (s-y)(s-z)}{\prod (s-x)} = \frac{s^2 + \sum xy}{s^3 - s^2 \cdot s + s\sum xy - xyz} = \frac{64 + 20}{160 - xyz}. Setting this equal to 32\tfrac32 gives 168=3(160−xyz)168 = 3(160 - xyz), so xyz=104xyz = 104.

The requested quantity is now immediate, since xyz+yzx+zxy=x2+y2+z2xyz\dfrac{x}{yz} + \dfrac{y}{zx} + \dfrac{z}{xy} = \dfrac{x^2+y^2+z^2}{xyz}: 24104=313.\frac{24}{104} = \frac{3}{13}. Hence m=3m = 3, n=13n = 13 and m+n=16m + n = 16.

Answer 16

Solution: IOQM 2022, Q16

Key idea

Each equation solves for one unknown in terms of the next, so composing the three round the cycle gives a single quadratic in aa.

Rearrange each equation to isolate a variable. From 3ab+2=6b3ab + 2 = 6b we get 3b(a−2)=−23b(a-2) = -2, so b=23(2−a).b = \frac{2}{3(2-a)}. From 3bc+2=5c3bc + 2 = 5c we get c(3b−5)=−2c(3b-5) = -2, so c=25−3bc = \dfrac{2}{5-3b}, and from 3ca+2=4a3ca + 2 = 4a we get a=24−3ca = \dfrac{2}{4-3c}.

Now walk round the cycle. Substituting 3b=22−a3b = \dfrac{2}{2-a}, 5−3b=5(2−a)−22−a=8−5a2−a,soc=2(2−a)8−5a.5 - 3b = \frac{5(2-a) - 2}{2-a} = \frac{8-5a}{2-a}, \qquad \text{so} \qquad c = \frac{2(2-a)}{8-5a}. Then 3c=6(2−a)8−5a3c = \dfrac{6(2-a)}{8-5a} and 4−3c=4(8−5a)−6(2−a)8−5a=20−14a8−5a,4 - 3c = \frac{4(8-5a) - 6(2-a)}{8-5a} = \frac{20 - 14a}{8-5a}, so the last equation becomes a=2(8−5a)20−14aa = \dfrac{2(8-5a)}{20-14a}. Cross-multiplying, a(20−14a)=16−10a⟹7a2−15a+8=0,a(20 - 14a) = 16 - 10a \quad\Longrightarrow\quad 7a^2 - 15a + 8 = 0, which factors as (7a−8)(a−1)=0(7a - 8)(a - 1) = 0. The two promised values of the product come from these two roots.

If a=1a = 1 then b=23b = \tfrac{2}{3} and c=23c = \tfrac{2}{3}, so abc=49abc = \tfrac49. If a=87a = \tfrac87 then b=79b = \tfrac79 and c=34c = \tfrac34, so abc=87⋅79⋅34=23abc = \tfrac87 \cdot \tfrac79 \cdot \tfrac34 = \tfrac23.

Both are already in lowest terms, so rs=49\tfrac{r}{s} = \tfrac49 and tu=23\tfrac{t}{u} = \tfrac23, giving r+s+t+u=4+9+2+3=18r + s + t + u = 4 + 9 + 2 + 3 = 18.

Answer 18

Solution: IOQM 2025 Part SEP, Q7

Key idea

Subtracting the outer two entries gives n2−2an−a+2=0n^2 - 2an - a + 2 = 0, and integrality forces nn to be an integer, after which 2n+12n+1 must divide 99.

The three entries are consecutive integers, so the outer two differ by 22: a(n+1)2−(a+1)n2=2.a(n+1)^2 - (a+1)n^2 = 2. Expanding, the an2an^2 terms cancel and this becomes n2−2an−a+2=0.(∗)n^2 - 2an - a + 2 = 0. \tag{$*$}

Now, why must nn be an integer? The middle entry m2m^2 is an integer and so is the first, (a+1)n2(a+1)n^2. From (∗)(*) we have n2=2an+a−2n^2 = 2an + a - 2, so (a+1)n2=(a+1)(2an+a−2),(a+1)n^2 = (a+1)(2an + a - 2), which is an integer only if 2a(a+1)n2a(a+1)n is, so nn is rational. Write n=h/kn=h/k in lowest terms with k>0k>0. Clearing denominators in (∗)(*) gives h2−2ahk+(−a+2)k2=0h^2-2ahk+(-a+2)k^2=0, so kk divides h2h^2. Since h,kh,k are coprime, k=1k=1, and nn is an integer.

With nn an integer, solve (∗)(*) for aa: a(2n+1)=n2+2,a=n2+22n+1.a(2n+1) = n^2 + 2, \qquad a = \frac{n^2+2}{2n+1}. Multiply by 44 to clear the denominator neatly: 4(n2+2)=(2n+1)(2n−1)+94(n^2+2) = (2n+1)(2n-1) + 9, so 2n+12n+1 must divide 99. Hence 2n+1∈{±1,±3,±9}2n + 1 \in \{\pm1, \pm3, \pm9\} and n∈{0,1,4,−1,−2,−5}n \in \{0, 1, 4, -1, -2, -5\}.

The negative values of nn all make aa negative, which is not allowed, so three cases remain: nathe three consecutive integers020, 1, 2112, 3, 44248, 49, 50\begin{array}{c|c|c} n & a & \text{the three consecutive integers} \\ \hline 0 & 2 & 0,\ 1,\ 2 \\ 1 & 1 & 2,\ 3,\ 4 \\ 4 & 2 & 48,\ 49,\ 50 \end{array} All three are genuine, and the largest middle entry is m2=49.m^2 = 49.

Answer 49

Solution: IOQM 2025 Part SEP, Q7

Key idea

Q(x)Q(x) factors as (x2+1)(x2+x+1)(x^2+1)(x^2+x+1), and modulo each factor the enormous power of xx collapses at once, because x2≡−1x^2 \equiv -1 and x3≡1x^3 \equiv 1.

First factor the divisor: x4+x3+2x2+x+1=(x2+1)(x2+x+1),x^4+x^3+2x^2+x+1 = (x^2+1)(x^2+x+1), which one checks by expanding. Now reduce x2025x^{2025} modulo each factor.

Modulo x2+1x^2+1 we have x2≡−1x^2 \equiv -1, so x2025=(x2)1012⋅x≡(−1)1012x=x.x^{2025} = \left(x^2\right)^{1012} \cdot x \equiv (-1)^{1012}x = x. Modulo x2+x+1x^2+x+1 we have x3≡1x^3 \equiv 1, because x3−1=(x−1)(x2+x+1)x^3 - 1 = (x-1)(x^2+x+1), and 20252025 is divisible by 33, so x2025=(x3)675≡1.x^{2025} = \left(x^3\right)^{675} \equiv 1.

The remainder RR has degree at most 33 and satisfies both congruences. Write R(x)=x+(x2+1)(αx+β)R(x) = x + (x^2+1)(\alpha x + \beta), which automatically handles the first. For the second, work modulo x2+x+1x^2+x+1, where x2≡−x−1x^2 \equiv -x-1 and therefore x2+1≡−xx^2 + 1 \equiv -x: x+(−x)(αx+β)=x−αx2−βx≡x+α(x+1)−βx=(1+α−β)x+α.x + (-x)(\alpha x + \beta) = x - \alpha x^2 - \beta x \equiv x + \alpha(x+1) - \beta x = (1 + \alpha - \beta)x + \alpha. This must be the constant 11, so α=1\alpha = 1 and β=2\beta = 2. Hence R(x)=x+(x2+1)(x+2)=x3+2x2+2x+2,R(x) = x + (x^2+1)(x+2) = x^3 + 2x^2 + 2x + 2, and R(3)=27+18+6+2=53.R(3) = 27 + 18 + 6 + 2 = 53.

Answer 53

Solution: IOQM 2026, Q21

Key idea

Subtracting the equations in pairs, each difference factorises with a factor x−yx - y, y−zy - z or z−xz - x. So two of the unknowns are equal or all three equal 11, and a short check of each case leaves three values of the product.

Move the lone unknown in each equation to the right: x2−xy+y2=−z,y2−yz+z2=−x,z2−zx+x2=−y.x^2 - xy + y^2 = -z, \qquad y^2 - yz + z^2 = -x, \qquad z^2 - zx + x^2 = -y. Subtracting the third from the first, y2−z2−xy+xz=y−zy^2 - z^2 - xy + xz = y - z, which factorises as (y−z)(y+z−x−1)=0.(y - z)(y + z - x - 1) = 0. In the same way the second minus the first gives (z−x)(z+x−y−1)=0(z - x)(z + x - y - 1) = 0, and the third minus the second gives (x−y)(x+y−z−1)=0(x - y)(x + y - z - 1) = 0.

All three different. Then y+z−xy + z - x, z+x−yz + x - y and x+y−zx + y - z all equal 11. Adding the first two gives 2z=22z = 2, so z=1z = 1, and in the same way x=y=1x = y = 1, which is not three different numbers. This case is empty.

All three equal, say to tt. The first equation reads t2=−tt^2 = -t, so t=0t = 0 or t=−1t = -1, and the product (t2−t)3(t^2 - t)^3 is 00 or 23=82^3 = 8.

Exactly two equal. Renaming x→y→z→xx \to y \to z \to x carries each equation to the next and leaves the product unchanged, so it is enough to take y=zy = z with xx different. Then (x−y)(x+y−z−1)=0(x - y)(x + y - z - 1) = 0 gives x=1x = 1, and the second equation becomes 1+2y2=y21 + 2y^2 = y^2, that is y2=−1y^2 = -1. The other two equations then hold as well. No real number has square −1-1, which is why the question allows complex numbers, but nothing beyond y2=−1y^2 = -1 is needed: (x2−y)(y2−z)(z2−x)=(1−y)(−1−y)(−2)=2(1−y2)=4.(x^2 - y)(y^2 - z)(z^2 - x) = (1 - y)(-1 - y)(-2) = 2(1 - y^2) = 4.

Every value of the product is real, so the modulus is the ordinary size, and the distinct values are 00, 88 and 44. Their sum is 1212, with or without the 00.

Answer 12

Report an error on this page

Reports are stored by Netlify. See the privacy note.