Library · Between the Challenge and the Olympiad · Chapter 7

Sequences, Series and Recursions

Revised Report an error
On this page
  1. Problems
  2. Solutions
  3. Solution: PRMO 2012, Q9
  4. Solution: PRMO 2014, Q19
  5. Solution: PRMO 2017, Q5
  6. Solution: PRMO 2017, Q6
  7. Solution: PRMO 2017, Q16
  8. Solution: PRMO 2019, Q3
  9. Solution: PRMO 2019, Q7
  10. Solution: IOQM 2020, Q3
  11. Solution: IOQM 2025 Part SEP, Q28
  12. Solution: PRMO 2013, Q2
  13. Solution: PRMO 2017, Q11
  14. Solution: PRMO 2017, Q14
  15. Solution: PRMO 2018, Q16
  16. Solution: IOQM 2020, Q18
  17. Solution: IOQM 2020, Q20
  18. Solution: IOQM 2023, Q10
  19. Solution: IOQM 2026, Q16

Problems

Problem 1

Suppose that 4X1=5,5X2=6,6X3=7,…,126X123=127,127X124=1284^{X_1} = 5, 5^{X_2} = 6, 6^{X_3} = 7, \ldots , 126^{X_{123}} = 127, 127^{X_{124}} = 128. What is the value of the product X1X2…X124X_1 X_2 \ldots X_{124}?

Problem 2

Let x1,x2,⋯ ,x2014x_1, x_2, \cdots , x_{2014} be real numbers different from 11, such that x1+x2+⋯+x2014=1x_1 + x_2 + \cdots + x_{2014} = 1 and x11−x1+x21−x2+⋯+x20141−x2014=1.\frac{x_1}{1 - x_1} + \frac{x_2}{1 - x_2} + \cdots + \frac{x_{2014}}{1 - x_{2014}} = 1. What is the value of x121−x1+x221−x2+x321−x3+⋯+x201421−x2014 ?\frac{x_1^2}{1 - x_1} + \frac{x_2^2}{1 - x_2} + \frac{x_3^2}{1 - x_3} + \cdots + \frac{x_{2014}^2}{1 - x_{2014}} \, ?

Problem 3

Let u,v,wu, v, w be real numbers in geometric progression such that u>v>wu > v > w. Suppose u40=vn=w60u^{40} = v^n = w^{60}. Find the value of nn.

Problem 4

Let the sum ∑n=191n(n+1)(n+2)\sum_{n=1}^{9} \frac{1}{n(n+1)(n+2)} written in its lowest terms be pq\frac{p}{q}. Find the value of q−pq - p.

Problem 5

Five distinct 2-digit numbers are in a geometric progression. Find the middle term.

Problem 6

Let x1x_1 be a positive real number and for every integer n≥1n \geq 1 let xn+1=1+x1x2…xn−1xnx_{n+1} = 1 + x_1 x_2 \ldots x_{n-1} x_n. If x5=43x_5 = 43, what is the sum of digits of the largest prime factor of x6x_6?

Problem 7

On a clock, there are two instants between 1212 noon and 11 PM, when the hour hand and the minute hand are at right angles. The difference in minutes between these two instants is written as a+bca + \frac{b}{c}, where a,b,ca, b, c are positive integers, with b<cb < c and b/cb/c in the reduced form. What is the value of a+b+ca + b + c?

Problem 8

If ∑k=1N2k+1(k2+k)2=0.9999\displaystyle\sum_{k=1}^{N} \frac{2k+1}{(k^2+k)^2} = 0.9999 then determine the value of NN.

Problem 9

Find the largest positive integer nn for which the inequality ∑k=12n(−1)kk2<100\sum_{k=1}^{2n} (-1)^k k^2 < 100 holds.

Problem 10

Let Sn=∑k=0n1k+1+kS_n = \sum_{k=0}^{n} \dfrac{1}{\sqrt{k+1} + \sqrt{k}}. What is the value of ∑n=1991Sn+Sn−1\sum_{n=1}^{99} \dfrac{1}{S_n + S_{n-1}}?

Problem 11

Let f(x)=sin⁡x3+cos⁡3x10f(x) = \sin\frac{x}{3} + \cos\frac{3x}{10} for all real xx. Find the least natural number nn such that f(nπ+x)=f(x)f(n\pi + x) = f(x) for all real xx.

Problem 12

Suppose xx is a positive real number such that {x},[x]\{x\}, [x] and xx are in a geometric progression. Find the least positive integer nn such that xn>100x^n > 100. (Here [x][x] denotes the integer part of xx and {x}=x−[x]\{x\} = x - [x].)

Problem 13

What is the value of ∑1≤i<j≤10i+j=odd(i+j)−∑1≤i<j≤10i+j=even(i+j)?\sum_{\substack{1 \leq i < j \leq 10 \\ i+j=\text{odd}}} (i+j) - \sum_{\substack{1 \leq i < j \leq 10 \\ i+j=\text{even}}} (i+j)?

Problem 14

If ∑k=140(1+1k2+1(k+1)2)=a+bc\sum_{k=1}^{40} \left( \sqrt{1 + \frac{1}{k^2} + \frac{1}{(k+1)^2}} \right) = a + \frac{b}{c} where a,b,c∈Na, b, c \in \mathbb{N}, b<cb < c, gcd⁡(b,c)=1\gcd(b,c) = 1, then what is the value of a+ba + b?

Problem 15

A group of women working together at the same rate can build a wall in 45 hours. When the work started, all the women did not start working together. They joined the work over a period of time, one by one, at equal intervals. Once at work, each one stayed till the work was complete. If the first woman worked 5 times as many hours as the last woman, for how many hours did the first woman work?

Problem 16

The sequence ⟨an⟩n≥0\langle a_n \rangle_{n \geq 0} is defined by a0=1,a1=−4a_0 = 1, a_1 = -4 and an+2=−4an+1−7ana_{n+2} = -4a_{n+1} - 7a_n, for n≥0n \geq 0. Find the number of positive integer divisors of a502−a49a51a_{50}^2 - a_{49}a_{51}.

Problem 17

A sequence a1,a2,a3,…a_1, a_2, a_3, \ldots of real numbers satisfies an+3−an+2an−an+1=an+3+an+2an+an+1\frac{a_{n+3} - a_{n+2}}{a_n - a_{n+1}} = \frac{a_{n+3} + a_{n+2}}{a_n + a_{n+1}} for all n≥1n \geq 1. Suppose a55=6a_{55} = 6, a66=2a_{66} = 2 and a77=1a_{77} = 1. Let NN denote the sum a12+a22+⋯+a20262a_1^2 + a_2^2 + \cdots + a_{2026}^2. What is the sum of the digits of NN?

Solutions

Solution: PRMO 2012, Q9

Key idea

Each XiX_i is a logarithm, and the product telescopes to log⁡4128\log_4 128.

From 4X1=54^{X_1} = 5 we get X1=log⁡45=ln⁡5ln⁡4X_1 = \log_4 5 = \dfrac{\ln 5}{\ln 4}, and in general Xi=ln⁡(i+4)ln⁡(i+3)X_i = \dfrac{\ln(i+4)}{\ln(i+3)}.

Multiplying them all, every logarithm cancels against the next: X1X2⋯X124=ln⁡5ln⁡4⋅ln⁡6ln⁡5⋯ln⁡128ln⁡127=ln⁡128ln⁡4=log⁡4128.X_1 X_2 \cdots X_{124} = \frac{\ln 5}{\ln 4}\cdot\frac{\ln 6}{\ln 5}\cdots\frac{\ln 128}{\ln 127} = \frac{\ln 128}{\ln 4} = \log_4 128.

Finally 128=27128 = 2^7 and 4=224 = 2^2, so log⁡4128=72.\log_4 128 = \frac{7}{2}.

Answer 7/27/2

Solution: PRMO 2014, Q19

Key idea

x21−x=x1−x−x\dfrac{x^2}{1-x} = \dfrac{x}{1-x} - x, so the sum asked for is the difference of the two sums given.

The identity to spot is x1−x−x=x−x(1−x)1−x=x21−x,\frac{x}{1-x} - x = \frac{x - x(1-x)}{1-x} = \frac{x^2}{1-x}, valid whenever x≠1x \ne 1, which the problem guarantees.

Summing it over all 20142014 values, ∑i=12014xi21−xi=∑i=12014xi1−xi−∑i=12014xi=1−1=0.\sum_{i=1}^{2014} \frac{x_i^2}{1-x_i} = \sum_{i=1}^{2014} \frac{x_i}{1-x_i} - \sum_{i=1}^{2014} x_i = 1 - 1 = 0.

Neither the number 20142014 nor the individual values played any part; any list with the two stated sums equal gives zero.

Answer 0

Solution: PRMO 2017, Q5

Key idea

Taking logarithms turns the geometric progression into an arithmetic one and the three equal powers into three reciprocals, after which the middle term is the harmonic mean.

A geometric progression has v2=uwv^2=uw. Neither uu nor ww can be zero: the equality u40=w60u^{40}=w^{60} would make both zero, contradicting u>wu>w. Thus uw=v2>0uw=v^2>0, so uu and ww have the same sign. The common value K=u40=w60K=u^{40}=w^{60} is positive. It cannot equal 11, because then ∣u∣=∣w∣=1|u|=|w|=1 and the common sign would give u=wu=w. Also n≠0n\ne0, since v0=1v^0=1 could not equal KK. Write L=log⁡K≠0L = \log K \ne 0 and take logarithms of absolute values: 40log⁡∣u∣=nlog⁡∣v∣=60log⁡∣w∣=L,40 \log|u| = n \log|v| = 60 \log|w| = L, so log⁡∣u∣=L40,log⁡∣v∣=Ln,log⁡∣w∣=L60.\log|u| = \frac{L}{40}, \qquad \log|v| = \frac{L}{n}, \qquad \log|w| = \frac{L}{60}.

A geometric progression has v2=uwv^2 = uw. Taking absolute values of both sides preserves that, since ∣v2∣=∣v∣2|v^2| = |v|^2 and ∣uw∣=∣u∣∣w∣|uw| = |u||w|, so ∣v∣2=∣u∣∣w∣|v|^2 = |u||w| and the logarithms are in arithmetic progression: 2log⁡∣v∣=log⁡∣u∣+log⁡∣w∣,2Ln=L40+L60.2\log|v| = \log|u| + \log|w|, \qquad \frac{2L}{n} = \frac{L}{40} + \frac{L}{60}. Cancelling LL and adding the fractions, 2n=3+2120=124,n=48.\frac2n = \frac{3 + 2}{120} = \frac{1}{24}, \qquad n = 48.

In words, nn is the harmonic mean of 4040 and 6060, which is what an arithmetic progression of logarithms always produces from equal powers.

Answer 48

Solution: PRMO 2017, Q6

Key idea

1n(n+1)(n+2)\dfrac{1}{n(n+1)(n+2)} is half the difference of two consecutive terms of 1n(n+1)\dfrac{1}{n(n+1)}, so the sum collapses to its first and last pieces.

The key identity is 1n(n+1)(n+2)=12(1n(n+1)−1(n+1)(n+2)),\frac{1}{n(n+1)(n+2)} = \frac12\left(\frac{1}{n(n+1)} - \frac{1}{(n+1)(n+2)}\right), which one checks by putting the right-hand side over a common denominator: the numerator is 12((n+2)−n)=1\tfrac12\bigl((n+2) - n\bigr) = 1.

Summing from n=1n = 1 to 99, every interior term cancels against its neighbour and only the ends survive: ∑n=191n(n+1)(n+2)=12(11⋅2−110⋅11)=12⋅54110=27110.\sum_{n=1}^{9} \frac{1}{n(n+1)(n+2)} = \frac12\left(\frac{1}{1 \cdot 2} - \frac{1}{10 \cdot 11}\right) = \frac12 \cdot \frac{54}{110} = \frac{27}{110}.

Since 110=2⋅5⋅11110 = 2 \cdot 5 \cdot 11 shares no factor with 2727, the fraction is already in lowest terms, so p=27p = 27, q=110q = 110 and q−p=83.q - p = 83.

Answer 83

Solution: PRMO 2017, Q16

Key idea

With ratio p/qp/q in lowest terms the first term must be a multiple of q4q^4 and the last of p4p^4, and the two-digit range leaves only q4=16q^4 = 16, p4=81p^4 = 81.

Let the ratio be rr. It is rational, since consecutive terms are integers, so write r=p/qr = p/q in lowest terms with p≠qp \ne q. Reversing the progression if necessary, take p>qp > q, so the terms increase.

For the fifth term ar4=ap4/q4a r^4 = a p^4/q^4 to be an integer with gcd⁡(p,q)=1\gcd(p, q) = 1, the first term aa must be a multiple of q4q^4, say a=cq4a = cq^4. The five terms are then cq4,cq3p,cq2p2,cqp3,cp4,cq^4, \quad cq^3p, \quad cq^2p^2, \quad cqp^3, \quad cp^4, all integers automatically.

Now impose the two-digit range. The smallest term is at least 1010 and the largest at most 9999, so cq4≥10,cp4≤99,hence(pq)4≤9910<10.cq^4 \ge 10, \qquad cp^4 \le 99, \qquad \text{hence} \qquad \left(\frac{p}{q}\right)^4 \le \frac{99}{10} < 10. Since 9910<16=24\tfrac{99}{10} < 16 = 2^4, that gives p/q<2p/q < 2. In particular qq cannot be 11, since p>qp > q would then make p/qp/q at least 22; so q≥2q \ge 2. And cp4≤99cp^4 \le 99 with c≥1c \ge 1 gives p4≤99p^4 \le 99, so p≤3p \le 3. The only coprime pair with 3≥p>q≥23 \ge p > q \ge 2 is (p,q)=(3,2)(p,q) = (3,2), and then cq4=16c≥10cq^4 = 16c \ge 10 and cp4=81c≤99cp^4 = 81c \le 99 force c=1c = 1.

The progression is therefore 16,24,36,54,81,16, \quad 24, \quad 36, \quad 54, \quad 81, five distinct two-digit numbers, and its middle term is 3636.

Answer 36

Solution: PRMO 2019, Q3

Key idea

The product of the first nn terms is xn+1−1x_{n+1} - 1, and it is also xn(xn−1)x_n(x_n - 1), so the sequence obeys xn+1=xn2−xn+1x_{n+1} = x_n^2 - x_n + 1 and the value of x1x_1 never has to be found.

Write Pn=x1x2⋯xnP_n = x_1 x_2 \cdots x_n, so that the rule reads xn+1=1+Pnx_{n+1} = 1 + P_n. For n≥2n \ge 2 this gives Pn−1=xn−1P_{n-1} = x_n - 1, and therefore Pn=Pn−1 xn=(xn−1)xn.P_n = P_{n-1}\, x_n = (x_n - 1)x_n. Substituting back, xn+1=1+xn(xn−1)=xn2−xn+1(n≥2).x_{n+1} = 1 + x_n(x_n - 1) = x_n^2 - x_n + 1 \qquad (n \ge 2).

That is all we need. From x5=43x_5 = 43, x6=432−43+1=1849−42=1807.x_6 = 43^2 - 43 + 1 = 1849 - 42 = 1807. Factorising, 1807=13×1391807 = 13 \times 139, and 139139 is prime, since it is divisible by none of 22, 33, 55, 77, 1111 and 132=16913^2 = 169 already exceeds it. The largest prime factor is 139139, whose digits sum to 1+3+9=131 + 3 + 9 = 13.

Recovering x1x_1 is possible but unnecessary, and that is the lesson. Running the one-step rule backwards, x5=43x_5 = 43 gives x4=7x_4 = 7, then x3=3x_3 = 3, x2=2x_2 = 2 and x1=1x_1 = 1, each from a quadratic with one positive root. None of that was needed, because the rule steps from x5x_5 to x6x_6 directly. Whenever a recursively defined sequence is pinned down by one of its later values, look for a rule connecting consecutive terms before trying to unwind the whole thing back to the start.

Answer 13

Solution: PRMO 2019, Q7

Key idea

The minute hand gains on the hour hand at a steady 5.55.5 degrees per minute, so the two right-angle instants are where that gain reaches 90∘90^{\circ} and 270∘270^{\circ}.

image

Measure time tt in minutes after noon, when both hands point at 1212. The minute hand turns 360∘360^{\circ} per hour, so it is at 6t6t degrees; the hour hand turns 30∘30^{\circ} per hour, so it is at 12t\tfrac12 t degrees. The angle the minute hand has gained is 6t−12t=112 t,6t - \tfrac12 t = \tfrac{11}{2}\,t, which increases steadily from 00 to 330∘330^{\circ} as tt runs from 00 to 6060.

The hands are perpendicular when this gain is 90∘90^{\circ} or 270∘270^{\circ}, both of which occur in range: 112t=90  ⟹  t=18011,112t=270  ⟹  t=54011.\tfrac{11}{2}t = 90 \implies t = \frac{180}{11}, \qquad \tfrac{11}{2}t = 270 \implies t = \frac{540}{11}. Their difference is 540−18011=36011=32+811.\frac{540 - 180}{11} = \frac{360}{11} = 32 + \frac{8}{11}. So a=32a = 32, b=8b = 8, c=11c = 11, and since gcd⁡(8,11)=1\gcd(8,11) = 1 and 8<118 < 11 this is the required reduced form. Hence a+b+c=51a + b + c = 51.

Answer 51

Solution: IOQM 2020, Q3

Key idea

The numerator 2k+12k+1 is the gap between two consecutive squares, and once that is noticed each term becomes a difference of two reciprocal squares, so the sum collapses.

A sum like this one is not meant to be evaluated term by term, and the shape of the numerator is the hint. Since k2+k=k(k+1)k^2 + k = k(k+1), the denominator is k2(k+1)2k^2(k+1)^2, and the numerator 2k+12k+1 is exactly (k+1)2−k2(k+1)^2 - k^2. Writing the term with that in mind, 2k+1(k2+k)2=(k+1)2−k2k2(k+1)2=1k2−1(k+1)2,\frac{2k+1}{(k^2+k)^2} = \frac{(k+1)^2 - k^2}{k^2 (k+1)^2} = \frac{1}{k^2} - \frac{1}{(k+1)^2}, where the last step is just splitting the fraction and cancelling.

Each term is now the difference of two consecutive members of the sequence 1/k21/k^2, so when we add them from k=1k=1 to k=Nk=N every interior quantity is created once and destroyed once, and only the two ends survive: ∑k=1N2k+1(k2+k)2=1−1(N+1)2.\sum_{k=1}^{N} \frac{2k+1}{(k^2+k)^2} = 1 - \frac{1}{(N+1)^2}.

Setting this against the given value 0.99990.9999, which is 1−1100001 - \tfrac{1}{10000}, we need (N+1)2=10000(N+1)^2 = 10000, so N+1=100N + 1 = 100 and N=99N = 99.

Answer 99

Solution: IOQM 2025 Part SEP, Q28

Key idea

Pair the terms two at a time: (2j)2−(2j−1)2=4j−1(2j)^2 - (2j-1)^2 = 4j-1, so the whole sum is n(2n+1)n(2n+1).

Group the 2n2n terms in consecutive pairs. For each jj the pair contributes −(2j−1)2+(2j)2=(2j−(2j−1))(2j+(2j−1))=4j−1,-(2j-1)^2 + (2j)^2 = (2j - (2j-1))(2j + (2j-1)) = 4j - 1, so ∑k=12n(−1)kk2=∑j=1n(4j−1)=4⋅n(n+1)2−n=2n2+n=n(2n+1).\sum_{k=1}^{2n} (-1)^k k^2 = \sum_{j=1}^{n} (4j-1) = 4 \cdot \frac{n(n+1)}{2} - n = 2n^2 + n = n(2n+1).

The inequality n(2n+1)<100n(2n+1) < 100 therefore holds for n=6n = 6, where the value is 7878, and fails at n=7n = 7, where it is 105105. Since n(2n+1)n(2n+1) increases with nn, the answer is 66.

Answer 6

Solution: PRMO 2013, Q2

Key idea

Rationalising each term makes Sn=n+1S_n = \sqrt{n+1}, and then 1/(Sn+Sn−1)1/(S_n + S_{n-1}) telescopes a second time.

Rationalise the general term of SnS_n: 1k+1+k=k+1−k(k+1)−k=k+1−k.\frac{1}{\sqrt{k+1} + \sqrt k} = \frac{\sqrt{k+1} - \sqrt k}{(k+1) - k} = \sqrt{k+1} - \sqrt k. Summing from k=0k = 0 to nn, everything cancels except the ends: Sn=n+1−0=n+1.S_n = \sqrt{n+1} - \sqrt0 = \sqrt{n+1}.

Now the outer sum. Since Sn+Sn−1=n+1+nS_n + S_{n-1} = \sqrt{n+1} + \sqrt n, the same rationalisation applies again: 1Sn+Sn−1=n+1−n,\frac{1}{S_n + S_{n-1}} = \sqrt{n+1} - \sqrt{n}, and summing from n=1n = 1 to 9999 telescopes once more: 100−1=10−1=9.\sqrt{100} - \sqrt1 = 10 - 1 = 9.

Two telescopes for the price of one rationalisation. The second is available only because the first collapsed SnS_n into a single square root, which is the point of the construction.

Answer 9

Solution: PRMO 2017, Q11

Key idea

Turning f(x+T)−f(x)f(x+T) - f(x) into products shows that TT must be a period of each wave separately, and then nn has to be a multiple of both 66 and 2020.

Write T=nπT = n\pi and use the sum-to-product identities on the difference: sin⁡x+T3−sin⁡x3=2cos⁡2x+T6 sin⁡T6,\sin\frac{x+T}{3} - \sin\frac{x}{3} = 2\cos\frac{2x+T}{6}\,\sin\frac{T}{6}, cos⁡3(x+T)10−cos⁡3x10=−2sin⁡3(2x+T)20 sin⁡3T20.\cos\frac{3(x+T)}{10} - \cos\frac{3x}{10} = -2\sin\frac{3(2x+T)}{20}\,\sin\frac{3T}{20}. So the condition f(x+T)=f(x)f(x+T) = f(x) for all xx says 2sin⁡T6cos⁡2x+T6=2sin⁡3T20sin⁡3(2x+T)20for all x.2\sin\frac{T}{6}\cos\frac{2x+T}{6} = 2\sin\frac{3T}{20}\sin\frac{3(2x+T)}{20} \qquad \text{for all } x.

Now compare least periods. A sine or cosine wave that is not constant has a least positive period, namely 2π2\pi divided by whatever multiplies xx inside it, and two functions that are equal at every xx are the same function and so have the same least period. That is the whole of the argument.

If the amplitude 2sin⁡T62\sin\tfrac T6 were not zero, the left side would be a wave of least period 6π6\pi; if 2sin⁡3T202\sin\tfrac{3T}{20} were not zero, the right side would be a wave of least period 20π3\tfrac{20\pi}{3}. But 6π≠20π36\pi \ne \tfrac{20\pi}{3}, so they cannot both be non-zero. And if just one were zero, that side would be identically zero and so would the other, forcing its amplitude to zero as well. So both amplitudes vanish: sin⁡nπ6=0andsin⁡3nπ20=0.\sin\frac{n\pi}{6} = 0 \quad\text{and}\quad \sin\frac{3n\pi}{20} = 0.

The first says 6∣n6 \mid n. The second says 20∣3n20 \mid 3n, and since gcd⁡(3,20)=1\gcd(3,20) = 1 that means 20∣n20 \mid n. Hence nn is a multiple of lcm⁡(6,20)=60\operatorname{lcm}(6,20) = 60, and n=60n = 60 works, because 60π60\pi is 1010 periods of the first wave and 99 of the second.

Answer 60

Solution: PRMO 2017, Q14

Key idea

The progression condition is [x]2={x} x[x]^2 = \{x\}\,x, which forces [x]=1[x] = 1 and makes xx the golden ratio.

Write m=[x]m = [x] and f={x}f = \{x\}, so x=m+fx = m + f with 0≤f<10 \le f < 1. For ff, mm, xx to be a geometric progression all three must be non-zero, so m≥1m \ge 1 and f>0f > 0, and the middle term squared is the product of the outer two: m2=f(m+f),that isf2+mf−m2=0.m^2 = f(m+f), \qquad \text{that is} \qquad f^2 + mf - m^2 = 0. Solving for the positive root, f=−m+5m22=m⋅5−12.f = \frac{-m + \sqrt{5m^2}}{2} = m \cdot \frac{\sqrt5 - 1}{2}.

Now use f<1f < 1, and no decimal is needed. If m≥2m \ge 2 then f≥2⋅5−12=5−1>1f \ge 2 \cdot \tfrac{\sqrt5-1}{2} = \sqrt5 - 1 > 1, since 5>2\sqrt 5 > 2. So m=1m = 1. Then f=5−12,x=1+f=1+52=φ,f = \frac{\sqrt5 - 1}{2}, \qquad x = 1 + f = \frac{1 + \sqrt5}{2} = \varphi, the golden ratio, whose defining property φ2=φ+1\varphi^2 = \varphi + 1 is exactly the progression condition in disguise.

Finally we need the least nn with φn>100\varphi^n > 100. Multiply a term Aφ+BA\varphi+B by φ\varphi. Since φ2=φ+1\varphi^2=\varphi+1, the result is (Aφ+B)φ=(A+B)φ+A.(A\varphi+B)\varphi=(A+B)\varphi+A. So the coefficient pair (A,B)(A,B) becomes (A+B,A)(A+B,A) at the next power. Starting from φ2=φ+1\varphi^2=\varphi+1, the powers needed are k2345678910A1235813213455B112358132134\begin{array}{c|rrrrrrrrr} k&2&3&4&5&6&7&8&9&10\\ \hline A&1&2&3&5&8&13&21&34&55\\ B&1&1&2&3&5&8&13&21&34 \end{array} In particular φ9=34φ+21\varphi^9=34\varphi+21 and φ10=55φ+34\varphi^{10}=55\varphi+34. Exact bounds on φ\varphi settle both. Since (115)2=12125<5\left(\tfrac{11}{5}\right)^2 = \tfrac{121}{25} < 5 we have 5>115\sqrt5 > \tfrac{11}{5} and so φ>85\varphi > \tfrac85; and 5<3\sqrt5 < 3 gives φ<2\varphi < 2. Hence φ9=34φ+21<68+21=89<100,\varphi^9 = 34\varphi + 21 < 68 + 21 = 89 < 100, φ10=55φ+34>55⋅85+34=122>100.\varphi^{10} = 55\varphi + 34 > 55 \cdot \tfrac85 + 34 = 122 > 100. So φ9<100<φ10\varphi^9 < 100 < \varphi^{10} and the answer is n=10.n = 10.

Answer 10

Solution: PRMO 2018, Q16

Key idea

Attaching the sign (−1)i+j+1(-1)^{i+j+1} to every pair turns the difference into a single sum; summing over ordered pairs rather than unordered ones then makes it collapse, because ∑(−1)i=0\sum(-1)^i = 0.

On the smaller list 1,2,3,41,2,3,4, the odd pair-sums are 3,5,5,73,5,5,7 and the even ones are 4,64,6. Their difference is 20−10=10=1+2+3+420-10=10=1+2+3+4. The ordinary total appearing here suggests that a signed count may cancel most of the pair contributions. The quantity asked for is T=∑1≤i<j≤10(i+j) (−1)i+j+1,T = \sum_{1 \le i < j \le 10} (i+j)\,(-1)^{i+j+1}, since a pair with i+ji+j odd contributes +(i+j)+(i+j) and one with i+ji+j even contributes −(i+j)-(i+j). Writing ai=(−1)ia_i = (-1)^i, the sign is −aiaj-a_ia_j, so T=−∑i<j(i+j) aiaj.T = -\sum_{i<j}(i+j)\,a_ia_j.

Now evaluate that sum by turning it into a sum over ordered pairs. For each unordered pair the two orderings contribute i aiaji\,a_ia_j and j ajaij\,a_ja_i, whose total is (i+j)aiaj(i+j)a_ia_j. Hence ∑i<j(i+j)aiaj=∑i≠ji aiaj=∑ii ai(S−ai),S=∑j=110aj.\sum_{i<j}(i+j)a_ia_j = \sum_{i \ne j} i\,a_ia_j = \sum_{i} i\,a_i\left(S - a_i\right), \qquad S = \sum_{j=1}^{10} a_j. Two evaluations finish it. First S=−1+1−1+⋯+1=0S = -1+1-1+\cdots+1 = 0, because the ten signs cancel in pairs. Second ai2=1a_i^2 = 1, so the bracket collapses and ∑i<j(i+j)aiaj=0−∑i=110i=−55.\sum_{i<j}(i+j)a_ia_j = 0 - \sum_{i=1}^{10} i = -55.

Therefore T=55T = 55.

The value ∑i ai=−1+2−3+⋯+10=5\sum i\,a_i = -1+2-3+\cdots+10 = 5 never appeared, because it was multiplied by S=0S = 0. That is the whole reason the answer is the plain triangular number 5555.

Answer 55

Solution: IOQM 2020, Q18

Key idea

The expression under the square root is a perfect square in disguise, because 1k−1k+1\tfrac1k - \tfrac1{k+1} happens to equal 1k⋅1k+1\tfrac1k \cdot \tfrac1{k+1}.

Whenever a sum of forty square roots is set as a competition problem, the roots are meant to disappear, so the first question to ask is whether the thing under each root is a square. Write u=1/ku = 1/k and v=1/(k+1)v = 1/(k+1), so that the expression is 1+u2+v21 + u^2 + v^2, and try to match it against (1+u−v)2(1 + u - v)^2. Expanding that guess, (1+u−v)2=1+u2+v2+2u−2v−2uv,(1 + u - v)^2 = 1 + u^2 + v^2 + 2u - 2v - 2uv, so the guess is correct precisely when u−v=uvu - v = uv. That is exactly the identity we have, because 1k−1k+1=1k(k+1)=1k⋅1k+1.\frac1k - \frac1{k+1} = \frac{1}{k(k+1)} = \frac1k \cdot \frac1{k+1}. The coincidence is what makes the problem work, and it is worth noticing that it holds only for consecutive integers.

Since 1+u−v1 + u - v is positive, the square root is exactly 1+1k2+1(k+1)2=1+1k−1k+1,\sqrt{1 + \frac{1}{k^2} + \frac{1}{(k+1)^2}} = 1 + \frac1k - \frac{1}{k+1}, and the sum now separates into forty ones plus a telescoping tail: ∑k=140(1+1k−1k+1)=40+(1−141)=40+4041.\sum_{k=1}^{40} \left(1 + \frac1k - \frac1{k+1}\right) = 40 + \left(1 - \frac{1}{41}\right) = 40 + \frac{40}{41}.

Matching this against a+bca + \tfrac{b}{c} with b<cb < c and gcd⁡(b,c)=1\gcd(b,c) = 1 gives a=40a = 40, b=40b = 40 and c=41c = 41, so a+b=80a + b = 80.

Answer 80

Solution: IOQM 2020, Q20

Key idea

The working times form an arithmetic progression, and the total labour is fixed, so the number of women cancels out of the equation entirely.

Let there be nn women, all working at the same rate. The phrase that the group can build the wall in 4545 hours means that nn women working together for 4545 hours complete the job, so the wall represents exactly 45n45n woman-hours of labour, whatever nn happens to be.

Now account for how those hours were actually supplied. The women arrive one at a time at equal intervals, and each stays until the end, so the first woman works the longest and each subsequent one works a fixed amount less. Their working times therefore form an arithmetic progression, with first term the time TT worked by the first woman and last term the time worked by the last. We are told the first worked five times as long as the last, so the last term is T/5T/5.

The total labour supplied is the sum of that progression, which for nn terms is the number of terms times the average of the first and last: n⋅T+T/52=45n.n \cdot \frac{T + T/5}{2} = 45n. The factor nn appears on both sides and cancels, which is the reason the problem never tells us how many women there were. What remains is 6T10=45,soT=75.\frac{6T}{10} = 45, \qquad \text{so} \qquad T = 75.

The first woman worked for 7575 hours.

Answer 75

Solution: IOQM 2023, Q10

Key idea

The combination an2+4anan−1+7an−12a_n^2 + 4a_na_{n-1} + 7a_{n-1}^2 is multiplied by exactly 77 at each step of the recursion, so the quantity asked for is a pure power of 77.

The first few terms are a0=1a_0=1, a1=−4a_1=-4 and a2=9a_2=9. The combination in the question begins a12−a0a2=16−9=7.a_1^2-a_0a_2=16-9=7. The next term is a3=−4⋅9−7(−4)=−8a_3=-4\cdot9-7(-4)=-8, giving a22−a1a3=81−32=49.a_2^2-a_1a_3=81-32=49. The individual terms are irregular, but this combination may have a much simpler rule. Substituting the recursion an+1=−4an−7an−1a_{n+1} = -4a_n - 7a_{n-1} into the expression we are asked about gives an2−an−1an+1=an2−an−1(−4an−7an−1)=an2+4anan−1+7an−12.a_n^2 - a_{n-1}a_{n+1} = a_n^2 - a_{n-1}(-4a_n - 7a_{n-1}) = a_n^2 + 4a_na_{n-1} + 7a_{n-1}^2. Call the right-hand side QnQ_n. It is worth tracking QnQ_n rather than the original expression, because it behaves so tidily. Using the recursion once more, Qn+1=an+12+4an+1an+7an2=an+1(an+1+4an)+7an2,Q_{n+1} = a_{n+1}^2 + 4a_{n+1}a_n + 7a_n^2 = a_{n+1}\bigl(a_{n+1} + 4a_n\bigr) + 7a_n^2, which is an+1(−7an−1)+7an2=7 Qna_{n+1}\bigl(-7a_{n-1}\bigr) + 7a_n^2 = 7\,Q_n, where the middle step used an+1+4an=−7an−1a_{n+1} + 4a_n = -7a_{n-1}.

So each step multiplies QQ by 77, and the starting value is Q1=a12+4a1a0+7a02=16−16+7=7.Q_1 = a_1^2 + 4a_1a_0 + 7a_0^2 = 16 - 16 + 7 = 7. Hence Qn=7nQ_n = 7^n for every n≥1n \ge 1, and in particular a502−a49a51=Q50=750.a_{50}^2 - a_{49}a_{51} = Q_{50} = 7^{50}.

The divisors of 7507^{50} are 70,71,…,7507^0, 7^1, \ldots, 7^{50}, so there are 5151 of them.

Answer 51

Solution: IOQM 2026, Q16

Key idea

Clearing the fractions, almost everything cancels, and what is left says that the product of two terms two places apart never changes. That makes the sequence repeat every four terms.

Cross-multiplying, (an+3−an+2)(an+an+1)=(an+3+an+2)(an−an+1).(a_{n+3} - a_{n+2})(a_n + a_{n+1}) = (a_{n+3} + a_{n+2})(a_n - a_{n+1}). Both sides contain an+3ana_{n+3}a_n and −an+2an+1-a_{n+2}a_{n+1}, which cancel. What remains is an+3an+1−an+2an=−an+3an+1+an+2ana_{n+3}a_{n+1} - a_{n+2}a_n = -a_{n+3}a_{n+1} + a_{n+2}a_n, that is an+1an+3=anan+2.a_{n+1}a_{n+3} = a_n a_{n+2}. So the product of two terms two places apart is the same number kk all the way along the sequence.

This kk is not zero. If it were, then a55a57=0a_{55}a_{57} = 0 with a55=6a_{55} = 6 would give a57=0a_{57} = 0, and a56a58=0a_{56}a_{58} = 0 would make a56a_{56} or a58a_{58} zero as well, next to a57a_{57}. Two neighbouring terms equal to 00 make a denominator an−an+1a_n - a_{n+1} in the statement vanish. So no term is zero, and an+4=kan+2=an,a_{n+4} = \frac{k}{a_{n+2}} = a_n, which means the sequence repeats with period 44.

Now 55=4⋅13+355 = 4 \cdot 13 + 3, 66=4⋅16+266 = 4 \cdot 16 + 2 and 77=4⋅19+177 = 4 \cdot 19 + 1, so a3=6a_3 = 6, a2=2a_2 = 2 and a1=1a_1 = 1. Then k=a1a3=6k = a_1a_3 = 6 and a4=k/a2=3a_4 = k/a_2 = 3. The sequence runs 1,2,6,3,1,2,6,3,…1, 2, 6, 3, 1, 2, 6, 3, \ldots, and no two neighbours are equal or opposite, so every fraction in the statement is defined.

Since 2026=4⋅506+22026 = 4 \cdot 506 + 2, N=506 (1+4+36+9)+1+4=25300+5=25305,N = 506\,(1 + 4 + 36 + 9) + 1 + 4 = 25300 + 5 = 25305, and the sum of its digits is 2+5+3+0+5=152 + 5 + 3 + 0 + 5 = 15.

Answer 15

Report an error on this page

Reports are stored by Netlify. See the privacy note.