The Canadian Mathematical Society

in collaboration with

The CENTRE for EDUCATION in MATHEMATICS and COMPUTING

Sun Life Financial Canadian Open Mathematics Challenge

Solutions

2010 Canadian Mathematical Society

2010 COMC Solutions Part A 1. Solution 1 142 42 196 16 180 (9 + 5)2 (9 5)2 = = = = 4. Calculating, (9)(5) 45 45 45 Solution 2 For a general x and y with x and y not zero, (x + y )2 (x y )2 (x2 + 2xy + y 2 ) (x2 2xy + y 2 ) 4xy = = =4 xy xy xy Since this expression equals 4 for any values of x and y , then (9 + 5)2 (9 5)2 = 4. (9)(5)

Page 2

Solution 3 For a general x and y with x and y not zero, we can factor as a dierence of squares: [(x + y ) + (x y )][(x + y ) (x y )] (2x)(2y ) (x + y )2 (x y )2 = = =4 xy xy xy Since this expression equals 4 for any values of x and y , then (9 + 5)2 (9 5)2 = 4. (9)(5) Answer: 4 2. Simplifying both sides, x (8 x) = 8 (x 8) x8+x = 8x+8 3x = 24 x = 8 Therefore, x = 8. Answer: x = 8 3. Solution 1 We call the ring between the middle and inner circles the inner ring. We reect the shaded portion of the inner ring across line segment CD. The area of the shaded region does not change when we do this. The shaded region is now the entire semi-circle to the right of CD. Thus, the area of the shaded region is half of the area of the outer circle. Since OC = 6, then the outer circle has radius 6 and so has area 62 = 36 . 1 Therefore, the area of the shaded region is 2 (36 ) = 18 .

2010 COMC Solutions

Page 3

Solution 2 We call the ring between the outer and middle circles the outer ring, and the ring between the middle and inner circles the inner ring. Since OC = 6, then the outer circle has radius 6 and so has area 62 = 36 . Since OB = 4, then the middle circle has radius 4 and so has area 42 = 16 . Since OA = 2, then the inner circle has radius 2 and so has area 22 = 4 . Since the outer circle has area 36 and the middle circle has area 16 , then the area of the outer ring is 36 16 = 20 . Since the diameter CD divides each ring into two parts of equal area, then the shaded region 1 (20 ) = 10 . of the outer ring has area 2 Since the middle circle has area 16 and the inner circle has area 4 , then the area of the inner ring is 16 4 = 12 . Since the diameter CD divides each ring into two parts of equal area, then the shaded region (12 ) = 6 . of the inner ring has area 1 2 Since the inner circle has area 4 and line segment CD passes through the centre of this circle, then the shaded region of the inner circle has area 1 (4 ) = 2 . 2 Therefore, the total shaded area is 10 + 6 + 2 = 18 . Answer: 18 4. First, we simplify the given expression: (3.1 107 )(8 108 ) 3.1 8 107 108 = = 3.1 4 107+83 = 12.4 1012 = 124 1011 3 3 2 10 2 10 Therefore, this integer consists of the digits 124 followed by 11 zeroes, so has 14 digits. Answer: 14 5. Solution 1 Let Q be the point on the line y = x that is closest to P (3, 9). Then P Q is perpendicular to the line y = x. Since the line with equation y = x has slope 1 and P Q is perpendicular to this line, then P Q has slope 1. Note that a general point Q on the line with equation y = x has coordinates (t, t) for some real number t. t9 For the slope of P Q to equal 1, we must have = 1 or t 9 = (t + 3) or 2t = 6 t (3) or t = 3. Therefore, the point on the line with equation y = x that is closest to P is the point (3, 3).

2010 COMC Solutions

Page 4

Solution 2 Let Q be the point on the line y = x that is closest to P (3, 9). Then P Q is perpendicular to the line y = x. Let R be the point on the line y = x so that P R is horizontal, as shown.

y
P Q R

Since P R is horizontal and P has y -coordinate 9, then R has y -coordinate 9. Since R lies on the line with equation y = x, then R has coordinates (9, 9). Since P R is horizontal and QR has slope 1 (because it lies along the line with equation y = x), then P RQ = 45 . Since P QR = 90 , then P QR is isosceles and right-angled. Let M be the midpoint of P R. Since P has coordinates (3, 9) and R has coordinates (9, 9), then M has coordinates (3, 9). Since P QR is isosceles and M is the midpoint of P R, then QM is perpendicular to P R. Thus, QM is vertical, so Q has x-coordinate 3. Since Q lies on the line with equation y = x, then Q has coordinates (3, 3). Solution 3 Note that a general point Q on the line with equation y = x has coordinates (t, t) for some real number t. Then P Q = (t (3))2 + (t 9)2 or P Q2 = (t + 3)2 + (t 9)2 . Since we want to nd the point on the line with equation y = x that is closest to P , then we want to minimize the value of P Q, or equivalently to minimize the value of P Q2 . In other words, we want to nd the value of t that minimizes the value of P Q2 = t2 + 6t + 9 + t2 18t + 81 = 2t2 12t + 90 Since this equation represents a parabola opening upwards, then its minimum occurs at its 12 vertex, which occurs at t = 2(2) = 3. Thus, t = 3 minimizes the length of P Q. Therefore, the point on the line with equation y = x that is closest to P is the point (3, 3). Answer: (3, 3)

2010 COMC Solutions

Page 5

6. Let x be the number of people who studied for the exam and let y be the number of people who did not study. We assume without loss of generality that the exam was out of 100 marks. Since the average of those who studied was 90%, then those who studied obtained a total of 90x marks. Since the average of those who did not study was 40%, then those who did not study obtained a total of 40y marks. 90x + 40y = 85. Since the overall average was 85%, then x+y Therefore, 90x + 40y = 85x + 85y or 5x = 45y or x = 9y . Therefore, x : y = 9 : 1 = 90 : 10. This means that 10% of the class did not study for the exam. Answer: 10% 7. Solution 1 Since ABCD is a rectangle, then AD = BC = 10 and DC = AB = 20. Since W A = 12, W B = 16, AB = 20, and 122 + 162 = 144 + 256 = 400 = 202 , then W A2 + W B 2 = AB 2 . Thus, AW B is right-angled at W . Note that CKD is congruent to AW B , so CKD is right-angled at K . Extend W A and KD to meet at Y and W B and KC to meet at Z .

W 12 A 20 10 Y D K
Suppose that W AB = KCD = . Then W BA = KDC = 90 . Now Y AD = 180 W AB BAD = 180 90 = 90 . Also, Y DA = 180 KDC ADC = 180 (90 ) 90 = . Therefore, Y DA is similar to W AB . This means that DY A = 90 . 1 Also, since DA = 2 AB , then the sides of Y DA are half as long as the corresponding sides of W AB . Thus, Y D = 1 W A = 6 and Y A = 1 W B = 8. 2 2 Similarly, BZC = 90 . Therefore, W Y KZ is a rectangle. We have W Y = W A + AY = 12 + 8 = 20 and Y K = Y D + DK = 6 + 16 = 22.

16 B Z

2010 COMC Solutions

Page 6

By the Pythagorean Theorem, since W K > 0, then W K = W Y 2 + Y K 2 = 202 + 222 = 400 + 484 = 884 = 2 221 as required. Solution 2 Since ABCD is a rectangle, then AD = BC = 10 and DC = AB = 20. We coordinatize the diagram, putting D at the origin, A at (0, 10), C at (20, 0), and B at (20, 10). Since W A = 12, W B = 16, AB = 20, and 122 + 162 = 144 + 256 = 400 = 202 , then W A2 + W B 2 = AB 2 . Thus, AW B is right-angled at W . Note that CKD is congruent to AW B , so CKD is right-angled at K . Since AW B is right-angled and we know its side lengths, then we can compute the trigonometric ratios of its angles. 16 4 WA 12 3 WB = = and cos(W AB ) = = = . In particular, sin(W AB ) = AB 20 5 AB 20 5 We drop perpendiculars from W to X on AB and from K to Y on DC .

y
W 12 A (0, 10) X 20 Y D K C (20, 0) 16 B (20, 10) 10

Then AX = W A cos(W AB ) = 12( 3 ) = 36 and W X = W A sin(W AB ) = 12( 4 ) = 48 . 5 5 5 5 Since A has coordinates (0, 10), then W has coordinates ( 36 , 10 + 48 ) = ( 36 , 98 ). 5 5 5 5 Since CKD is congruent to AW B , then in a similar way we can nd that the coordinates , 48 ) = ( 64 , 48 ). of K are (20 36 5 5 5 5 Since we have the coordinates of W and K , then the distance between W and K is WK = ( 64 36 )2 + ( 48 98 )2 = 5 5 5 5 2 = 2 142 + 732 = 5 196 + 5329 5 = 2 5525 = 2 5525 5 25 = 2 221
282 52

1462 52

2010 COMC Solutions Therefore, W K = 2 221.

Page 7

Answer: W K = 2 221 8. Solution 1 First, we factor the rst and third quadratic factors to obtain (x + 1)(x + 2)(x2 2x 1)(x 3)(x 4) + 24 = 0 Next, we rearrange the factors to obtain (x + 1)(x 3)(x2 2x 1)(x + 2)(x 4) + 24 = 0 and expand to obtain (x2 2x 3)(x2 2x 1)(x2 2x 8) + 24 = 0 Next, we make the substitution w = x2 2x to obtain (w 3)(w 1)(w 8) + 24 = 0 This is a cubic equation in w so we expand, simplify and factor: (w2 4w + 3)(w 8) + 24 = 0 w3 12w2 + 35w = 0 w(w2 12w + 35) = 0 w(w 5)(w 7) = 0 Therefore, the solutions in terms of w are w = 0 or w = 5 or w = 7. If w = x2 2x = 0, then x(x 2) = 0 which gives x = 0 or x = 2. If w = x2 2x = 5, then x2 2x 5 = 0. 2 24 = 1 6. The quadratic formula gives the roots x = 2 If w = x2 2x = 7, then x2 2x 7 = 0. 2 32 The quadratic formula gives the roots x = = 1 8. 2 Therefore, x = 0 or x = 2 or x = 1 6 or x = 1 8. (This last pair can be rewritten as x = 1 2 2.) Solution 2 First, we factor the rst and third quadratic factors and complete the square in the second quadratic factor to obtain (x + 1)(x + 2)((x 1)2 2)(x 3)(x 4) + 24 = 0

2010 COMC Solutions

Page 8

Next, we make the substitution y = x 1 (which makes x + 1 = y + 2 and x + 2 = y + 3 and x 3 = y 2 and x 4 = y 3) to obtain (y + 2)(y + 3)(y 2 2)(y 2)(y 3) + 24 = 0 (We made this substitution because it made the algebra more symmetric; that is, after making this substitution, factors of the form y a are paired with factors of the form y + a.) Next, we rearrange the factors to obtain (y + 2)(y 2)(y + 3)(y 3)(y 2 2) + 24 = 0 Next, we multiply out pairs of factors to obtain (y 2 4)(y 2 9)(y 2 2) + 24 = 0 Next, we make the substitution z = y 2 to obtain (z 4)(z 9)(z 2) + 24 = 0 This is a cubic equation in z so we expand and simplify: (z 2 13z + 36)(z 2) + 24 = 0 z 3 15z 2 + 62z 48 = 0 By inspection, z = 1 is a solution so z 1 is a factor of the cubic equation. We factor out this linear factor to obtain (z 1)(z 2 14z + 48) = 0. The quadratic factor can be factored as (z 6)(z 8). Therefore, we have (z 1)(z 6)(z 8) = 0. Therefore, the solutions in terms of z are z = 1 or z = 6 or z = 8. Since z = y 2 , then the solutions in terms of y are y = 1 or y = 6 or y = 8. Since y = x 1, then x = y + 1, and so the solutions in terms of x are x = 0 or x = 2 or x = 1 6 or x = 1 8. (This last pair can be rewritten as x = 1 2 2.) Answer: x = 0, 2, 1 6, 1 2 2

2010 COMC Solutions

Page 9

Part B 1. (a) Solution 1 From the rst row, A + A = 50 or A = 25. From the second column, A + C = 57. Since A = 25, then C = 57 25 = 32. Solution 2 From the rst row, A + A = 50 or A = 25. From the rst column, A + B = 37. Since A = 25, then B = 37 25 = 12. From the second row, B + C = 44. Since B = 12, then C = 44 12 = 32.

(b) Solution 1 The sum of the nine entries in the table equals the sum of the column sums, or 50+ n +40 = 90 + n. (This is because each entry in the table is part of exactly one column sum.) Similarly, the sum of the nine entries in the table also equals the sum of the row sums, or 30 + 55 + 50 = 135. Therefore, 90 + n = 135 or n = 45. Solution 2 The sum of the nine entries in the table equals the sum of the row sums, or 30 + 55 + 50 = 135. (This is because each entry in the table is part of exactly one row sum.) Since the entries in the table include three entries equal to each of D, E and F , then the sum of the entries in the table is also 3D + 3E + 3F = 3(D + E + F ). Therefore, 3(D + E + F ) = 135 or D + E + F = 45. From the second column, D + E + F = n. Thus, n = 45. Solution 3 From the rst row, D + D + D = 30 or D = 10. From the rst column, D + 2F = 50. Since D = 10, then 2F = 50 10 and so F = 20. From the third column, D + 2E = 40. Since D = 10, then 2E = 40 10 and so E = 15. Therefore, n = D + E + F = 10 + 15 + 20 = 45.

(c) Solution 1 From the third row, 3R + T = 33. From the fourth row, R + 3T = 19. Adding these equations, we obtain 4R + 4T = 52 or R + T = 13. From the rst row, P + Q + R + T = 20. Since R + T = 13, then P + Q = 20 13 = 7.

2010 COMC Solutions

Page 10

Solution 2 The sum of the sixteen entries in the table equals the sum of the row sums, or 20 + 20 + 33 + 19 = 92. (This is because each entry in the table is part of exactly one row sum.) The table includes two entries equal to each of P and Q and six entries equal to each of R and T . Therefore, 2P + 2Q + 6R + 6T = 92. The last two rows of the table include four entries equal to each of R and T , so 4R + 4T = 33 + 19 = 52, or R + T = 13. Therefore, 2P + 2Q = 92 6(R + T ) = 92 6(13) = 14, and so P + Q = 7. Solution 3 From the third row, 3R + T = 33. From the fourth row, R + 3T = 19. Multiplying the rst equation by 3 and subtracting the second equation gives (9R + 3T ) (R + 3T ) = 99 19 or 8R = 80 or R = 10. Since 3R + T = 33, then T = 33 3(10) = 3. From the rst row, P + Q + R + T = 20. Since R = 10 and T = 3, then P + Q = 20 10 3 = 7.

2. (a) To determine the coordinates of A and B , we equate values of y using the equations y = x2 4x + 12 and y = 2x + 20 to obtain x2 4x + 12 = 2x + 20 x2 2x 8 = 0 (x 4)(x + 2) = 0 Therefore, x = 4 or x = 2. To determine the y -coordinates of points A and B , we can use the equation of the line. If x = 4, then y = 2(4) + 20 = 12. If x = 2, then y = 2(2) + 20 = 24. Therefore, the coordinates of A and B are (4, 12) and (2, 24). (b) Using the coordinates of A and B from (a), the coordinates of the midpoint M of AB are 1 (4 + (2)), 1 (24 + 12)) or (1, 18). (2 2 (c) Solution 1 The line with equation y = 2x + 20 has slope 2. Therefore, we have a line with slope 2 that intersects the parabola at points

2010 COMC Solutions P (p, p2 4p + 12) and Q(q, q 2 4q + 12). In other words, line segment P Q has slope 2. Therefore, (p2 4p + 12) (q 2 4q + 12) pq p2 q 2 4p + 4q pq (p q )(p + q ) 4(p q ) pq (p + q ) 4 = 2 = 2 = 2 = 2 (since p = q )

Page 11

p+q = 2 Therefore, p + q = 2, as required. Solution 2 The line with equation y = 2x + 20 has slope 2. Therefore, we have a line with slope 2 (say with equation y = 2x + b) that intersects the parabola at points P and Q. Since y = 2x + b and y = x2 4x + 12 intersect when x = p, then p2 4p + 12 = 2p + b, which gives p2 2p + 12 b = 0. Since y = 2x + b and y = x2 4x + 12 intersect when x = q , then q 2 4q + 12 = 2q + b, which gives q 2 2q + 12 b = 0. Since we have two expressions equal to 0, then p2 2p + 12 b = q 2 2q + 12 b p2 2p = q 2 2q p2 q 2 2p + 2q = 0 (p q )(p + q ) 2(p q ) = 0 (p q )(p + q 2) = 0 Therefore, p q = 0 or p + q 2 = 0. Since p = q , then p + q = 2. (d) Since P has coordinates (p, p2 4p + 12) and Q has coordinates (q, q 2 4q + 12), then the 1 x-coordinate of the midpoint N of P Q is 2 (p + q ). Since p + q = 2 by (c), then the x-coordinate of N is 1. Since the x-coordinate of M is 1 and the x-coordinate of N is 1, then line segment M N is vertical.

2010 COMC Solutions

Page 12

3. (a) Solution 1 Let U be the point in S vertically above O and let B be the point where AU intersects the circle. (There will be one other point U in S with U O perpendicular to AC ; this point will be vertically below O. By symmetry, the length of U O is the same in either case.) Join U O and BO.
U

Let BU O = . Note that AO = BO = 1 since they are radii and BU = 1 by denition. Therefore, U BO is isosceles and so BOU = BU O = . Now ABO is an exterior angle in this triangle, so ABO = BU O + BOU = 2. Since OB = OA, then ABO is isosceles and so BAO = ABO = 2. But U AO is right-angled at O, and so U AO + AU O = 90 or 2 + = 90 . Therefore, 3 = 90 or = 30 . This tells us that U AO is a 30 -60 -90 triangle, and so U O = 3AO = 3. Solution 2 Let U be the point in S vertically above O and let B be the point where AU intersects the circle. (There will be one other point U in S with U O perpendicular to AC ; this point will be vertically below O. By symmetry, the length of U O is the same in either case.) Join U O and BC .
U

2010 COMC Solutions

Page 13

Since AC is a diameter, then ABC = 90 . Therefore, ABC is similar to AOU since each is right-angled and each includes the angle at A. AB AO Thus, = . AC AU Since U is in S , then BU = 1, so AU = AB + BU = AB + 1. Also, AO = 1 and AC = 2 since the radius of the circle is 1. 1 AB = or AB 2 + AB 2 = 0. Therefore, 2 AB + 1 Factoring, we obtain (AB 1)(AB + 2) = 0. Since AB > 0, then AB = 1 and so AU = 2. By the Pythagorean Theorem, since U O > 0, then U O = AU 2 AO2 = 22 12 = 3. (b) As in (a), we can choose the point V in S that is vertically above C . B is the point where AV intersects the circle. Note that BV = 1 by denition. Join V C and BC . Since AC is a diameter, then ABC = 90 .
V B

Let V C = x. Since V CA is right-angled at C and AV > 0, then by the Pythagorean Theorem, AV = AC 2 + CV 2 = 4 + x2 . Now V BC is similar to V CA since both are right-angled and the triangles share a common angle at V . Since these triangles are similar, then VB VC = VC VA 1 x = x x2 + 4 x2 + 4 = x2 x2 + 4 = x4 0 = x4 x2 4 0 = (x2 )2 x2 4

2010 COMC Solutions

Page 14

This is a quadratic equation in x2 . By the quadratic formula, 1 (1)2 4(1)(4) 1 17 2 = x = 2 2 1 + 17 2 2 Since x is positive, then x = . 2 1 + 17 . Since V C = x is positive, then V C = x = 2 (c) We prove that such a circle does not exist by contradiction. Suppose that there is a circle Z on which all of the points in S lie. We coordinatize the original diagram, putting O at (0, 0), C at (1, 0), and A at (1, 0). For every point X in S that is above AC , there will be a corresponding point Y in S that is below AC which is the reection of X in AC . Therefore, S is symmetric across the x-axis. Thus, Z is also symmetric across the x-axis and so its centre lies on the x-axis. Suppose Z has centre (p, 0) and radius r. Then the equation of Z is (x p)2 + y 2 = r2 . From (a), the point (0, 3) lies on Z . Thus, p2 + 3 = r2 . Also, the point W (2, 0) lies on Z . This point comes from choosing B to coincide with C and extending AB horizontally by 1 unit. Thus, (2 p)2 + 02 = r2 or p2 4p + 4 = r2 . From the equations p2 + 3 = r2 and p2 4p + 4 = r2 , we equate values of r2 to obtain 1 p2 + 3 = p2 4p + 4 or 4p = 1 or p = 4 . 1 Thus, r2 = p2 + 3 = 16 + 3 = 49 and so r = 7 since r > 0. 16 4 7 2 1 2 ). Therefore, the equation of Z must be (x 4 ) + y 2 = ( 4 From (b), the point Therefore, (1
1 2 ) 4 2 1+ 17 + 2 3 2 (4 ) + 1+2 17 17 9 8 + + 16 16 2 17 2

1,

1+ 17 2

lies on the circle.

= (7 )2 4 = =
49 16 49 16

= 2 17 = 4

This statement is false, so we have reached a contradiction. Therefore, our assumption is false and there is no circle on which all of the points in S lie. 1 1 1 4. (a) First we note that if x > 0, then x + > 1, since if x 1, then x + 1 + > 1 and if x x x 1 1 0 < x < 1, then > 1, so x + > 1. x x

2010 COMC Solutions

Page 15

We note that f (x) = x is equivalent to x+ 1 x x+ 1 x = x 1 = x x+ 1 x

In this last equation, the right side is a positive integer, so the left side is also a positive integer. 1 1 Suppose that = n for some positive integer n. Then x = . x n 1 1 1 is equivalent to the equation n = +n . Therefore, the equation = x + x x n 1 1 1 Note that if n 2, then < 1, so n < n + < n + 1, which says that + n = n so n n n this result is true for all positive integers n 2. 1 1 Note also that if n = 1, then n + = 2, so + n = n. n n 1 + n if and only if n 2. Therefore, if n is a positive integer, then n = n 1 Therefore, the solution set of the equation f (x) = x is x = , where n is a positive integer n with n 2. a (b) Suppose that x = for some positive integer a > 1. a+1 First, we calculate f (x). Note that x+ a a+1 a2 + (a + 1)2 2a2 + 2a + 1 2(a2 + a) + 1 1 1 = + = = = =2+ 2 2 2 x a+1 a a(a + 1) a +a a +a a +a a2 1 1 1 1 < 2 = and so 2 < 2 + 2 < 3. +a 1 +1 2 a +a 1 = 2+ 2 = 2. a +a 1 1 1 x+ = 2+ 2 then f (x) = x + x x a +a

Since a > 1, then Therefore, x + Thus, if x = 1 x

a , a+1

2=

a2

1 . +a

Second, we show that x = f (x). a a2 1 Note that x = = and f (x) = , so x and f (x) would be equal if a+1 a(a + 1) a(a + 1) a2 1 and only if = which is true if and only if a2 = 1. a(a + 1) a(a + 1) Since a > 1, this is not true, so x = f (x). 1 (Alternatively, we could note that, from (a), x = f (x) if and only if x is of the form for n

2010 COMC Solutions

Page 16 a which is not of this form when a > 1, so a+1

some positive integer n > 1. Here, x = x = f (x).)

Third, we show that f (x) = f (f (x)). 1 We set y = f (x) = 2 . a +a 1 Since a is a positive integer with a > 1, then y is of the form for some positive integer n n with n > 2. Thus, y is of the form discovered in (a), so f (y ) = y ; in other words, f (f (x)) = f (x). Therefore, if x = a for some positive integer a > 1, then x = f (x), but f (x) = f (f (x)). a+1

(c) Solution 1 We want to nd an innite family of rational numbers u with the properties that 0 < u < 1, u, f (u), and f (f (u)) are all distinct, and f (f (u)) = f (f (f (u))). We will do this by nding an innite family of rational numbers u with 0 < u < 1 with a for some positive integer a > 1. the property that f (u) = a+1 1 1 In this case, (b) shows that f (f (u)) = 2 and that f (f (f (u))) = 2 . a +a a +a Thus, we will have f (f (u)) = f (f (f (u))) and f (u) = f (f (u)). 1 a and u = 2 , then we will have found a family of rational As long as we have u = a+1 a +a numbers u with the required properties. 1 Note that in fact we cannot have u = 2 because in this case we would have f (u) = u a +a a and so we would not have f (u) = . a+1 We now show the existence of an innite family of rational numbers u with 0 < u < 1 a with f (u) = for some positive integer a > 1. a+1 b Let us consider candidate rational numbers u = with b and c positive integers and b+c c > 1. Since b + c > b, then each is a rational number with 0 < u < 1. b b+c b2 + ( b + c ) 2 2b2 + 2bc + c2 c2 1 In this case, u + = + = = = 2 + . u b+c b b(b + c) b2 + bc b2 + bc c2 1 c2 If we suppose further that c2 < b2 + bc, then 2 < 1 and so u + = 2 + 2 < 3, b + bc u b + bc which gives 1 1 c2 c2 f (u) = u + u+ =2+ 2 2= 2 u u b + bc b + bc

2010 COMC Solutions

Page 17

a c2 We want f (u) to be of the form . In other words, we want 2 to be of the form a+1 b + bc a , which would be true if b2 + bc c2 = 1. a+1 Note that if b2 + bc c2 = 1, then c2 = b2 + bc 1 < b2 + bc, so the additional assumption above is included in this equation. Also, if b2 + bc c2 = 1, then b and c can have no b common divisor larger than 1 so u = is irreducible. Combining this with the fact b+c b a that c = 1, we see that cannot be of the form . b+c a+1 To summarize so far, if b2 + bc c2 = 1 has an innite family of positive integer solutions b has the required properties. (b, c), then the innite family of rational numbers u = b+c Consider the equation b2 + bc c2 = 1. This is equivalent to the equations 4b2 + 4bc 4c2 = 4 and 4b2 + 4bc + c2 5c2 = 4 and (2b + c)2 5c2 = 4. If we let d = 2b + c, we obtain the equation d2 5c2 = 4. This is a version of Pells equation. It is known that if such an equation has one positive integer solution, then it has innitely many positive integer solutions. Since d2 5c2 = 4 has one positive integer solution (d, c) = (7, 3), then it has innitely many positive integer solutions (d, c). If d2 = 5c2 + 4 and c is odd, then c2 is odd, so d2 = 5c2 + 4 is odd, which means that d is odd. If d2 = 5c2 + 4 and c is even, then c2 is even, so d2 = 5c2 + 4 is even, which means that d is even. 1 ( d c) Therefore, if (d, c) satises d2 5c2 = 4, then d and c have the same parity so b = 2 is an integer. In addition, since d2 = 5c2 + 4 > c2 then d > c which means that b = 1 (d c) is a positive 2 integer. Therefore, each positive integer solution (d, c) of the equation d2 5c2 = 4 gives a solution (b, c) of the equation b2 + bc c2 = 1 which is also a positive integer solution. b with the required Therefore, there exists an innite family of rational numbers u = b+c properties. Solution 2 As in Solution 1, we want to show the existence of an innite family of rational numbers a u with 0 < u < 1 with f (u) = for some positive integer a > 1. a+1 Consider the Fibonacci sequence which has F1 = 1, F2 = 1, and Fn = Fn1 + Fn2 for n 3.

2010 COMC Solutions

Page 18

F2n1 for each positive integer n 2. F2n+1 Note that 0 < F2n1 < F2n+1 so 0 < u < 1. 2 F3 For example, u2 = = . F5 5 2 29 9 In this case, f (u2 ) = ( 5 + 5 ) 2 +5 = 10 29 = 10 , which has the desired properties. 2 5 2 10 a 1 We must show that un is not of the form or of the form 2 : a+1 a +a F2n1 a If = , then aF2n1 + F2n1 = aF2n+1 or F2n1 = a(F2n+1 F2n1 ) or F2n+1 a+1 F2n1 = aF2n . Since a is a positive integer and F2n > F2n1 , this cannot be the case. 1 F2n1 = 2 , then F2n+1 is divisible by F2n1 . But Fj +1 and Fj 1 never If F2n+1 a +a have a common divisor larger than 1, so this cannot be the case. (If Fj +1 and Fj 1 have a common divisor larger than 1, then Fj = Fj +1 Fj 1 also has this divisor. We can continue this process using the equation Fj 2 = Fj Fj 1 to show that Fj 2 also has this divisor, and so on, until we obtain that F2 and F1 both have this divisor. Since F2 = F1 = 1, we have a contradiction.) Dene un = In general, note that un + F2n1 F2n+1 1 = + un F2n+1 F2n1 (F2n1 )2 + (F2n+1 )2 = F2n1 F2n+1 (F2n1 )2 + (F2n + F2n1 )2 = F2n1 (F2n + F2n1 ) 2(F2n1 )2 + 2F2n F2n1 + (F2n )2 = (F2n1 )2 + F2n F2n1 (F2n )2 = 2+ (F2n1 )2 + F2n F2n1 (F2n )2 = 2+ F2n1 F2n+1

It is known that (F2n )2 F2n1 F2n+1 = 1 for all positive integers n. (See the end of the solution for a proof of this.) Set an = (F2n )2 , which is a positive integer. 1 an Then, un + =2+ . un an + 1

2010 COMC Solutions

Page 19

Therefore, f (un ) = 1 1 un + un un an an = 2+ 2+ an + 1 an + 1 an 2 = 2+ an + 1 an = an + 1 un +

Therefore, the innite family of rational numbers un has the desired properties. As a postscript, we prove that (Fm )2 Fm1 Fm+1 = (1)m+1 for all positive integers m 2. We prove this result by induction on m. When m = 2, we obtain (F2 )2 F1 F3 = 12 1(2) = 1 = (1)2+1 , as required. Suppose that the result is true for m = k , for some positive integer k 2. That is, suppose that (Fk )2 Fk1 Fk+1 = (1)k+1 . Consider m = k + 1. Then (Fk+1 )2 Fk Fk+2 = (Fk + Fk1 )2 Fk (Fk + Fk+1 ) = (Fk )2 + 2Fk Fk1 + (Fk1 )2 (Fk )2 Fk Fk+1 = 2Fk Fk1 + (Fk1 )2 Fk Fk+1 = 2Fk Fk1 + (Fk1 )2 Fk (Fk + Fk1 ) = 2Fk Fk1 + (Fk1 )2 (Fk )2 Fk Fk1 = Fk Fk1 + (Fk1 )2 (Fk )2 = Fk1 (Fk + Fk1 ) (Fk )2 = Fk1 Fk+1 (Fk )2 = (1)((Fk )2 Fk1 Fk+1 ) = (1)(1)k+1 = (1)(k+1)+1 as required. Therefore, (Fm )2 Fm1 Fm+1 = (1)m+1 for all positive integers m 2 by induction, which shows that (F2k )2 F2k1 F2k+1 = (1)2k+1 = 1. (by our inductive assumption)