Let a∈R be a parameter. (1) Prove that the curves of y=x2+(a+2)x−2a+1 pass through a fixed point; also, the vertices of these parabolas all lie on the curve of a certain parabola. (2) If the function x2+(a+2)x−2a+1=0 has two distinct real roots, find the value range of the larger root.
2005 South East Mathematical Olympiad
July 10th - Day 1
Click for solution (1) It is clear that all parabolas pass through (2,9), and the vertices lie on the parabola y=−x2+4x+5. (2) The larger root is −a−2+√a2+12a2, so a∈(−∞,−12) or (0,∞). Now, as a gets smaller, the root gets bigger. On the other hand, the root is negative in (0,12), and the function is concave, so the lower bound is −1.
Circle C (with center O) does not have common point with line l. Draw OP perpendicular to l, P∈l. Let Q be a point on l (Q is different from P), QA and QB are tangent to circle C, and intersect the circle at A and B respectively. AB intersects OP at K. PM, PN are perpendicular to QB, QA, respectively, M∈QB, N∈QA. Prove that segment KP is bisected by line MN.
Click for solution draw PX perpendicular to AB ( X is a point on AB) We know that O, A, P, Q, B are cyclic. Then apply Simson's theorem from point P to the triangle ABQ. K, N, M are on the same line. angle QOP = angle QAP = angle NXP = angle KPX (because OQ are parallel to XP) So KP is bisected by line NM
Let n be positive integer, set M={1,2,…,2n}. Find the minimum positive integer k such that for any subset A (with k elements) of set M, there exist four pairwise distinct elements in A whose sum is 4n+1.
Click for solution k=n+3 will do. to start with note that the set {n−1,n,n+1,n+2,…,2n} is a set of n+2 elements and has a minimum sum (with 4 distinct elements) of n−1+n+n+1+n+2=4n+2, so k≥n+3.also for a set of size atleast n+2 we can form pairs of distinct elements that sum to 2n and pigeonhole ⇒ there exists a pair of distinct elements with sum 2n.once we remove such a pair from the n+3-set and we are still left with an n+1 set from which another pair,this time with sum 2n+1 (again pigeonhole) can be extracted and so we are through.
Find all positive integer solutions (a,b,c) to the function a2+b2+c2=2005, where a≤b≤c.
Click for solution I do not know the official solution, but here's my solution anyway. We consider \mod 8. Since all squares are \equiv 0, 1 or 4 \pmod{8} and 2005 \equiv 5 \pmod{8} it follows that a^2, b^2, c^2 are \equiv 0, 1, 4 \pmod{8} in some order. WLOG, we ignore the order of a,b,c (we drop the assumption of a \leq b \leq c) and assume a^2 \equiv 0 \pmod{8}, b^2 \equiv 1 \pmod{8}, c^2 \equiv 4 \pmod{8}. Let a=4a_1, b=2b_1+1, c=4c_1+2. We then have a^2+b^2+c^2=16{a_1}^2+4{b_1}^2+4b_1+16{c_1}^2+16c_1+5 = 2005 \Leftrightarrow 4{a_1}^2+{b_1}^2+b_1 + 4{c_1}^2+4c_1 = 500. Taking \mod 4, we have {b_1}^2 + b_1 \equiv 0 \pmod{4} \Rightarrow b_1 \equiv 0 or -1 \pmod{4}. Case 1: b_1 \equiv 0 \pmod{4}. Let b_1 = 4b_2 We have 4{a_1}^2+16{b_2}^2+4b_2 + 4{c_1}^2+4c_1 = 500 \Leftrightarrow {a_1}^2+4{b_2}^2+b_2 + {c_1}^2+c_1 = 125. Since b_2 \in \mathbb{Z}^+ and 4{b_2}^2 > 125 \ \forall \ b_2 > 5, we have 1 \leq b_2 \leq 5. When b_2 = 5, we have {a_1}^2 +{c_1}^2+c_1=20. Note that {c_1}^2+c_1 \equiv 0 \pmod{2}, thus {a_1}^2 = 4 or 16. We then have {c_1}^2+c_1 = 16 or 4 respectively, which yield no solutions. When b_2=4, we have {a_1}^2 +{c_1}^2+c_1=57. Note that {c_1}^2+c_1 \equiv 0 \pmod{2}, thus {a_1}^2 = 1, 9, 25 or 49. We then have {c_1}^2+c_1 = 56, 48, 32 or 8 respectively, which yield the only possible solution c_1 = 7 \Rightarrow (a,b,c) = (4,33,30). When b_2=3, we have {a_1}^2 +{c_1}^2+c_1=86. Similarly, {a_1}^2 can only be 4, 16, 36, or 64, which means {c_1}^2+c_1 = 82, 70, 50 or 22 respectively, which yield no solutions. When b_2=2, we have {a_1}^2 +{c_1}^2+c_1=107. Similarly, {a_1}^2 can only be 1, 9, 25, 49 or 81, which means {c_1}^2+c_1 = 106, 98, 82, 58 or 26 respectively, which yield no solutions. When b_2=1, we have {a_1}^2 +{c_1}^2+c_1=120. Similarly, {a_1}^2 can only be 4, 16, 36, 64 or 100, which means {c_1}^2+c_1 = 116, 104, 84, 56 or 20 respectively, which yield the possible solutions c_1 = 7, 4 \Rightarrow (a,b,c) = (32, 9, 30), (40, 9, 18). Case 2: b_1 \equiv -1 \pmod{4}. Let b_1 = 4b_2 -1 We have 4{a_1}^2+16{b_2}^2-4b_2 + 4{c_1}^2+4c_1 = 500 \Leftrightarrow {a_1}^2+4{b_2}^2-b_2 + {c_1}^2+c_1 = 125. Since b_2 \in \mathbb{Z}^+ and 4{b_2}^2 -b_2 > 125 \forall b_2 > 5, we have 1 \leq b_2 \leq 5 When b_2 = 5, we have {a_1}^2 +{c_1}^2+c_1=30. Similarly, by considering parity, we must have {a_1}^2 = 4 or 16, which means {c_1}^2+c_1 = 26 or 14 respectively, which yield no solutions. When b_2=4, we have {a_1}^2 +{c_1}^2+c_1=65. Similarly, we have {a_1}^2 = 1, 9, 25 or 49, meaning {c_1}^2+c_1 = 64, 56, 40 or 16 respectively, which yield the only possible solution c_1 = 7 \Rightarrow (a,b,c) = (12, 31, 30). When b_2=3, we have {a_1}^2 +{c_1}^2+c_1=92. Similarly, we have {a_1}^2 = 4, 16, 36 or 64, meaning {c_1}^2+c_1 = 88, 76, 56 or 28 respectively, which yield the only possible solution c_1 = 7 \Rightarrow (a,b,c) = (24, 23, 30). When b_2=2, we have {a_1}^2 +{c_1}^2+c_1=111. Similarly, we have {a_1}^2 = 1, 9, 25, 49 or 81, meaning {c_1}^2+c_1 = 110, 102, 86, 62 or 30 respectively, which yield the possible solutions c_1 = 5, 10 \Rightarrow (a,b,c) = (4, 15, 22), (36, 15, 42). When b_2=1, we have {a_1}^2 +{c_1}^2+c_1=122. Similarly, {a_1}^2 can only be 1, 9, 25, 49, 81 or 121, which means {c_1}^2+c_1 = 121, 113, 97, 73, 41 or 1 respectively, which yield no solutions. Therefore, by considering the above 10 cases, the only seven possible solutions are (4,33,30), (32,9,30), (40,9,18), (12,31,30), (24,23,30), (4,15,22), (36,15,42). As a side note, we can actually make use of what Singular has mentioned regarding the solvability of a^2+b^2 =k to reduce the number of cases, but this does not necessarily make the solution shorter. Using the "Dirichlet Class-Number Theorem", or rather, finding the appropriate class numbers, to compute \frac{r_3(2005)}{2^3 \cdot 3!} - \frac{r_2(2005)}{2^2 \cdot 2!} = 7 is but an unnecessary complication. And besides, I haven't seen any olympiad problems that ask "Find the number of solutions..." instead of the usual "Find all solutions to...", so I doubt this theorem will be used that often.
July 11th - Day 2
Line l tangents unit circle S in point P. Point A and circle S are on the same side of l, and the distance from A to l is h (h > 2). Two tangents of circle S are drawn from A, and intersect line l at points B and C respectively. Find the value of PB \cdot PC.
Click for solution Consider the circle (S) of radius r = 1 and the line l directed. Let k be a directed line parallel to the line l at a distance h and A a point on this line outside of the circle (S). Let A be a point on the line k outside of the circle (S) and H the foot of a normal from the point A to the line l. Let b, c be the 2 directed tangents from the point A to the circle (S) intersecting the line l at points B, C and let \beta = \widehat{(b, k)} = \widehat{(b, l)} = \measuredangle ABH \gamma = \widehat{(c, k)} = \widehat{(c, l)} = 180^\circ - \measuredangle ACH be the directed angles formed by the directed tangents b, c with the directed lines k, l. (A directed tangent must have the same direction as the directed circle at their tangency point. As opposed to 2 undirected lines, which form 2 different complementary angles, 2 directed lines form a unique angle.) Then \frac{r}{PB} = \frac{PS}{PB} = \tan \widehat{SBP} = \tan \left(90^\circ - \frac{\widehat{ABH}}{2}\right) = \tan \left(90^\circ - \frac{\beta}{2}\right) = \frac{1}{\tan \frac \beta 2} {\frac{r}{PC} = \frac{PS}{PC} = \tan \widehat{SCP}} = \tan \frac{\widehat{ACH}}{2} = \tan \left(90^\circ - \frac{\gamma}{2}\right) = \frac{1}{\tan \frac \gamma 2} The power of the directed line k to the directed circle (S) is equal to \tan \frac{\widehat{(b, k)}}{2} \cdot \tan \frac{\widehat{(c, k)}}{2} = \tan \frac{\beta}{2} \cdot \tan \frac{ \gamma}{2} = \frac{PB \cdot PC}{r^2} and it does not depend on the position of the point A on the directed line k. Obviously, this is a dual concept of the power of a point to an undirected circle, which does not depend on the angle of the circle secant from this point. I guess it is not considered a pure geometry, because it contains trigonometric functions . To calculate the power of the directed line k to the directed circle (S), we can move the point A into the foot A_0 of a normal from the tangency point P to the line k, because then \gamma_0 = \beta_0 and \tan \frac{\beta}{2} \cdot \tan \frac{ \gamma}{2} = \tan \frac{\beta_0}{2} \cdot \tan \frac{ \gamma_0}{2} = \tan^2 \frac{\beta_0}{2} Let U_0, V_0 be the tangency points of the directed tangents from the point A_0 to the directed circle (S) intersecting the directed line l at points B_0, C_0. \frac{r}{h - r} = \frac{U_0S}{A_0S} = \cos \widehat{A_0SU_0} = \cos \widehat{A_0BP} = \cos (180^\circ - \beta_0) = -\cos \beta_0 \cos \beta_0 = \cos^2 \frac{\beta_0}{2} - \sin^2 \frac{\beta_0}{2} = \cos^2 \frac{\beta_0}{2} \left(1 - \tan^2 \frac{\beta_0}{2}\right) = \frac{1 - \tan^2 \frac{\beta_0}{2}}{1 + \tan^2 \frac{\beta_0}{2}} = \frac{1 - x}{1 + x} Solving for x, we get x = \tan^2 \frac{\beta_0}{2} = \frac{1 - \cos \beta_0}{1 + \cos \beta_0} = \frac{1 + \frac{r}{h - r}}{1 - \frac{r}{h - r}} = \frac{h - r + r}{h - r - r} = \frac{h}{h - 2r} PB \cdot PC = r^2 \tan^2 \frac{\beta_0}{2} = \frac{r^2h}{h - 2r} and for a unit circle r = 1, PB \cdot PC = \frac{h}{h - 2}
Let P(A) be the arithmetic-means of all elements of set A = \{ a_1, a_2, \ldots, a_n \}, namely P(A) = \frac{1}{n} \sum^{n}_{i=1}a_i. We denote B "balanced subset" of A, if B is a non-empty subset of A and P(B) = P(A). Let set M = \{ 1, 2, 3, 4, 5, 6, 7, 8, 9 \}. Find the number of all "balanced subset" of M.
Click for solution I'd look at the numbers not as they are written down now, but at their difference with 5. So: 4, 3, 2, 1, 0, 1, 2, 3, 4 At both the left and the right part, the sum of the differences must be the same. D is the total difference at one side: If D = 0: One possibility at both sides: (empty) 1*1=1 If D = 1: One possibility at both sides: 4, 6 1*1=1 If D = 2: One possibility at both sides: 3, 7 1*1=1 If D = 3: Two possibilities at both sides: 3+4 / 2, 6+7 / 8 2*2=4 If D = 4: Two possibilities at both sides: 2+4 / 1, 6+8 / 9 2*2=4 If D = 5: Two possibilities at both sides: 1+4 / 2+3, 6+9 / 7+8 2*2=4 If D = 6: Two possibilities at both sides: 1+3 / 2+3+4, 7+9 / 6+7+8 2*2=4 If D = 7: Two possibilities at both sides: 1+2 / 1+3+4, 8+9 / 6+7+9 2*2=4 If D = 8: One possibility: no 3, 7 1*1=1 If D=9: One possibility: no 4,6 1*1=1 If D=10: all numbers: 1 Counting gives us 26. Multiplied by 2 for including or excluding 5 gives 52, minus the empty subset gives us 51 as well!
(1) Find the possible number of roots for the equation |x + 1| + |x + 2| + |x + 3| = a, where x \in R and a is parameter. (2) Let \{ a_1, a_2, \ldots, a_n \} be an arithmetic progression, n \in \mathbb{N}, and satisfy the condition \sum^{n}_{i=1}|a_i| = \sum^{n}_{i=1}|a_{i} + 1| = \sum^{n}_{i=1}|a_{i} - 2| = 507. Find the maximum value of n.
Click for solution First, observe that if a_{1}<a_{2}<\ldots<a_{n} are real numbers and f:\mathbb{R}\rightarrow\mathbb{R} is defined as f\left( x\right) =\left| x-a_{1}\right| +\left| x-a_{2}\right| +\ldots+\left| x-a_{n}\right| , then f is first decreasing, and then increasing. More precisely, if n is odd, say n=2k+1, then f is strictly decreasing on the interval (-\infty,a_{k}] and strictly increasing on [a_{k},+\infty), while if n is even, say n=2k, then f is strictly decreasing on the interval (-\infty,a_{k}], constant on [a_{k},a_{k+1}] and strictly increasing on [a_{k+1},+\infty). Now, let WLOG a_{1}\leq\ldots\leq a_{n} be our arithmetic sequence and f defined as above. We see that f(0)=f\left( -1\right) =f\left( 2\right), hence three values of f are equal. This is possible only if n is even (say n=2k) and 0,-1,2\in\lbrack a_{k},a_{k+1}]. We deduce that the common difference d is at least 3, and a_{k}\leq-1,a_{k+1}\geq2. A short computation shows that in this case 507=\sum\left| a_{i}\right| \geq3k^{2}, hence k\leq13, that is, n\leq26. Equality is obtained when a_{1}=-37,a_{2}=-34,\ldots,a_{26}=38.
Let 0 < \alpha, \beta, \gamma < \frac{\pi}{2} and \sin^{3} \alpha + \sin^{3} \beta + \sin^3 \gamma = 1. Prove that \tan^{2} \alpha + \tan^{2} \beta + \tan^{2} \gamma \geq \frac{3 \sqrt{3}}{2} .