返回题库

AIME 2017 I · 第 13 题

AIME 2017 I — Problem 13

专题
Contest Math
难度
L4
来源
AIME

题目详情

Problem 13

For every m2m \geq 2, let Q(m)Q(m) be the least positive integer with the following property: For every nQ(m)n \geq Q(m), there is always a perfect cube k3k^3 in the range n<k3mnn < k^3 \leq m \cdot n. Find the remainder when

m=22017Q(m)\sum_{m = 2}^{2017} Q(m) is divided by 1000.

解析

Solution 1

Lemma 1: The ratio between k3k^3 and (k+1)3(k+1)^3 decreases as kk increases.

Lemma 2: If the range (n,mn](n,mn] includes yy cubes, (p,mp](p,mp] will always contain at least y1y-1 cubes for all pp in [n,+)[n,+\infty).

If m=14m=14, the range (1,14](1,14] includes one cube. The range (2,28](2,28] includes 2 cubes, which fulfills the Lemma. Since n=1n=1 also included a cube, we can assume that Q(m)=1Q(m)=1 for all m>14m>14. Two groups of 1000 are included in the sum modulo 1000. They do not count since Q(m)=1Q(m)=1 for all of them, therefore

m=22017Q(m)m=217Q(m)mod1000\sum_{m = 2}^{2017} Q(m) \equiv \sum_{m = 2}^{17} Q(m) \mod 1000 Now that we know this we will find the smallest nn that causes (n,mn](n,mn] to contain two cubes and work backwards (recursion) until there is no cube in (n,mn](n,mn].

For m=2m=2 there are two cubes in (n,2n](n,2n] for n=63n=63. There are no cubes in (31,62](31,62] but there is one in (32,64](32,64]. Therefore Q(2)=32Q(2)=32.

For m=3m=3 there are two cubes in (n,3n](n,3n] for n=22n=22. There are no cubes in (8,24](8,24] but there is one in (9,27](9,27]. Therefore Q(3)=9Q(3)=9.

For mm in {4,5,6,7}\{4,5,6,7\} there are two cubes in (n,4n](n,4n] for n=7n=7. There are no cubes in (1,4](1,4] but there is one in (2,8](2,8]. Therefore Q(4)=2Q(4)=2, and the same for Q(5)Q(5), Q(6)Q(6), and Q(7)Q(7) for a sum of 88.

For all other mm there is one cube in (1,8](1,8], (2,16](2,16], (3,24](3,24], and there are two in (4,32](4,32]. Therefore, since there are 10 values of mm in the sum, this part sums to 1010.

When the partial sums are added, we get 059QED\boxed{059}\hspace{2 mm}QED\hspace{2 mm} \blacksquare

This solution is brought to you by a1b2

Solution 2

We claim that Q(m)=1Q(m) = 1 when m8m \ge 8.

When m8m \ge 8, for every nQ(m)=1n \ge Q(m) = 1, we need to prove there exists an integer kk, such that n<k3mnn < k^3 \le m \cdot n.

That's because mn3n32n3n3=n31\sqrt[3]{m \cdot n} - \sqrt[3]{n} \ge 2\sqrt[3]{n} - \sqrt[3]{n} = \sqrt[3]{n} \ge 1, so k exists between mn3\sqrt[3]{m \cdot n} and n3\sqrt[3]{n}

n3<kmn3\sqrt[3]{n} < k \le \sqrt[3]{m \cdot n}.

We can then hand evaluate Q(m)Q(m) for m=2,3,4,5,6,7m = 2,3,4,5,6,7, and get Q(2)=32Q(2) = 32, Q(3)=9Q(3) = 9, and all the others equal 2.

There are a total of 2010 integers from 8 to 2017.

m=22017Q(m)m=27Q(m)+201032+9+2+2+2+2+10=059mod1000\sum_{m = 2}^{2017} Q(m) \equiv \sum_{m = 2}^{7} Q(m) + 2010 \equiv 32+9+2+2+2+2+10 = \boxed{059} \mod 1000 -AlexLikeMath

Solution 3

Note that the problem is just asking for every m2m \geq 2, find the least nn such that there lies a perfect cube in between of nn and mnmn for every nn after that minimum including the minimum. We consider m=2m = 2 first. Then, we need to find the least nn such that for every nn after this minimum nn, including the minimum, there lies a perfect cube in between of nn and 2n2n. Let's denote the range notation as (n,2n)(n, 2n). Hence, we start with n=1n = 1 to get the range of (1,2)(1, 2). Clearly, no perfect cube lies between these numbers. Then we go to n=2n = 2 to get (2,4)(2, 4) and here also nothing works. We can see that the perfect cubes are 1,8,2764,125,216,343,512,729,1000,...1, 8, 27 64, 125, 216, 343, 512, 729, 1000, .... Because 11 can't lie between two positive integers, we need to find the minimum nn such that 88 is between nn and 2n2n. Clearly, n=4n = 4 will yield this case. We have the range (4,8)(4, 8). This works. We now list every nn starting from 44 and see if all of these ranges host a perfect cube in between them. We have (5,10)(5, 10) works, (6,12)(6, 12) works (7,14)(7, 14) works, but (8,16)(8, 16) doesn't work. This is because the next perfect cube has to be 2727. Aha! Now we see the case. If we have a range in the following form (a32,a3)(\frac{a^{3}}{2}, a^{3}), this will clearly host a perfect cube between them which is a3a^{3}. But as we keep going, we need to put the next perfect cube after a3a^{3} into the range which is (a+1)3(a + 1)^{3}. We know that the next range after (a32,a3)(\frac{a^{3}}{2}, a^{3}) is just (a3,2a3)(a^{3}, 2a^{3}) because we double everything. Hence, (a+1)3(a + 1)^{3} must be in this range. If the perfect cube in between a3a^{3} and 2a32a^{3} is of the form a3+αa^{3} + \alpha, we know that this added constant α\alpha is greater than or equal to (a+1)3a3=3a2+3a+1(a + 1)^{3} - a^{3} = 3a^{2} + 3a + 1. Hence, we let α=a\alpha = a to solve a33a2+3a+1a^{3} \geq 3a^{2} + 3a + 1 and for integer solutions, we claim the answer is a4a \geq 4. We can prove this by induction.

Claim: a33a2+3a+1a^{3} \geq 3a^{2} + 3a + 1 for integer solutions exists when a4a \geq 4.

Proof: Let's plug in a=4a = 4 for our base case. We get: 6448+12+1=6164 \geq 48 + 12 + 1 = 61 which holds true. Hence, the inductive hypothesis is that (a+1)33(a+1)2+3(a+1)+1(a + 1)^{3} \geq 3(a + 1)^{2} + 3(a + 1) + 1 for all a3a \geq 3. In fact, we simplify to prove a36a+6a^{3} \geq 6a + 6 for all a3a \geq 3. Now, from our inductive assumption, we need to prove that 3a2+3a+16a+63a^{2} + 3a + 1 \geq 6a + 6. Hence, 3a23a53a^{2} - 3a \geq 5 for all a3a \geq 3. Hence, we need to show a2a53a^{2} - a \geq \frac{5}{3} for all a3a \geq 3. But since we're proving this in the integer world, this is equivalent to proving that a2a2a^{2} -a \geq 2 for all a3a \geq 3. Now we need to show that a2a+2a^{2} \geq a + 2. Dividing by aa because aa is a positive integer means we have to show a1+2aa \geq 1 + \frac{2}{a}. Now since a3a \geq 3, 2a<1\frac{2}{a} < 1 and so the RHSRHS of the inequality can't even exceed 22 whereas the LHSLHS of the inequality is already exceeding 22. Therefore, the inductive hypothesis holds true and we are done.

What we have just shown now is that a4a \geq 4. Now note that the minimum range we found that worked was (a32,a3)(\frac{a^{3}}{2}, a^{3}) where n=a32n = \frac{a^{3}}{2}. We plug in a=4a = 4 to get n=32n = 32 and hence Q(2)=32Q(2) = 32. We can now apply the same procedure to the other numbers and finish as the above solutions.

~ilikemath247365