Difference between revisions of "2011 AIME I Problems/Problem 7"

m (Solution 2)
(Solution)
Line 3: Line 3:
 
<cmath>m^{x_0} = \sum_{k = 1}^{2011} m^{x_k}.</cmath>
 
<cmath>m^{x_0} = \sum_{k = 1}^{2011} m^{x_k}.</cmath>
  
==Solution==
+
==Solution ==
 +
<math> m^{x_0}= m^{x_1} +m^{x_2} + .... + m^{x_{2011}}</math>. Now, divide by <math>m^{x_0}</math> to get <math>1= m^{x_1-x_0} +m^{x_2-x_0} + .... + m^{x_{2011}-x_0}</math>. Notice that since we can choose all nonnegative <math>x_0,...,x_{2011}</math>, we can make <math>x_n-x_0</math> whatever we desire. WLOG, let <math>x_0\geq...\geq x_{2011}</math> and let <math>a_n=x_n-x_0</math>.  Notice that, also, <math>m^{a_2011}</math> doesn't matter if we are able to make <math> m^{a_1} +m^{a_2} + .... + m^{a_2011}</math> equal to <math>1-(\frac{1}{m})^x</math> for any power of <math>x</math>. Consider <math>m=2</math>. We can achieve a sum of <math>1-(\frac{1}{2})^x</math> by doing <math>\frac{1/2}+\frac{1/4}+...</math> (the simplest" sequence). If we don't have <math>\frac{1}{2}</math>, to compensate, we need <math>2\cdot 1\frac{1}{4}</math>'s. Now, let's try to generalize. The "simplest" sequence is having <math>\frac{1}{m}</math> <math>m-1</math> times, <math>\frac{1}{m^2}</math> <math>m-1</math> times, <math>\ldots</math>. To make other sequences, we can split <math>m-1</math> <math>\frac{1}{m^i}</math>s into <math>m(m-1)</math> <math>\frac{1}{m^{i+1}}</math>s since <math>\frac{1}{m^{i+1}}\cdot =\frac{1}{m^{i}}</math>. Since we want <math>2010</math> terms, we have <math>\sum (m-1)\cdotm^x=2010</math>. However, since we can set <math>x</math> to be anything we want (including 0), all we care about is that <math>m-1 | 2010</math> which happens <math>\boxed{016}</math> times.
 +
 
 +
==Solution 1==
 
Let <math>P(m) = m^{x_0} - m^{x_1} -m^{x_2} - .... - m^{x_{2011}}</math>. The problem then becomes finding the number of positive integer roots <math>m</math> for which <math>P(m) = 0</math> and <math>x_0, x_1, ..., x_{2011}</math> are nonnegative integers. We plug in <math>m = 1</math> and see that <math>P(1) = 1 - 1 - 1... -1 = 1-2011 = -2010</math>. Now, we can say that <math>P(m) = (m-1)Q(m) - 2010</math> for some polynomial <math>Q(m)</math> with integer coefficients. Then if <math>P(m) = 0</math>, <math>(m-1)Q(m) = 2010</math>. Thus, if <math>P(m) = 0</math>, then <math>m-1 | 2010</math> .
 
Let <math>P(m) = m^{x_0} - m^{x_1} -m^{x_2} - .... - m^{x_{2011}}</math>. The problem then becomes finding the number of positive integer roots <math>m</math> for which <math>P(m) = 0</math> and <math>x_0, x_1, ..., x_{2011}</math> are nonnegative integers. We plug in <math>m = 1</math> and see that <math>P(1) = 1 - 1 - 1... -1 = 1-2011 = -2010</math>. Now, we can say that <math>P(m) = (m-1)Q(m) - 2010</math> for some polynomial <math>Q(m)</math> with integer coefficients. Then if <math>P(m) = 0</math>, <math>(m-1)Q(m) = 2010</math>. Thus, if <math>P(m) = 0</math>, then <math>m-1 | 2010</math> .
 
Now, we need to show that for all <math>m-1 | 2010</math>, <math> m^{x_{0}}=\sum_{k = 1}^{2011}m^{x_{k}}. </math>. We try with the first few <math>m</math> that satisfy this.
 
Now, we need to show that for all <math>m-1 | 2010</math>, <math> m^{x_{0}}=\sum_{k = 1}^{2011}m^{x_{k}}. </math>. We try with the first few <math>m</math> that satisfy this.

Revision as of 02:00, 7 February 2016

Problem 7

Find the number of positive integers $m$ for which there exist nonnegative integers $x_0$, $x_1$ , $\dots$ , $x_{2011}$ such that \[m^{x_0} = \sum_{k = 1}^{2011} m^{x_k}.\]

Solution

$m^{x_0}= m^{x_1} +m^{x_2} + .... + m^{x_{2011}}$. Now, divide by $m^{x_0}$ to get $1= m^{x_1-x_0} +m^{x_2-x_0} + .... + m^{x_{2011}-x_0}$. Notice that since we can choose all nonnegative $x_0,...,x_{2011}$, we can make $x_n-x_0$ whatever we desire. WLOG, let $x_0\geq...\geq x_{2011}$ and let $a_n=x_n-x_0$. Notice that, also, $m^{a_2011}$ doesn't matter if we are able to make $m^{a_1} +m^{a_2} + .... + m^{a_2011}$ equal to $1-(\frac{1}{m})^x$ for any power of $x$. Consider $m=2$. We can achieve a sum of $1-(\frac{1}{2})^x$ by doing $\frac{1/2}+\frac{1/4}+...$ (the simplest" sequence). If we don't have $\frac{1}{2}$, to compensate, we need $2\cdot 1\frac{1}{4}$'s. Now, let's try to generalize. The "simplest" sequence is having $\frac{1}{m}$ $m-1$ times, $\frac{1}{m^2}$ $m-1$ times, $\ldots$. To make other sequences, we can split $m-1$ $\frac{1}{m^i}$s into $m(m-1)$ $\frac{1}{m^{i+1}}$s since $\frac{1}{m^{i+1}}\cdot =\frac{1}{m^{i}}$. Since we want $2010$ terms, we have $\sum (m-1)\cdotm^x=2010$ (Error compiling LaTeX. Unknown error_msg). However, since we can set $x$ to be anything we want (including 0), all we care about is that $m-1 | 2010$ which happens $\boxed{016}$ times.

Solution 1

Let $P(m) = m^{x_0} - m^{x_1} -m^{x_2} - .... - m^{x_{2011}}$. The problem then becomes finding the number of positive integer roots $m$ for which $P(m) = 0$ and $x_0, x_1, ..., x_{2011}$ are nonnegative integers. We plug in $m = 1$ and see that $P(1) = 1 - 1 - 1... -1 = 1-2011 = -2010$. Now, we can say that $P(m) = (m-1)Q(m) - 2010$ for some polynomial $Q(m)$ with integer coefficients. Then if $P(m) = 0$, $(m-1)Q(m) = 2010$. Thus, if $P(m) = 0$, then $m-1 | 2010$ . Now, we need to show that for all $m-1 | 2010$, $m^{x_{0}}=\sum_{k = 1}^{2011}m^{x_{k}}.$. We try with the first few $m$ that satisfy this. For $m = 2$, we see we can satisfy this if $x_0 = 2010$, $x_1 = 2009$, $x_2 = 2008$, $\cdots$ , $x_{2008} = 2$, $x_{2009} = 1$, $x_{2010} = 0$, $x_{2011} = 0$, because $2^{2009} + 2^{2008} + \cdots + 2^1 + 2^0 +2^ 0 = 2^{2009} + 2^{2008} + \cdots + 2^1 + 2^1 = \cdots$ (based on the idea $2^n + 2^n = 2^{n+1}$, leading to a chain of substitutions of this kind) $= 2^{2009} + 2^{2008} + 2^{2008} = 2^{2009} + 2^{2009} = 2^{2010}$. Thus $2$ is a possible value of $m$. For other values, for example $m = 3$, we can use the same strategy, with $x_{2011} = x_{2010} = x_{2009} = 0$, $x_{2008} = x_{2007} = 1$, $x_{2006} = x_{2005} = 2$, $\cdots$, $x_2 = x_1 = 1004$ and $x_0 = 1005$, because $3^0 + 3^0 + 3^0 +3^1+3^1+3^2+3^2+\cdots+3^{1004} +3^{1004} = 3^1+3^1+3^1+3^2+3^2+\cdots+3^{1004} +3^{1004} = 3^2+3^2+3^2+\cdots+3^{1004} +3^{1004} = \cdots$ $=3^{1004} +3^{1004}+3^{1004} = 3^{1005}$. It's clearly seen we can use the same strategy for all $m-1 |2010$. We count all positive $m$ satisfying $m-1 |2010$, and see there are $\boxed{016}$

Solution 2

One notices that $m-1 \mid 2010$ if and only if there exist non-negative integers $x_0,x_1,\ldots,x_{2011}$ such that $m^{x_0} = \sum_{k=1}^{2011}m^{x_k}$.

To prove the forward case, we proceed by directly finding $x_0,x_1,\ldots,x_{2011}$. Suppose $m$ is an integer such that $m^{x_0} = \sum_{k=1}^{2011}m^{x_k}$. We will count how many $x_k = 0$, how many $x_k = 1$, etc. Suppose the number of $x_k = 0$ is non-zero. Then, there must be at least $m$ such $x_k$ since $m$ divides all the remaining terms, so $m$ must also divide the sum of all the $m^0$ terms. Thus, if we let $x_k = 0$ for $k = 1,2,\ldots,m$, we have, \[m^{x_0} = m + \sum_{k=m+1}^{2011}m^{x_k}.\] Well clearly, $m^{x_0}$ is greater than $m$, so $m^2 \mid m^{x_0}$. $m^2$ will also divide every term, $m^{x_k}$, where $x_k \geq 2$. So, all the terms, $m^{x_k}$, where $x_k < 2$ must sum to a multiple of $m^2$. If there are exactly $m$ terms where $x_k = 0$, then we must have at least $m-1$ terms where $x_k = 1$. Suppose there are exactly $m-1$ such terms and $x_k = 1$ for $k = m+1,m+2,2m-1$. Now, we have, \[m^{x_0} = m^2 + \sum_{k=2m}^{2011}m^{x_k}.\] One can repeat this process for successive powers of $m$ until the number of terms reaches 2011. Since there are $m + j(m-1)$ terms after the $j$th power, we will only hit exactly 2011 terms if $m-1$ is a factor of 2010. To see this, \[m+j(m-1) = 2011 \Rightarrow m-1+j(m-1) = 2010 \Rightarrow (m-1)(j+1) = 2010.\] Thus, when $j = 2010/(m-1) - 1$ (which is an integer since $m-1 \mid 2010$ by assumption, there are exactly 2011 terms. To see that these terms sum to a power of $m$, we realize that the sum is a geometric series: \[1 + (m-1) + (m-1)m+(m-1)m^2 + \cdots + (m-1)m^j = 1+(m-1)\frac{m^{j+1}-1}{m-1} = m^{j+1}.\] Thus, we have found a solution for the case $m-1 \mid 2010$.

Now, for the reverse case, we use the formula \[x^k-1 = (x-1)(x^{k-1}+x^{k-2}+\cdots+1).\] Suppose $m^{x_0} = \sum_{k=1}^{2011}m^{x^k}$ has a solution. Subtract 2011 from both sides to get \[m^{x_0}-1-2010 = \sum_{k=1}^{2011}(m^{x^k}-1).\] Now apply the formula to get \[(m-1)a_0-2010 = \sum_{k=1}^{2011}[(m-1)a_k],\] where $a_k$ are some integers. Rearranging this equation, we find \[(m-1)A = 2010,\] where $A = a_0 - \sum_{k=1}^{2011}a_k$. Thus, if $m$ is a solution, then $m-1 \mid 2010$.

So, there is one positive integer solution corresponding to each factor of 2010. Since $2010 = 2\cdot 3\cdot 5\cdot 67$, the number of solutions is $2^4 = \boxed{016}$.

See also

2011 AIME I (ProblemsAnswer KeyResources)
Preceded by
Problem 6
Followed by
Problem 8
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
All AIME Problems and Solutions

The problems on this page are copyrighted by the Mathematical Association of America's American Mathematics Competitions. AMC logo.png