Find the Upper Bound

Algebra Level 2

There exists a smallest possible positive integer N N such that

( x 1 + 2 x 2 + + 2014 x 2014 ) 2 x 1 2 + x 2 2 + + x 2014 2 N \dfrac{(x_1+2x_2+\cdots +2014x_{2014})^2}{x_1^2+x_2^2+\cdots +x_{2014}^2}\le N

for all real sequences { x i } i = 1 2014 \large \{x_i\}_{i=1}^{2014} .

Find the sum of digits of N N .


The answer is 38.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

9 solutions

Cody Johnson
Dec 22, 2013

Titu's lemma (Cauchy):

n 2 d ( n ) 2 d \sum \frac{n^2}{d} \ge \frac{(\sum n)^2}{\sum d}

Hence,

( x 1 + 2 x 2 + + 2014 x 2014 ) 2 x 1 2 + x 2 2 + + x 2014 2 1 2 + 2 2 + + 201 4 2 \frac{(x_1+2x_2+\dots+2014x_{2014})^2}{x_1^2+x_2^2+\dots+x_{2014}^2}\le1^2+2^2+\dots+2014^2 whose digit sum is 38 \boxed{38} .

MOT!!!!

Matthew Yu - 7 years, 5 months ago
Daniel Chiu
Dec 21, 2013

By Titu's lemma (see here ) or Cauchy-Schwarz (multiply by the RHS on the right side), i = 1 2014 i 2 x i 2 x i 2 = 1 2 x 1 2 x 1 2 + 2 2 x 2 2 x 2 2 + + 201 4 2 x 2014 2 x 2014 2 ( x 1 + 2 x 2 + + 2014 x 2014 ) 2 x 1 2 + x 2 2 + + x 2014 2 \sum_{i=1}^{2014}\dfrac{i^2x_i^2}{x_i^2}=\dfrac{1^2x_1^2}{x_1^2}+\dfrac{2^2x_2^2}{x_2^2}+\cdots+\dfrac{2014^2x_{2014}^2}{x_{2014}^2}\ge \dfrac{(x_1+2x_2+\cdots+2014x_{2014})^2}{x_1^2+x_2^2+\cdots+x_{2014}^2} Then, N = i = 1 2014 i 2 x i 2 x i 2 = i = 1 2014 i 2 = ( 2014 ) ( 2015 ) ( 4029 ) 6 = 2725088015 N=\sum_{i=1}^{2014}\dfrac{i^2x_i^2}{x_i^2}=\sum_{i=1}^{2014} i^2=\dfrac{(2014)(2015)(4029)}{6}=2725088015 which has digit sum of 38 \boxed{38} .

Motivation: The expression looks suspiciously like Titu's lemma, and also if I see a square that is less than something, I think Cauchy-Schwarz.

It follows directly from Cauchy-Schwarz that ( x 1 + 2 x 2 + 3 x 3 + . . . + 2014 x 2014 ) 2 ( x 1 2 + x 2 2 + . . . + x 2014 2 ) ( 1 2 + 2 2 + 3 3 + . . . + 201 4 2 ) (x_1+2x_2+3x_3+...+2014x_{2014})^2 \le (x_1^2+x_2^2+...+x_{2014}^2)(1^2+2^2+3^3+...+2014^2)

minimario minimario - 7 years, 5 months ago

Log in to reply

Well yeah since the x x 's cancel.

Also you know you can edit a comment

Daniel Chiu - 7 years, 5 months ago

Don't you need to find an equality case for this N , N, to show that N N cannot be lower? Take, for instance, x i = i x_i = i for all i , i, which shows that N 1 2 + + 201 4 2 . N \ge 1^2 + \ldots + 2014^2.

Michael Tang - 7 years, 5 months ago

Log in to reply

Yeah, you are correct. Equality holds whenever x i = k i x_i=ki for all i i and some nonzero constant k k .

Daniel Chiu - 7 years, 5 months ago

Wouldn't you also have to mention the rearrangement inequality to say how the LHS is maximized? i.e., why can't we have 201 4 2 x 2014 2 x 1 2 + \frac{2014^2 x_{2014}^2}{x_{1}^2} + \cdots or something like that?

Michael Tong - 7 years, 5 months ago

Log in to reply

I think citing an equality case as Michael Tang gave means that the LHS can be maximized.

Daniel Chiu - 7 years, 5 months ago
Mike Kong
Dec 23, 2013

It's not 2014 yet!...

Since we want a maximum value of the LHS, assume that x 1 x 2 x 3 x 2014 x_1 \leq x_2 \leq x_3 \leq \cdots \leq x_{2014} . We can assume that because if x i > x j x_i > x_j for i < j i < j , then we can optimize the expression further by switching values for x i x_i and x j x_j .

The LHS looks a lot like a rearrangement of Cauchy-Schwarz / Titu's Lemma. On the RHS = N, we will have an expression made up of 2014 2014 fractions, with numerators chosen from the set { x 1 2 , ( 2 x 2 ) 2 , ( 3 x 3 ) 2 , , ( 2014 x 2014 ) 2 } \{x_1^2, (2x_2)^2, (3x_3)^2, \cdots, (2014x_{2014})^2\} and denominators chosen from the set { x 1 2 , x 2 2 , x 3 2 , , x 2014 2 } \{x_1^2, x_2^2, x_3^2, \cdots, x_{2014}^2\} . We wish to minimize N N , and so by the rearrangement inequality we know that it is minimized when the greatest numerator is paired up with the greatest denominator. Thus, we come to

( x 1 + 2 x 2 + + 2014 x 2014 ) 2 x 1 2 + x 2 2 + + x 2014 2 n = 1 2014 n 2 x n 2 x n 2 = n = 1 2014 n 2 = 2725088015 \frac{(x_1 + 2x_2 + \cdots + 2014x_{2014})^2}{x_1^2 + x_2^2 + \cdots + x_{2014}^2} \leq \displaystyle \sum_{n = 1}^{2014} \frac{n^2 x_n^2}{x_n^2} = \displaystyle \sum_{n = 1}^{2014} n^2 = 2 725 088 015

Equality holds when they are linearly dependent, i.e. k n x n = x n 2 k*nx_n = x_n^2 , and taking k = 1 k = 1 we find easily x n = n x_n = n .

Mike Kong - 7 years, 5 months ago
Meet Udeshi
Dec 22, 2013

It is a simple application of the Cauchy-Shwarz Inequality .

( i = 1 i = 2014 i x i ) 2 i = 1 i = 2014 i 2 i = 1 i = 2014 x i 2 (\sum_{i=1}^{i=2014} ix_i)^2 \le \sum_{i=1}^{i=2014} i^2 \sum_{i=1}^{i=2014} x_i^2 This can be rearranged to the inequality given in the question and value of N N is N = i = 1 i = 2014 i 2 = 2014 × 2015 × 4029 6 = 2725088015 N=\sum_{i=1}^{i=2014} i^2=\frac{2014\times 2015\times 4029}{6}=2725088015 Thus sum of digits is 38

You forgot the C in Schwarz :)

Michael Tang - 7 years, 5 months ago
Akshaj Kadaveru
Dec 23, 2013

Notice by Cauchy-Schwarz, ( x 1 2 + x 2 2 + + x 2014 2 ) ( 1 2 + 2 2 + + 201 4 2 ) ( x 1 + 2 x 2 + + 2014 x 2014 ) 2 (x_1^2 + x_2^2 + \cdots + x_{2014}^2)(1^2 + 2^2 + \cdots + 2014^2) \ge (x_1 + 2x_2 + \cdots + 2014x_{2014})^2 Therefore, ( x 1 + 2 x 2 + + 2014 x 2014 ) 2 ( x 1 2 + x 2 2 + + x 2014 2 ) ( 1 2 + 2 2 + + 201 4 2 ) = 2014 2015 4029 6 = 2725088015 \begin{aligned}\dfrac{(x_1 + 2x_2 + \cdots + 2014x_{2014})^2}{(x_1^2 + x_2^2 + \cdots + x_{2014}^2)} &\le& (1^2 + 2^2 + \cdots + 2014^2) \\ &=& \dfrac{2014 \cdot 2015 \cdot 4029}{6}\\ &=& 2725088015\end{aligned} Equality occurs when x 1 = 1 , x 2 = 2 , , x 2014 = 2014 x_1 = 1, \ x_2 = 2, \ \cdots, x_{2014} = 2014 . Therefore, N = 2725088015 N = 2725088015 so our answer is 38 \boxed{38} . \blacksquare

Muhammad Shariq
Jan 28, 2014

By the look of this problem, it's pretty clear that it wants to be killed by the Cauchy-Schwarz Inequality. Letting { a i } i = 1 2014 = x i \{a_i\}_{i=1}^{2014}=x_i and { b i } i = 1 2014 = i \{b_i\}_{i=1}^{2014}=i , by Cauchy we get:

( x 1 + 2 x 2 + . . . + 2014 x 2014 ) 2 ( 1 2 + 2 2 + 3 2 + . . . + 201 4 2 ) ( x 1 2 + x 2 2 + . . . + x 2014 2 ) (x_1+2x_2+...+2014x_{2014})^2 \le (1^2+2^2+3^2+...+2014^2)(x_1^2+x_2^2+...+x_{2014}^2)

Rearranging yields:

( x 1 + 2 x 2 + . . . + 2014 x 2014 ) 2 x 1 2 + x 2 2 + . . . + x 2014 2 1 2 + 2 2 + 3 2 + . . . + 201 4 2 = N \frac{(x_1+2x_2+...+2014x_{2014})^2}{x_1^2+x_2^2+...+x_{2014}^2} \le 1^2+2^2+3^2+...+2014^2=N

At this point (actually at the very beginning of the problem lol =P), the solution becomes trivial...

N = 1 2 + 2 2 + 3 2 + . . . + 201 4 2 = ( 2 ( 2014 ) + 1 ) ( 2014 + 1 ) ( 2014 ) 6 = 272508815 N=1^2+2^2+3^2+...+2014^2=\frac{(2(2014)+1)(2014+1)(2014)}{6}=272508815 .

Then the sum of the digits is: 2 + 7 + 2 + 5 + 8 + 8 + 1 + 5 = 38 2+7+2+5+8+8+1+5=\boxed{38} .

Akash Gupta
Jan 23, 2014

as it is applicable to all real series let x1=1 therefore,it becomes {1^2+2^2+3^2...................+2014^2014}^2 divided by 1^2+2^2+3^2+4^2.......................+2014^2014 therefore the expression becomes1^2+2^2+3^2+4^2........2014^2014====which can be calculated by the formulae for the sum of squares of consecutive integers======n(n+1)(2n+1)/6.the sum of digits will come 38

Andres Saez
Dec 30, 2013

By Cauchy-Schwarz, we have [ i = 1 2014 i x i ] 2 i = 1 2014 i 2 i = 1 2014 x i 2 \left[\sum_{i=1}^{2014} i \cdot x_i\right]^2 \leq \sum_{i=1}^{2014} i^2 \cdot \sum_{i=1}^{2014} x_i^2 , which rearranges to yield [ i = 1 2014 i x i ] 2 i = 1 2014 x i 2 i = 1 2014 i 2 = 2725088015 \frac{\left[\sum_{i=1}^{2014} i \cdot x_i\right]^2}{\sum_{i=1}^{2014} x_i^2} \leq \sum_{i=1}^{2014} i^2 = 2725088015 .Hence, the sum of the digits of N N is 38 \boxed{38}

Vijay Krishna
Dec 24, 2013

By Cauchy Swartz inequality,

[ (x1)^2 + (x2)^2 + ......... + (x2014)^2 ] [ 1 +4 +9 +16 + ...........] ≥ [x1 + 2(x2) +.........+ 2014(x2014)]^2

So the given expression in the question in less than or equal to 1+4 +9 +........+2014^2 and we can get N by using sum of squares of first n naturals and then we can find the sum of digits as 38

1 pending report

Vote up reports you agree with

×

Problem Loading...

Note Loading...

Set Loading...