Help please! Medal and fan!
Assume that \[ 1a_1+2a_2+\cdots+na_n=1, \] where the a_j are real numbers. As a function of $n$, what is the minimum value of \[1a_1^2+2a_2^2+\cdots+na_n^2?\]
Sum of products => want to use Cauchy-Scharz ?
I know that I'm supposed to use Cauchy-Scharz, but I'm not sure how to apply it.
decompose \(ka_k^2\) in the right way... if you don't see right away, test them all! Try: \(ka_k^2 = ka_k \times a_k\).
I see, it's tricky. Gotta think :D
Well before I had $$( 1a_1 + 2a_2 + 3a_3 + \cdots + na_n )^2 \le (1+2+3+ \cdots + n)(a_1^2 + a_2^2 + a_3^2 + \cdots + a_n^2 ),$$ and I simplified to $$ 1/(1+2+3+ \cdots + n) \le (a_1^2 + a_2^2 + a_3^2 + \cdots + a_n^2)., would this way also result in the correct answer?
Hmm I don't know. (you forgot "to the power 2" on the RHS). The question itself is a hint. You want a lower bound, while usually CS gives you the upper bound. That means your expression is on the RHS of CS. You must find \(c_k\) and \(d_k\) such that \((c_k d_k)^2\) can be written as \(ka_k^2 \times \star\).
So that \((\sum_k c_k d_k)^2 \le (\sum_k \star_k^2) (\sum_k ka_k^2)\).
Ummm...I'm not sure I know how to do that.
Find the things backwards. CS: \((\sum_k |x_ky_k|)^2 \le (\sum_k x_k^2) (\sum y_k^2)\). Here, the sum \(\sum_k y_k^2\) has terms \(ka_k^2\). That means \(y_k = \sqrt{k}a_k\). Now find the "natural" complement \(x_k\) to this \(y_k\).
Sorry I probably seem kind of dumb but I still don't fully understand what I should do to solve the problem
In the resolution, there must be a guess. Turns out the natural guess works fine. 1. What I said above: if you want to find a lower bound for a sum, that means this sum is on the RHS of CS. 2. therefore the sum has terms "\(y_k^2\)". -> \(y_k^2\) 3. Find a "good" \(x_k\) and look at CS's inequality with this \(x_k,y_k\).
*2. "\(y_k^2\) -> \(y_k = \sqrt{k} \,a_k\).
If you multiply \(\sqrt{k}\,a_k\) by \(\sqrt{k}\), you obtain a term with a familiar face: \(ka_k\). Try to write CS with this choice of \(x_k,y_k\): \((x_k,y_k)=(\sqrt{k},\sqrt{k}\,a_k)\).
Okay and once I do that what do I do?
Write CS: \((\sum_k \sqrt k \sqrt k a_k)^2 \le (\sum_k (\sqrt k)^2)(\sum_k (\sqrt k a_k)^2)\)
This is almost the end.
Sorry it's just that I haven't been taught much about summations (except what they are called and what they look like).
ok i'll rewrite it
CS: \[ (\sqrt 1\sqrt1a_1 +\sqrt 2\sqrt 2a_2+\dotsb+\sqrt n \sqrt na_n)^2 \\ \le \biggl((\sqrt 1)^2 + (\sqrt 2)^2 + \dotsb + (\sqrt n)^2\biggr) \biggl((\sqrt1a_1)^2+(\sqrt2a_2)^2+\dotsb+(\sqrt na_n)^2\biggr) \] Compute what's inside each the parnetheses.
Wait how do I compute what is in the parentheses if "n" could be any number?
You have the assumption for the LeftHandSide, and there is a "well-known" formula: \(1+2+3+\dotsb+n = n(n+1)/2\).
Just a clarification for the LHS you are saying sqrt(n)*sqrt(n)*a_n not sqrt(n)*sqrt(n*a_n)? Right?
Yes. \(\sqrt{k} \times (\sqrt k \times a_k)\)
Do you just mean to simplify it: LHS=$$(1a_1+2a_2+...+na_n)^2$$
That's it. The assumption now says that \(1a_1+\dots+na_n = 1\). -> \(1^2=1\) :)
Wait so is that the answer or is there more?
@reemii
1 isn't the correct answer though...
This is the LHS. Now: CS: \(1^2 \le (1+2+\dotsb+n)\times (\text{your sum})\) i.e.: \(\frac{1}{n(n+1)} \le \text{your sum} \).
because 1+2+...+n = n(n+1)/2.
typo.. \(\frac{2}{n(n+1)} \le \text{your sum}\)
Wait but how do I find the minimum of the sum. The inequality suggests I have to find the maximum of $$\frac{2}{n(n+1)}$$
I thought you wanted to find a minimal for \(1a_1^2+2_a_2^2+\dotsb +na_n^2\).
\(1a_1^2+2 a_2^2+\dotsb +na_n^2\)
oh yeah
Wait but then using the inequality how do I find the answer?
CS has given you this: \( 1 \le \text{sum} \times \text{yoursum}\). -> \(\text{yoursum}\ge 1/\text{sum} = 1/\frac{n(n+1)}{2}\).
= 2/(n(n+1)).
You're done
Oh. Thanks so much!
Join our real-time social learning platform and learn together with your friends!