MHB Can high school math prove this inequality?

AI Thread Summary
The discussion centers on proving the inequality involving Euclidean vectors using high school mathematics. Participants express skepticism about the feasibility of such a proof without invoking advanced concepts like summation or the Cauchy-Schwarz inequality. The reverse triangle inequality is mentioned as a potential approach, but there is uncertainty about providing a purely high school-level proof. Overall, the conversation highlights the challenge of proving the inequality with limited mathematical tools. The thread concludes with anticipation for the original poster's solution.
solakis1
Messages
407
Reaction score
0
Using high school mathematics prove the following inequality:

$$\sqrt{a_{1}^2+...+a_{n}^2}\leq\sqrt{(a_{1}-b_{1})^2+...+(a_{n}-b_{n})^2}+\sqrt{b_{1}^2+...+b_{n}^2}$$
 
Last edited:
Mathematics news on Phys.org
My attempt:
Squaring both sides:

\[\sum_{i}a_i^2\leq \sum_{i}(a_i-b_i)^2+\sum_{i}b_i^2+2\sqrt{\left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}(a_i-b_i)^2\right )} \\\\ =\sum_{i}a_i^2+2\sum_{i}b_i^2-2\sum_{i}a_ib_i+2\sqrt{\left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}(a_i-b_i)^2\right )} \\\\ \Rightarrow \sum_{i}a_ib_i -\sum_{i}b_i^2\leq \sqrt{\left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}(a_i-b_i)^2\right )}\]

Squaring again:

\[\left ( \sum_{i}a_ib_i \right )^2+\left ( \sum_{i}b_i^2 \right )^2-2\left ( \sum_{i}a_ib_i \right )\left ( \sum_{i}b_i^2 \right )\leq \left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}(a_i^2+b_i^2-2a_ib_i) \right ) \\\\ =\left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}a_i^2 \right )+\left ( \sum_{i}b_i^2 \right )^2-2\left ( \sum_{i}a_ib_i \right )\left ( \sum_{i}b_i^2 \right ) \\\\ \Rightarrow \left ( \sum_{i}a_ib_i \right )^2\leq \left ( \sum_{i}a_i^2 \right )\left ( \sum_{i}b_i^2 \right )\]

This statement is the Cauchy-Schwarz inequality. Thus the stated inequality holds.
 
lfdahl said:
My attempt:
Squaring both sides:

\[\sum_{i}a_i^2\leq \sum_{i}(a_i-b_i)^2+\sum_{i}b_i^2+2\sqrt{\left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}(a_i-b_i)^2\right )} \\\\ =\sum_{i}a_i^2+2\sum_{i}b_i^2-2\sum_{i}a_ib_i+2\sqrt{\left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}(a_i-b_i)^2\right )} \\\\ \Rightarrow \sum_{i}a_ib_i -\sum_{i}b_i^2\leq \sqrt{\left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}(a_i-b_i)^2\right )}\]

Squaring again:

\[\left ( \sum_{i}a_ib_i \right )^2+\left ( \sum_{i}b_i^2 \right )^2-2\left ( \sum_{i}a_ib_i \right )\left ( \sum_{i}b_i^2 \right )\leq \left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}(a_i^2+b_i^2-2a_ib_i) \right ) \\\\ =\left ( \sum_{i}b_i^2 \right )\left ( \sum_{i}a_i^2 \right )+\left ( \sum_{i}b_i^2 \right )^2-2\left ( \sum_{i}a_ib_i \right )\left ( \sum_{i}b_i^2 \right ) \\\\ \Rightarrow \left ( \sum_{i}a_ib_i \right )^2\leq \left ( \sum_{i}a_i^2 \right )\left ( \sum_{i}b_i^2 \right )\]

This statement is the Cauchy-Schwarz inequality. Thus the stated inequality holds.
Yes,but summation and the Cauchy-Schwarz inequality are not high school mathematics.

How would you prove the CSB inequality using high school mathematics
 
If we let $a$ and $b$ be $n$ dimensional (euclidean) vectors, then your statement follows directly from the reverse triangle inequality, which is a high-school concept. Now if you're asking for the proof of the triangle inequality, since I'm lazy I'll just say it is geometrically intuitive.
 
solakis said:
Yes,but summation and the Cauchy-Schwarz inequality are not high school mathematics.

How would you prove the CSB inequality using high school mathematics

Good question! I´m afraid I cannot come up with a real high school solution then ...:(
 
lfdahl said:
Good question! I´m afraid I cannot come up with a real high school solution then ...:(

It will be interesting to see the OP's solution. :)
 
solakis said:
Using high school mathematics prove the following inequality:

$$\sqrt{a_{1}^2+...+a_{n}^2}\leq\sqrt{(a_{1}-b_{1})^2+...+(a_{n}-b_{n})^2}+\sqrt{b_{1}^2+...+b_{n}^2}$$
my solution:
using inductive method
$\sqrt{a^2_1}=\left | a_1 \right |=\left | a_1-b_1+b_1 \right |\leq \left | a_1-b_1\right |+\left |b_1 \right |
=\sqrt{(a_1-b_1)^2}+\sqrt{(b_1)^2}$

n=1 is true
suppose n=n is true
$$\sqrt{a_{1}^2+...+a_{n}^2}\leq\sqrt{(a_{1}-b_{1})^2+...+(a_{n}-b_{n})^2}+\sqrt{b_{1}^2+...+b_{n}^2}$$
let $P={a_{1}^2+...+a_{n}^2},Q={(a_{1}^2-b_1)^2+...+(a_{n}-b_n)^2},R={b_{1}^2+...+b_{n}^2}$
that is $\sqrt P\leq \sqrt Q+\sqrt R$
for n=n+1
we will prove :$\sqrt {P+a^2_{n+1}}\leq\sqrt {Q+(a_{n+1}-b_{n+1})^2}+\sqrt {R+b^2_{n+1}}$
with the help of the following diagram it is easy to get the proof
here CF=$\sqrt {P+a^2_{n+1}}-\sqrt P$
BD=CE=$\sqrt {Q+(a_{n+1}-b_{n+1})^2}-\sqrt Q$
and BD//CE
BC=DE
and BC//DE
EF=$\sqrt {R+b^2_{n+1}}-\sqrt R$
that is $AF\leq AD+DF$
View attachment 6614
IF point D between A and F then equality will hold
 

Attachments

  • inequality 3.jpg
    inequality 3.jpg
    4.5 KB · Views: 106
Last edited:
MarkFL said:
It will be interesting to see the OP's solution. :)
$$\sqrt{a_{1}^2+...+a_{n}^2}\leq\sqrt{(a_{1}-b_{1})^2+...+(a_{n}-b_{n})^2}+\sqrt{b_{1}^2+...+b_{n}^2}\Longleftrightarrow\sqrt{a_{1}^2+...+a_{n}^2}-\sqrt{b_{1}^2+...+b_{n}^2}\leq\sqrt{(a_{1}-b_{1})^2+...+(a_{n}-b_{n})^2}$$ .................1
And squaring and then cancelling equal terms we end up with:

$$\sqrt{(a_{1}^2+...a_{n}^2)(b_{1}^2+...b_{n}^2)}\geq(a_{1}b_{1}+...a_{n}b_{n})\Longleftrightarrow(a_{1}^2+...a_{n}^2)(b_{1}^2+...b_{n}^2)\geq(a_{1}b_{1}+...a_{n}b_{n})^2$$............2

Put :
$$A= a_{1}^2+...+a_{n}^2$$,

$$B= a_{1}b_{1}+...a_{n}b_{n}$$

$$C= b_{1}^2+...+b_{n}^2$$

Note: A and C are not zero

And (2) becomes:

$$AC\geq B^2\Longleftrightarrow AC-B^2\geq 0\Longleftrightarrow\frac{AC-B^2}{A}\geq 0\Longleftrightarrow A(x+\frac{B}{A})^2+\frac{AC-B^2}{A}\geq 0\Longleftrightarrow Ax^2+2Bx+C\geq 0\Longleftrightarrow (a_{1}^2+...a_{n}^2)x^2+2(a_{1}b_{1}+...a_{n}b_{n})x+ (b_{1}^2+...b_{n}^2)\geq 0$$$$\Longleftrightarrow (a_{1}x+b_{1})^2+...(a_{n}x+b_{n})^2\geq 0$$ which is true hence the intial inequality which is equivalent to the final one is also true

NOTE: $$\frac{AC-B^2}{A}\geq 0\Longrightarrow A(x+\frac{B}{A})^2+\frac{AC-B^2}{A}\geq 0$$ FOR ALL x's hence for x=-(B/A) WE have : $$ A(x+\frac{B}{A})^2+\frac{AC-B^2}{A}\geq 0\Longrightarrow\frac{AC-B^2}{A}\geq 0$$
 
Last edited:
Back
Top