Page 1 of 2

An Angle Bisector and the two sides...

Posted: Fri Sep 09, 2011 11:16 pm
by Labib
In the triangle $\triangle ABC$, let $AU$ be the angle bisector of the angle $ \angle BAC$ with $U$ lying on the side $BC$.
Prove that,
$AU \leq \frac{1}{2}(AB+AC).$

Re: An Angle Bisector and the two sides...

Posted: Fri Sep 09, 2011 11:40 pm
by Labib
Here's how I got going...
If I can prove, $4{AU}^2 \leq (AB+AC)^2$, it would be enough.
Let $AU=p$, $BC=a$, $AB=c$, $AC=b$, $BU=m$, $CU=n$.
With the help of Stewart's theorem, I can deduce-
${p}^2=bc-mn$
$\Rightarrow 4p^2=4bc-4mn$...............(i)
Again,
$(b+c)^2=b^2+c^2+2bc$...........(ii)
As, $b^2+c^2 \geq 2bc$
$\Rightarrow b^2+c^2+2bc > 4bc-4mn$
So from (i) and (ii), we get-
$(b+c)^2 > 4p^2$
$\Rightarrow 4AU^2 < (AB+AC)^2$
$\Rightarrow 2AU < (AB+AC)$
$\Rightarrow AU < \frac{1}{2}(AB+AC)$
Then where's the equality part??
Where's the problem??

Re: An Angle Bisector and the two sides...

Posted: Fri Sep 09, 2011 11:57 pm
by sourav das
Step 1: Prove that $AU^2 = bc(1 - \frac{a^2}{(b+c)^2}) $ using Stewart's theorem and ratio theorem of angle bisector. Next substitute $a= x + y, b= y + z, c= z + x$ and find out that
$AU^2 = \frac {4y(x+y)(y+z)(x+y+z)}{(x+2y+z)^2}$. Now $(x + y) + ( y + z) \geq 2 \sqrt{(x+y)(y+z)}$ and
$( x + y + z) + y \geq 2 \sqrt{y(x+y+z)}$
And it will lead you to,
$\frac {b+c}{2} = \frac {x+2y+z}{2} \geq AU$

Stewart's theorem :http://en.wikipedia.org/wiki/Stewart%27s_theorem

Re: An Angle Bisector and the two sides...

Posted: Sat Sep 10, 2011 12:03 am
by Labib
Sourav, please check my solution for any possible mistakes!! :'(

Re: An Angle Bisector and the two sides...

Posted: Sat Sep 10, 2011 12:17 am
by *Mahi*
I was about to post the same solution Sourav Das posted, so I'd rather not waste my time writing the same solution again.
@Labib:
You said $b^2+c^2>2bc$
But when it's $b=c$? ;)
And Stewert's Theorem: http://en.wikipedia.org/wiki/Stewart%27s_theorem

Re: An Angle Bisector and the two sides...

Posted: Sat Sep 10, 2011 12:30 am
by Labib
Mahi...
It was a typing mistake of mine...
I've edited it...
Now... If you could re-check... ;)

Re: An Angle Bisector and the two sides...

Posted: Sat Sep 10, 2011 1:01 am
by *Mahi*
Shit! I should have seen this earlier! The equality case never occurs other than degenerate cases, that's why you never found that part. I'm using Sourav's solution for this fact.
sourav das wrote:Step 1: Prove that $AU^2 = bc(1 - \frac{a^2}{(b+c)^2}) $ using Stewart's theorem and ratio theorem of angle bisector. Next substitute $a= x + y, b= y + z, c= z + x$ and find out that
$AU^2 = \frac {4y(x+y)(y+z)(x+y+z)}{(x+2y+z)^2}$. Now $(x + y) + ( y + z) \geq 2 \sqrt{(x+y)(y+z)}$ and
$( x + y + z) + y \geq 2 \sqrt{y(x+y+z)}$
And it will lead you to,
$\frac {b+c}{2} = \frac {x+2y+z}{2} \geq AU$
But the equality part will only hold if and only if $(x + y) + ( y + z) = 2 \sqrt{(x+y)(y+z)}$ and $( x + y + z) + y = 2 \sqrt{y(x+y+z)}$ happens together, that is both $x + y= y + z$ and $ x + y + z= y$ happens together, i.e. the degenerate case, and so ,that solves it, doesn't it? ;)

Again then in your solution, $4mn$ will be $0$, so the equality will hold...$\text{sigh}$

Re: An Angle Bisector and the two sides...

Posted: Sat Sep 10, 2011 8:26 am
by bristy1588
Now, the only problem is that the original problem was:
2AU>= AB+BC.
What modification is needed to solve this?

Re: An Angle Bisector and the two sides...

Posted: Sat Sep 10, 2011 8:27 am
by bristy1588
Now, the only problem is that the original problem was:
2AU>= AB+BC.
What modification is needed to solve this?

Re: An Angle Bisector and the two sides...

Posted: Sun Sep 11, 2011 12:26 am
by *Mahi*
WTH ?! $AU \geq \frac 1 2 (AB+BC)??$ It's not possible.
Try the simplest counter-example, a isosceles triangle.