A Triangle Inequality

Algebra Level 5

Let a triangle with sides of length a , b , c a,b,c have perimeter 2 2 . What is the maximum value of k k such that 1 a b + 1 b c + 1 c a k \dfrac{1-a}{b}+\dfrac{1-b}{c}+\dfrac{1-c}{a}\ge k is always true? Prove your claim.


The answer is 1.000.

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.

4 solutions

Jon Haussmann
May 29, 2014

The best bound is actually given by 1 a b + 1 b c + 1 c a > 1. \frac{1 - a}{b} + \frac{1 - b}{c} + \frac{1 - c}{a} > 1.

Since a a , b b , and c c are the sides of a triangle, there exist positive real numbers such that a = y + z a = y + z , b = x + z b = x + z , and c = x + y c = x + y . Then 2 x + 2 y + 2 z = a + b + c = 2 2x + 2y + 2z = a + b + c = 2 , so x + y + z = 1 x + y + z = 1 , and 1 a b + 1 b c + 1 c a = 1 y z x + z + 1 x z x + y + 1 x y y + z = x z + x + y x + y + z y + z . \begin{aligned} \frac{1 - a}{b} + \frac{1 - b}{c} + \frac{1 - c}{a} &= \frac{1 - y - z}{x + z} + \frac{1 - x - z}{x + y} + \frac{1 - x - y}{y + z} \\ &= \frac{x}{z + x} + \frac{y}{x + y} + \frac{z}{y + z}. \end{aligned} Note that this expression is homogeneous in x x , y y , and z z , so we can discard the condition x + y + z = 1 x + y + z = 1 .

We see that x z + x + y x + y + z y + z > x x + y + z + y x + y + z + z x + y + z = 1. \frac{x}{z + x} + \frac{y}{x + y} + \frac{z}{y + z} > \frac{x}{x + y + z} + \frac{y}{x + y + z} + \frac{z}{x + y + z} = 1.

Now, let x = 1 x = 1 , y = ϵ y = \epsilon , and z = ϵ 2 z = \epsilon^2 , where ϵ \epsilon is a positive real number. As ϵ 0 + \epsilon \to 0^+ , x z + x + y x + y + z y + z = 1 ϵ 2 + 1 + ϵ 1 + ϵ + ϵ 2 ϵ + ϵ 2 = 1 ϵ 2 + 1 + ϵ 1 + ϵ + ϵ 1 + ϵ 1 , \begin{aligned} \frac{x}{z + x} + \frac{y}{x + y} + \frac{z}{y + z} &= \frac{1}{\epsilon^2 + 1} + \frac{\epsilon}{1 + \epsilon} + \frac{\epsilon^2}{\epsilon + \epsilon^2} \\ &= \frac{1}{\epsilon^2 + 1} + \frac{\epsilon}{1 + \epsilon} + \frac{\epsilon}{1 + \epsilon} \\ &\to 1, \\ \end{aligned} so the bound of 1 cannot be replaced by a larger number.

But how do you know that the minimum of the expression occurs at x=1??

Arvind Chander - 7 years ago

Log in to reply

I don't understand your comment; I never claim that the minimum occurs only when x = 1 x = 1 .

If you are asking why I set x = 1 x = 1 to show that the expression can come arbitrarily close to 1, it's just a nice value to work with. I could have set x x to any value, and the argument would still work.

Jon Haussmann - 7 years ago

Sir how do approached for such a nice solution?

Department 8 - 5 years, 8 months ago
Daniel Liu
May 28, 2014

IGNORE THIS SOLUTION. IT IS INCORRECT.LOOK AT JON HAUSSMAN'S SOLUTION.


Because a , b , c a,b,c are the side lengths of the triangle, we use the well-known substitution a = x + y a=x+y , b = y + z b=y+z , and c = z + x c=z+x (draw a triangle and its incircle to understand why this substitution works).

Thus, we have 2 x + 2 y + 2 z = 2 2x+2y+2z=2 or x + y + z = 1 x+y+z=1 for positive reals x , y , z x,y,z . Also, we have c y c 1 a b = c y c 1 x y y + z = c y c z y + z \displaystyle\sum_{cyc}\dfrac{1-a}{b}=\sum_{cyc}\dfrac{1-x-y}{y+z}=\sum_{cyc}\dfrac{z}{y+z} . Thus we just need to find the minimum of c y c y x + y \sum_{cyc}\dfrac{y}{x+y}

By Cauchy, we have ( c y c y x + y ) ( c y c y ( x + y ) ) ( x + y + z ) 2 = 1 \left(\sum_{cyc}\dfrac{y}{x+y}\right)\left(\sum_{cyc}y(x+y)\right)\ge (x+y+z)^2=1

Thus, c y c y x + y 1 c y c y ( x + y ) = 1 x 2 + y 2 + z 2 + x y + y z + z x \sum_{cyc}\dfrac{y}{x+y}\ge \dfrac{1}{\sum\limits_{cyc}y(x+y)}=\dfrac{1}{x^2+y^2+z^2+xy+yz+zx}

But x 2 + y 2 + z 2 + x y + y z + z x = ( x + y + z ) 2 ( x y + y z + z x ) = 1 ( x y + y z + z x ) x^2+y^2+z^2+xy+yz+zx=(x+y+z)^2-(xy+yz+zx)=1-(xy+yz+zx) so 1 x 2 + y 2 + z 2 + x y + y z + z x = 1 1 ( x y + y z + z x ) \dfrac{1}{x^2+y^2+z^2+xy+yz+zx}=\dfrac{1}{1-(xy+yz+zx)}

But x y + y z + z x ( x + y + z ) 2 3 xy+yz+zx\le \dfrac{(x+y+z)^2}{3} because when expanded and rearranged it becomes x 2 + y 2 + z 2 x y y z z x 0 x^2+y^2+z^2-xy-yz-zx\ge 0 which factors into 1 2 ( ( x y ) 2 + ( y z ) 2 + ( z x ) 2 ) 0 \dfrac{1}{2}((x-y)^2+(y-z)^2+(z-x)^2)\ge 0 which is always true. Thus, x y + y z + z x 1 3 xy+yz+zx\le \dfrac{1}{3} so 1 1 ( x y + y z + z x ) 1 1 1 3 = 3 2 \dfrac{1}{1-(xy+yz+zx)}\ge \dfrac{1}{1-\dfrac{1}{3}}=\dfrac{3}{2}

Thus, 1 a b + 1 b c + 1 c a 3 2 = 1.5 \dfrac{1-a}{b}+\dfrac{1-b}{c}+\dfrac{1-c}{a}\ge \dfrac{3}{2}=\boxed{1.5} and we are done.

Although I successfully guessed "correct" answer, unfortunately 1.5 is a wrong answer. It is easy to come up with a counterexample (actually any successful example will lead to a counter one as minimum value + maximum value is 3). For example, choose a=0.8, b=0.9 and c=0.3. You can calculate the value of the expression is 103/72, which is less than 1.5.

It is easy to find out where your logic is wrong. I leave it as an exercise to you:-).

I do not know correct answer at this moment, though.

Roger Lu - 7 years ago

Log in to reply

Indeed, since

( 1 a b + 1 b c + 1 c a ) + ( 1 a c + 1 b a + 1 a b ) = 3 ( \frac{ 1-a} { b} + \frac{1-b}{c} + \frac{1-c}{a} ) + ( \frac{1-a}{c} + \frac{ 1-b}{a} + \frac{ 1-a} { b} ) = 3

this tells us that there is at least one ordering of the lengths which produces an expression that is at most 3 2 \frac{3}{2} .

Calvin Lin Staff - 7 years ago

Log in to reply

Could anyone tell me what the problem with Daniel's solution is? Where is his logic wrong?

mathh mathh - 6 years, 12 months ago

I'd like to know how many people legitimately solved the question and how many people just assumed a = b = c a=b=c .

See, this is the one thing I don't like about making inequality problems on Brilliant. One can just assume a = b = c a=b=c ; and for me it is too hard to create a nice inequality without that as the equality case.

Daniel Liu - 7 years ago

Log in to reply

That happens, you should get used to it. Most of the people who solve my geometry problems usually get some crucial observations from the accurate diagrams. :P

But anyways, you shouldn't care. If someone guessed a problem and got it right, he cheated himself from the joy of solving that problem.

I liked this question because the solution set isn't a = b = c a = b =c . Your logic is wrong, and the correct value of k k is 1 (which will not be achieved).

Calvin Lin Staff - 7 years ago

Log in to reply

Thanks. Yes, correct value of k is 1. It is a good problem.

Roger Lu - 7 years ago

Do we really need the condition that a , b , c a,b,c are sides of a triangle? Note that you never used the fact that x , y , z x,y,z are positive.

I saw one of the tags named 'Nesbitt's Inequality'. Can you please explain it by posting a note on the site?:)

Samarpit Swain - 7 years ago

Log in to reply

It states that given any three positive reals a , b , c , a,b,c, a b + c + b c + a + c a + b 3 2 . \dfrac{a}{b+c} + \dfrac{b}{c+a} + \dfrac{c}{a+b} \geq \dfrac{3}{2}.

Log in to reply

Thanks a lot!!!

Samarpit Swain - 7 years ago

Isn't this what the question is asking?

Avineil Jain - 7 years ago

Log in to reply

@Avineil Jain Nope. Read the question more carefully.

Log in to reply

@Sreejato Bhattacharya Got it. Thanks!

Avineil Jain - 7 years ago

We can also do it by AM,GM and HM inequality.i

Prateek Bontha - 7 years ago

Log in to reply

Care to elaborate?

this is the correct answer ..... my ans is also 1.5

Owendrilla Chowdhury - 7 years ago

I found 1 too.

Guilherme Silva - 7 years ago

A Note, The substitution used here that a = x + y a=x+y and similar things is called Ravi substitution. Also
@Daniel Liu , If only u wanted to get greater than 3 2 \frac{3}{2} , then u could also use AM-HM (Hint:let 1 = s 1=s i.e. semiperimeter. Then write all 1 b a = s a b a \frac{1-b}{a}=\frac{s}{a}-\frac{b}{a} , Move the semiperimeter terms in one side, the remaining side cancels out by AM-GM, then put semiperimeter common, apply AM-HM, and u get result 1.5 Actually, I got it wrong forst time, But when I used Ravi Substitution method, I got answer as 1

Dinesh Chavan - 7 years ago
Aaaaa Bbbbb
May 30, 2014

It is not reduced generalization, assume that: a b c , k = 1 a b + 1 b c + 1 c a a \geq b \geq c, k=\frac{1-a}{b}+\frac{1-b}{c}+\frac{1-c}{a} . Having: c 2 b c ( b + a ( 1 a ) + k a b ) a b ( 1 b ) = 0 \Leftrightarrow c^2b-c(b+a(1-a)+kab)-ab(1-b)=0 d = ( b + a ( 1 a ) + k a b ) 2 + 4 a b 2 ( 1 b ) \Rightarrow d=(b+a(1-a)+kab)^2+4ab^2(1-b) 0 ( b + a ( 1 a ) + k a b ) 2 + 4 a b 2 ( 1 b ) + ( b + a ( 1 a ) + k a b ) 2 b 1 \Rightarrow 0 \leq \frac{\sqrt{(b+a(1-a)+kab)^2+4ab^2(1-b)}+(b+a(1-a)+kab)}{2b} \leq 1 ( b + a ( 1 a ) + k a b ) 2 + 4 a b 2 ( 1 b ) b ( a ( 1 a ) + k a b ) \Rightarrow \sqrt{(b+a(1-a)+kab)^2+4ab^2(1-b)} \leq b-(a(1-a)+kab) b ( 1 k a ) > 0 k < 1 a \Rightarrow b(1-ka) \gt 0 \Rightarrow k<\frac{1}{a} When a 1 k 1 lim a 1 k = 1 a \rightarrow 1\Rightarrow k \rightarrow 1\Leftrightarrow \lim_{a \rightarrow 1}{k} = 1 k m a x = 1 k_{max}=\boxed{1}

As in Daniel's solution, we make the substitution a = x + y , b = y + z , c = z + x , a= x+y, b= y+z, c= z+x, with the reverse substitution given by x = b + c a 2 , y = c + a b 2 , z = a + b c 2 . x= \dfrac{b+c-a}{2}, y= \dfrac{c+a-b}{2}, z= \dfrac{a+b-c}{2}. The given condition transforms to 2 ( x + y + z ) = 2 x + y + z = 1 , 2 (x+y+z) = 2 \implies x+y+z= 1, and hence 1 a b + 1 b c + 1 c a = x z + x + y x + y + z y + z . \dfrac{1-a}{b} + \dfrac{1-b}{c} + \dfrac{1-c}{a} = \dfrac{x}{z+x} + \dfrac{y}{x+y} + \dfrac{z}{y+z}. Since the inequality is cyclic, WLOG assume x y z , x \geq y \geq z, so 1 y + z 1 z + x 1 x + y . \dfrac{1}{y+z} \geq \dfrac{1}{z+x} \geq \dfrac{1}{x+y}. By rearrangement, x z + x + y x + y + z y + z x x + y + y y + z + z z + x . \dfrac{x}{z+x} + \dfrac{y}{x+y} + \dfrac{z}{y+z} \geq \dfrac{x}{x+y} + \dfrac{y}{y+z} + \dfrac{z}{z+x}.

Note that ( x z + x + y x + y + z y + z ) + ( z z + x + x x + y + y y + z ) = x + z x + z + y + z y + z + z + x z + x = 3 , \left( \dfrac{x}{z+x} + \dfrac{y}{x+y} + \dfrac{z}{y+z} \right) + \left( \dfrac{z}{z+x} + \dfrac{x}{x+y} + \dfrac{y}{y+z} \right) = \dfrac{x+z}{x+z} + \dfrac{y+z}{y+z} + \dfrac{z+x}{z+x} = 3, so it follows that x z + x + y x + y + z y + z 3 2 . \dfrac{x}{z+x} + \dfrac{y}{x+y} + \dfrac{z}{y+z} \geq \boxed{ \dfrac{3}{2} }.

Equality holds for a = b = c = 2 3 . a=b=c= \dfrac{2}{3}.

Rearrangement doesn't work that way. The minimum is actually 1.

Calvin Lin Staff - 7 years ago

Log in to reply

Oh yeah, right. I messed up. So sorry for the mistake.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...