Inspired by Gurido Cuong

Algebra Level 5

3 a b + 1 a + b + 3 b c + 1 b + c + 3 a c + 1 a + c K a b c \large \frac{3ab+1}{a+b}+\frac{3bc+1}{b+c}+\frac{3ac+1}{a+c} \geq Kabc

If a , b a,b and c c are positive real numbers satisfying a b + b c + c a = 1 ab+bc+ca=1 , find the maximum K K for which the inequality holds.


Inspiration


The answer is 27.

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.

2 solutions

Zk Lin
Feb 11, 2016

We make the substitution x = a b , y = b c , z = a c x=ab, y=bc, z=ac .

Verify the following relationships:

x y = a b 2 c , y z = a b c 2 , x z = a 2 b c , x y z = a b c , x + y + z = 1 xy=ab^{2}c, yz=abc^{2}, xz=a^{2}bc, \sqrt{xyz}=abc, x+y+z=1

Therefore,

a + b = x ( y + z ) x y z a+b=\frac{x(y+z)}{\sqrt{xyz}}

b + c = y ( x + z ) x y z b+c=\frac{y(x+z)}{\sqrt{xyz}}

c + a = z ( x + y ) x y z c+a=\frac{z(x+y)}{\sqrt{xyz}}

Substitute this into our inequality:

3 a b + 1 a + b + 3 b c + 1 b + c + 3 a c + 1 a + c \frac{3ab+1}{a+b}+\frac{3bc+1}{b+c}+\frac{3ac+1}{a+c}

= x y z ( 3 x + 1 x ( y + z ) + 3 y + 1 y ( x + z ) + 3 z + 1 z ( x + y ) ) =\sqrt{xyz}(\frac{3x+1}{x(y+z)}+\frac{3y+1}{y(x+z)}+\frac{3z+1}{z(x+y)})

= x y z ( 3 x + 1 x ( 1 x ) + 3 y + 1 y ( 1 y ) + 3 z + 1 z ( 1 z ) ) =\sqrt{xyz}(\frac{3x+1}{x(1-x)}+\frac{3y+1}{y(1-y)}+\frac{3z+1}{z(1-z)})

Let's attempt to evaluate the minimum of the function f ( x ) = 3 x + 1 x ( 1 x ) f(x)=\frac{3x+1}{x(1-x)} .

The first derivative of the function is evaluated as f ( x ) = ( 3 x 1 ) ( x + 1 ) x 2 ( 1 x ) 2 f'(x)=\frac{(3x-1)(x+1)}{x^{2}{(1-x)}^{2}} . Setting f ( x ) = 0 f'(x)=0 , we find that x = 1 3 x=\frac{1}{3} or x = 1 x=-1 . We reject the latter since x x must be positive.

Consider the shape of the function when graphed. f ( x ) f(x)\rightarrow \infty as x 0 + x \rightarrow 0^{+} or x 1 x \rightarrow 1^{-} . Therefore, we conclude that f ( x ) f(x) does not have a maximum, so the turning point we evaluate must be a minimum instead. This saves us the hassle of evaluating the second derivative.

Therefore, we conclude that the minimum of the function is 9 9 , attained when x = 1 3 x=\frac{1}{3} . Repeating the same thing to the function in terms of y y and z z gives us the same results. We find that x = y = z = 1 3 x=y=z=\frac{1}{3} if equality is to hold. Note also that x + y + z = 1 x+y+z=1 , so the constraint of the problem is fulfilled.

Therefore, we have

x y z ( 3 x + 1 x ( 1 x ) + 3 y + 1 y ( 1 y ) + 3 z + 1 z ( 1 z ) ) \sqrt{xyz}(\frac{3x+1}{x(1-x)}+\frac{3y+1}{y(1-y)}+\frac{3z+1}{z(1-z)})

x y z ( 9 + 9 + 9 ) \geq \sqrt{xyz}(9+9+9)

= 27 a b c = 27abc , hence the answer.

Equality holds when x = y = z = 1 3 x=y=z=\frac{1}{3} , implying a = b = c = 1 3 a=b=c=\frac{1}{\sqrt{3}}


This problem is motivated by the process of solving This will give you a run for your money by Gurido Cuong, which is an excellent problem. I play around with the conditions and come up with the substitution above.

Moderator note:

The proper way to deal with the later half of the solution is to apply Jensen's to f ( x ) = 3 x + 1 x ( 1 x ) f(x) = \frac{ 3x+1 } { x ( 1 - x ) } , since the second derivative is positive in the interval 0 x 1 0 \leq x \leq 1 . Then, f ( x ) + f ( y ) + f ( z ) 3 f ( x + y + z 3 ) = 3 f ( 1 3 ) f(x) + f(y) + f(z) \geq 3 f ( \frac{x + y + z } { 3} ) = 3 f ( \frac{1}{3} ) as desired.

The benefit of this approach is that we are not at the mercy of "We find that x = y = z = 1 3 x=y=z=\frac{1}{3} if equality is to hold. Note also that x + y + z = 1 x+y+z=1 , so the constraint of the problem is fulfilled.". IE, this will work even if x + y + z = 0.5 x + y + z = 0.5 .

There seems to be a problem when You deal with the largest A .

Since 0<x<1 ,it's unreasonable to just simply make A satisfy (3-A)²-4A≤0 .(If so, x ∈R)

What's more, although the largest A is picked when x=1/3 which exactly makes sense it's still unreasonable to conclude that the largest A is 9. Because(3-A)²-4A≤0 is the unnecessary and sufficient condition. If A>9,there still are chances that ョA>9 satisfies the condition. However,in this question ,it is impossible to find an A>9 that makes sense.

John Swift - 5 years, 4 months ago

Log in to reply

That is a very sharp observation. You are correct, indeed, ( 3 A ) 2 4 A 0 (3-A)^{2}-4A\leq0 is sufficient but unnecessary due to the range, guess I neglect that because I was too carried away writing the solution. I suppose we will have to revert to calculus then to find the minimum.

EDIT: I have fixed the solution. If you still find any discrepancies, please let me know. Thanks!

ZK LIn - 5 years, 4 months ago

Log in to reply

It is perfect now I suppose .

John Swift - 5 years, 4 months ago

The proper way to deal with the later half of the solution is to apply Jensen's to f ( x ) = 3 x + 1 x ( 1 x ) f(x) = \frac{ 3x+1 } { x ( 1 - x ) } , since the second derivative is positive in the interval 0 x 1 0 \leq x \leq 1 . Then, f ( x ) + f ( y ) + f ( z ) 3 f ( x + y + z 3 ) = 3 f ( 1 3 ) f(x) + f(y) + f(z) \geq 3 f ( \frac{x + y + z } { 3} ) = 3 f ( \frac{1}{3} ) as desired.

The benefit of this approach is that we are not at the mercy of "We find that x = y = z = 1 3 x=y=z=\frac{1}{3} if equality is to hold. Note also that x + y + z = 1 x+y+z=1 , so the constraint of the problem is fulfilled.". IE, this will work even if x + y + z = 0.5 x + y + z = 0.5 .

Calvin Lin Staff - 5 years, 3 months ago

Log in to reply

Indeed. Jensen's inequality is an excellent option especially when x + y + z x+y+z conveniently equals 1 1 . Thanks for pointing this out!

ZK LIn - 5 years, 3 months ago
John Swift
Feb 13, 2016

∵( 3 a b + 1 a + b \frac{3ab+1}{a+b} + 3 b c + 1 b + c \frac{3bc+1}{b+c} + 3 a c + 1 a + c \frac{3ac+1}{a+c} )[(a+b)(3ab+1)+(b+c)(3bc+1)+(a+c)(3ac+1)]≥(3ab+3bc+3ac+3)²=36 ( Cauchy inequality)

(we know ab+bc+ac=1 but 3ab+1,3ac+1,3bc+1 are numerators which makes it difficult for them to interact.In order to break the "barriers" Cauchy inequality is a good choice. I still don't know whether this will work but it's not a bad idea to have a try.)

3 a b + 1 a + b \frac{3ab+1}{a+b} + 3 b c + 1 b + c \frac{3bc+1}{b+c} + 3 a c + 1 a + c \frac{3ac+1}{a+c} 36 ( a + b ) ( 3 a b + 1 ) + ( b + c ) ( 3 b c + 1 ) + ( a + c ) ( 3 a c + 1 ) \frac{36}{(a+b)(3ab+1)+(b+c)(3bc+1)+(a+c)(3ac+1)} …….(*)

(it seems like a monster isn't it? Maybe you'll choose to give up this kind of solution and it's exactly what I thought at first. it's ok .There do be the chances that it's nonsense and it's a tough decision .So it's the atmosphere that I use to decide this kind of decision.In this question,what I've done at first makes "a b c" get together which provides the chances for them to interact.Moreover, the symmetry still exists during the "operation".All above provide a sign of success.So it is worth keeping going.

How to deal with it is the next problem. The fewer structures the beter.Because it makes the problem simple to operate generally. Inspired by this ,I tried to get rid of a+b b+c a+c and replaced them with ab bc ac . )

∵ab+bc+ca=1 ∴a+b= 1 a b c \frac{1-ab}{c} ∴(a+b)(3ab+1)= ( 1 a b ) ( 3 a b + 1 ) c \frac{(1-ab)(3ab+1)}{c}

(after this ,it's obvious that we can apply Gn-An to this to attain its maximum. the direction of the inequality is still right. though we still have no idea what will happen it's not bad to have a try)

∵(1-ab)(3ab+1)= 1 3 \frac{1}{3} (3-3ab)(1+3ab)≤ 1 3 \frac{1}{3} ( 3 3 a b + 1 + 3 a b 2 \frac{3-3ab+1+3ab}{2} )²= 4 3 \frac{4}{3}

Repeating the same thing to the b+c a+c, we can conclude that (*)≥ 36 3 4 \frac{3}{4} ( 1 c \frac{1}{c} + 1 a \frac{1}{a} + 1 b \frac{1}{b} ) ^(-1)=27abc

(after doing the same thing it's such a coincidence that ( 1 c \frac{1}{c} + 1 a \frac{1}{a} + 1 b \frac{1}{b} )^(-1) could be abc .How would You know it that If You didn't try )

Equality holds when a=b=c= (1/3)½

hence the largest K is 27

(as stated above it is good to try.If you don't try it's difficult to find the magic behind it .But not All thoughts are worth trying If Time limited.At that Time the atmosphere really helps. )

This is a very nice solution. Do you mind typesetting it properly (putting all the relevant Math parts in LATEX) so others can learn from it more easily?

ZK LIn - 5 years, 4 months ago

Log in to reply

sorry for the inconvenience,I've tried and probably this is the best I can do ....QAQ

I've not got accustomed to using this software.And to my great surprise and happiness,my poor English could make sense to others ,which is really beyond my expectations.(really :) )

maybe I can expand on my account more clearly to make up for it .I'll have a try .

John Swift - 5 years, 4 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...