Pretty simple, maybe?

Algebra Level 4

Let x , y > 0 x, y>0 be different real numbers satisfying x y = y x {x}^{y} = {y}^{x} .

Find the infimum value of x + y x+y .

Give your answer to 3 decimal places.


The answer is 5.437.

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.

3 solutions

Guilherme Niedu
Jan 2, 2020

Without loss of generality, let y = t x y = tx , where t t is a real number. Then:

x t x = ( t x ) x \large \displaystyle x^{tx} = (tx)^x

x = ( t x ) 1 t \large \displaystyle x = (tx)^{\frac{1}{t}}

x = t 1 t x 1 t \large \displaystyle x = t^{\frac{1}{t}} \cdot x^{\frac{1}{t}}

x 1 1 t = t 1 t \large \displaystyle x^{1 - \frac{1}{t}} = t^{\frac{1}{t}}

x = t 1 t 1 \color{#20A900} \boxed{ \large \displaystyle x = t^{\frac{1}{t-1}} }

y = t x = t t t 1 \color{#20A900} \boxed{ \large \displaystyle y = tx = t^{\frac{t}{t-1}} }

This is a family of solutions based in a parametric parameter t t . So we're looking to minimize:

S = t 1 t 1 + t t t 1 \large \displaystyle S = t^{\frac{1}{t-1}} + t^{\frac{t}{t-1}}

Differtiating and making it equal to zero:

d S d t = t 1 t 1 ( 1 t ( t 1 ) ln ( t ) ( t 1 ) 2 ) + t t t 1 ( ( 1 + ln ( t ) ) ( t 1 ) t ln ( t ) ( t 1 ) 2 ) = 0 \large \displaystyle \frac{dS}{dt} = t^{\frac{1}{t-1}} \left ( \frac{\frac{1}{t} (t-1) - \ln(t) }{(t-1)^2} \right ) + t^{\frac{t}{t-1}} \left ( \frac{(1 + \ln(t))(t-1) - t \ln(t)}{(t-1)^2} \right ) = 0

Multiplying the first term above and below by t t :

d S d t = t t t 1 1 t ( 1 t ( t 1 ) ln ( t ) ( t 1 ) 2 ) + t t t 1 ( ( 1 + ln ( t ) ) ( t 1 ) t ln ( t ) ( t 1 ) 2 ) = 0 \large \displaystyle \frac{dS}{dt} = t^{\frac{t}{t-1}} \frac{1}{t} \left ( \frac{\frac{1}{t} (t-1) - \ln(t) }{(t-1)^2} \right ) + t^{\frac{t}{t-1}} \left ( \frac{(1 + \ln(t))(t-1) - t \ln(t)}{(t-1)^2} \right ) = 0

Multiplying above and below by ( t 1 ) 2 t t t 1 \frac{ (t-1)^2 }{ t^{\frac{t}{t-1}} } (which implies t 0 t \neq 0 and t 1 x y t \neq 1 \rightarrow x \neq y , as the problem states):

1 t ( t 1 t ln ( t ) ) + ( t 1 ) ( 1 + ln ( t ) ) t ln ( t ) = 0 \large \displaystyle \frac{1}{t} \left ( \frac{t-1}{t} - \ln(t) \right ) + (t-1)(1 + \ln(t)) - t\ln(t) = 0

Multiplying above and below by t 2 t^2 :

t 3 t 2 t 2 ln ( t ) + t t ln ( t ) 1 = 0 \large \displaystyle t^3 - t^2 - t^2 \ln(t) + t - t \ln(t) - 1 = 0

The only solution to this is t = 1 t=1 , which is forbidden by the steps above and by the problem itself ( x y x \neq y ). This means we're looking for a limiting condition, that's never achieved. I.e., the minimum M never is attained, but is tends to the following value:

M = lim t 1 S \large \displaystyle M = \lim_{t \rightarrow 1} S

M = lim t 1 [ t 1 t 1 + t t t 1 ] \large \displaystyle M = \lim_{t \rightarrow 1} \left [ t^{\frac{1}{t-1}} + t^{\frac{t}{t-1}} \right ]

M = lim t 1 [ t t t 1 ( 1 t + 1 ) ] \large \displaystyle M = \lim_{t \rightarrow 1} \left [ t^{\frac{t}{t-1}} \left ( \frac{1}{t} + 1 \right ) \right ]

ln ( M ) = lim t 1 [ ln ( 1 t + 1 ) + t t 1 ln ( t ) ] \large \displaystyle \ln(M) = \lim_{t \rightarrow 1} \left [ \ln \left ( \frac{1}{t} + 1 \right ) + \frac{t}{t-1} \ln(t) \right ]

ln ( M ) = ln ( 2 ) + lim t 1 [ t ln ( t ) t 1 ] \large \displaystyle \ln(M) = \ln(2) + \lim_{t \rightarrow 1} \left [ \frac{t \ln(t)}{t-1} \right ]

L'Hôpital's rule applies:

ln ( M ) = ln ( 2 ) + lim t 1 [ 1 + ln ( t ) 1 ] \large \displaystyle \ln(M) = \ln(2) + \lim_{t \rightarrow 1} \left [ \frac{1 + \ln(t)}{1} \right ]

ln ( M ) = ln ( 2 ) + 1 \large \displaystyle \ln(M) = \ln(2) + 1

M = 2 e 5.4366 \color{#3D99F6} \boxed{ \large \displaystyle M = 2e \approx 5.4366 }

But the question says x x and y y are "different". The minimum is attained when they are equal. Doesn't it lead to a contradiction?

A Former Brilliant Member - 1 year, 5 months ago

Log in to reply

Theoretically yes, since this condition of x + y = 2 e x+y=2e will never actually be achieved. But it is the limiting case. The minimum is an infinitesimal away from it.

Guilherme Niedu - 1 year, 5 months ago

Log in to reply

It will never actually be achieved with distinct x x and y y , to be precise. :)

A Former Brilliant Member - 1 year, 5 months ago

Log in to reply

@A Former Brilliant Member Exactly! :D

Guilherme Niedu - 1 year, 5 months ago
Alapan Das
Jan 7, 2020

We are intended to find two solutions of equation f ( x ) = l n ( x ) x = r , r < 1 e f(x)=\frac{ln(x)}{x}=r ,r<\frac{1}{e} . That is possible for all x > 1 x>1 and not for any x 1 x≤1 . As the maximum value of f ( x ) f(x) is 1 e \frac{1}{e} .

Now, looking at the graph of f ( x ) = l n ( x ) x f(x)=\frac{ln(x)}{x} we see that the curve less steep on positive side of x = e x=e , the maximum point. Which means for some x e x≠e , and y y ( the companion of x x , or f ( x ) = f ( y ) f(x)=f(y) x < y x<y ), y e > e x x + y > 2 e y-e>e-x \Rightarrow x+y>2e . Hence the x x for which x + y x+y is minimum is e e .

So, the answer is 2 e = 5.437 2*e=5.437 .

The curve of the function f ( x ) = ( x ) 1 x f(x)=(x)^{\dfrac{1}{x}} is positively skewed. So, with growth of x x the average of two values of x x where f ( x ) f(x) assumes the same value shifts towards right. Hence this average is minimum when f ( x ) f(x) attains it's peak value of ( e ) 1 e (e)^{\dfrac{1}{e}} at x = e x=e . Therefore the minimum value of x + y x+y is 2 e \boxed {2e}

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...