Constantly (well, not really) 0 integral

Level 2

The non-linear relation f ( a R + ) b f(a\in \mathbb{R}^+)\rightarrow b is a function that takes a value a a and returns a value b b such that a b ln x x d x = 0 \int_a^b\dfrac{\ln x}{x}\text{ }dx=0 What is the minimum value of a + b ? a+b\text{?} \text{ }

Details and Assumptions \textbf{Details and Assumptions}

f ( n ) f(n) is differentiable over its domain.


The answer is 2.

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

Trevor B.
Feb 2, 2014

Since we aren’t given any values for a a or b , b, let’s see if we can find what the integral is equal to.

Let u = ln x . u=\ln x. Then d u = 1 x d x . du=\frac{1}{x}dx. Proceeding with the u -substitution, u\text{-substitution,} the integral becomes this. ln a ln b u d u \int_{\ln a}^{\ln b}u\text{ }du This is equal to 1 2 u 2 ] ln a ln b \left.\frac{1}{2}u^2\right]_{\ln a}^{\ln b} = = ( ln a ) 2 ( ln b ) 2 2 = 0 \frac{(\ln a)^2-(\ln b)^2}{2}=0

A little rearranging gives ( ln a ) 2 = ( ln b ) 2 , (\ln a)^2=(\ln b)^2, so ln a = ± ln b a = b ± 1 \ln a=\pm\ln b\Rightarrow a=b^{\pm1} We have two possible relations that could be f ( a ) . f(a). These are f ( a ) = a f(a)=a and f ( a ) = 1 a . f(a)=\frac{1}{a}. However, since f ( a ) f(a) is non-linear, we can eliminate the possibility that f ( a ) = a f(a)=a and know that f ( a ) = 1 a . f(a)=\frac{1}{a}.

Now we need to find the minimum value of a new relation g ( a ) = a + f ( a ) = a + 1 a . g(a)=a+f(a)=a+\frac{1}{a}. Deriving, we get g ( a ) = 1 1 a 2 g’(a)=1-\frac{1}{a^2} This equals 0 0 when a = ± 1 , a=\pm1, but 1 -1 is not in the domain of f ( a ) f(a) and is therefore not in the domain of g ( a ) g(a) either, so we can disregard it. Applying the First Derivative Test, we find that a = 1 a=1 is the a -coordinate a\text{-coordinate} of the local minimum of g ( a ) , g(a), but since g ( a ) g(a) does not change direction again and is continuous on its domain, g ( a ) = 2 g(a)=2 is also the absolute minimum.

Therefore, the point ( a , b ) (a,b) we are looking for is ( 1 , 1 ) (1,1) so a + b = 2 a+b=\boxed{2}

I found out that ab=1 and used AM-GM to minimize a+b.

Kenneth Wang - 7 years, 4 months ago

Log in to reply

Also a good way to do it!

Trevor B. - 7 years, 4 months ago

You need to be more careful with rejecting the other possibility. All that you know is that the point wise value is f ( a ) = a f(a) = a or 1 a \frac{1}{a} . This does not imply that f ( x ) = x f(x) = x or f ( x ) = 1 x f(x) = \frac{1}{x} for all (positive) real values. (This is a very common mistake made in functional equations.)

Without the restriction that f ( n ) f(n) is differentiable, there would be infinitely many solutions. For example, if all we had was continuity, then we could have f ( x ) = 1 x f(x) = \frac{1}{x} for 0 < x < 1 0<x<1 and f ( x ) = x f(x) = x for 1 x 1 \leq x .

Edit: I see in that you addressed this in comments to Finn's solution.

Calvin Lin Staff - 7 years, 4 months ago

One thing Trevor, does the condition imply that f f always takes a a and always gives b b as the value? I mean every function takes one value in its domain and gives another value as the output, so why did you write the statement that the function takes a value a a and gives a value b ? b?

Jit Ganguly - 7 years, 4 months ago
Jatin Yadav
Feb 2, 2014

If we put ln x = t \ln x = t , we get d x x = d t \frac{dx}{x} = dt and the integral as 1 / 2 ( ( ln b ) 2 ( ln a ) 2 1/2((\ln b)^2 - (\ln a)^2 ) = 1 2 ln ( b a ) ln ( a b ) \frac{1}{2} \ln (\frac{b}{a}) \ln (ab) .

For it to be 0 0 , either b = 1 a b = \frac{1}{a} or b = a b=a . I assume b = 1 a b = \frac{1}{a} , otherwise there would be no min. value.

Hence, a + b = a + 1 a 2 a + b = a + \frac{1}{a} \geq \boxed{2} for a > 0 a > 0

Finn Hulse
Feb 2, 2014

First we integrate. We find that the indefinite integral is 1/2 times ln(x) squared. Let's start just by setting this equal to zero. Ln of 1 is zero (anything to the zero is one), therefore if a and b both were one, then the integral will come down to zero. Of course, we must remember: this is the same thing as asking what the limit of the size of delta x is. Of course, this will be true for any numbers that are equal, because it's a tiny sliver of space that is actually not quite zero, but we'll just say that it is. That was the easy part. Now, this is going to require some differential equations and a bit of topology. We are going to do one of the Millenium Challenges. We are going to add 1 to 1. I know, I know, it's impossible. Nonetheless! The first thing we do is add: 1+1. We get 2. Done.

By this logic, something like this integral would be an acceptable answer as well. What's stopping the answer from being 2 π < 2 ? \frac{2}{\pi}<2\text{?} 1 π 1 π ln x x d x \int_{\frac{1}{\pi}}^{\frac{1}{\pi}}\dfrac{\ln x}{x}\text{ }dx In general, if f ( x ) f(x) is continuous at and around n , n, then lim h 0 n h n + h f ( x ) d x = 0 \lim_{h\rightarrow0}\int_{n-h}^{n+h}f(x)\text{ }dx=0 This would make the relation linear, which cannot happen.

Trevor B. - 7 years, 4 months ago

Log in to reply

Why would it make f f linear?

Jit Ganguly - 7 years, 4 months ago

Yes, but you could have f ( x ) = x f(x) = x for some values of x x and not others. E.g. the function f ( x ) = x f(x) = x for x 1 x \le 1 , f ( x ) = 1 / x f(x) = 1/x for x 1 x \ge 1 is non-linear. It seems like you need to make some more stringent restrictions on your function/relation/whatever.

Patrick Corn - 7 years, 4 months ago

Log in to reply

I added that it is differentiable over its domain, eliminating that possibility.

Trevor B. - 7 years, 4 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...