Advanced Mean Value Theorem - Part 1

Calculus Level 3

Given that f ( x ) f(x) is a differentiable function defined at R \mathbb R , f ( 0 ) = 1 f(0)=1 , and f ( 1 ) = 1 2 f(1)=\dfrac{1}{2} .

Does there always exist a ξ ( 0 , 1 ) \xi \in (0,1) such that f 2 ( ξ ) + f ( ξ ) = 0 ? f^2(\xi)+f'(\xi)=0\ ?

Yes, always Yes, but only sometimes No, never

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

We solve the problem using two different functions. One for the case f f doesn’t have any zeros in ( 0 , 1 ) \left( 0,1 \right) and another one in the opposite case.

\ \

Case 1 : Suppose that for the given function f f it holds f ( x ) 0 f\left( x \right)\ne 0 , x ( 0 , 1 ) \forall x\in \left( 0,\ 1 \right) .
Then, the function g ( x ) = 1 f ( x ) x {\color{#D61F06}{g\left( x \right)=\dfrac{1}{f\left( x \right)}-x}} can be defined on [ 0 , 1 ] \left[ 0,\ 1 \right] , g g is continuous within the interval [ 0 , 1 ] \left[ 0,\ 1 \right] , differentiable within the interval ( 0 , 1 ) \left( 0,\ 1 \right) , with g ( x ) = f ( x ) f 2 ( x ) 1 {g}'\left( x \right)=-\dfrac{{f}'\left( x \right)}{{{f}^{2}}\left( x \right)}-1 and g ( 0 ) = g ( 1 ) = 1 g\left( 0 \right)=g\left( 1 \right)=1 . Hence, by Rolle’s theorem , there exists a ξ ( 0 , 1 ) \xi\in \left( 0,\ 1 \right) such that
g ( ξ ) = 0 f ( ξ ) f 2 ( ξ ) 1 = 0 f ( ξ ) + f 2 ( ξ ) = 0 {g}'\left( \xi \right)=0\Rightarrow -\dfrac{{f}'\left( \xi \right)} {{{f}^{2}}\left( \xi \right)}-1=0\Rightarrow {f}'\left( \xi \right)+{{f}^{2}}\left( \xi \right)=0

\ \

Case 2 : Suppose the given function f f has at least one zero x 0 ( 0 , 1 ) {{x}_{0}}\in \left( 0,\ 1 \right) .
Define h ( x ) = f ( x ) e F ( x ) {\color{#20A900}{h\left( x \right)=f\left( x \right)\cdot {{e}^{F\left( x \right)}}}} on the interval [ 0 , 1 ] \left[ 0,\ 1 \right] , where F ( x ) = 0 x f ( t ) d t F\left( x \right)=\int\limits_{0}^{x}{f\left( t \right)}dt , x [ 0 , 1 ] \forall x\in \left[ 0,\ 1 \right] .

By the Fundamental Theorem of Calculus F F is differentiable with F ( x ) = f ( x ) {F}'\left( x \right)=f\left( x \right) . Consequently, h h is differentiable (thus, continuous) within the interval [ 0 , 1 ] \left[ 0,\ 1 \right] , with h ( x ) = f ( x ) e F ( x ) + f ( x ) ( e F ( x ) ) = f ( x ) e F ( x ) + f ( x ) F ( x ) e F ( x ) = f ( x ) e F ( x ) + f 2 ( x ) e F ( x ) = ( f ( x ) + f 2 ( x ) ) e F ( x ) \begin{aligned} {h}'\left( x \right) & ={f}'\left( x \right)\cdot {{e}^{F\left( x \right)}}+f\left( x \right)\cdot {{\left( {{e}^{F\left( x \right)}} \right)}^{\prime }} \\ & ={f}'\left( x \right)\cdot {{e}^{F\left( x \right)}}+f\left( x \right)\cdot {F}'\left( x \right){{e}^{F\left( x \right)}} \\ & ={f}'\left( x \right)\cdot {{e}^{F\left( x \right)}}+{{f}^{2}}\left( x \right)\cdot {{e}^{F\left( x \right)}} \\ & =\left( {f}'\left( x \right)+{{f}^{2}}\left( x \right) \right)\cdot {{e}^{F\left( x \right)}} \\ \end{aligned}

Moreover, h ( 0 ) = 1 h\left( 0 \right)=1 , h ( 1 ) = 1 2 e F ( 1 ) > 0 h\left( 1 \right)=\frac{1}{2}{{e}^{F\left( 1 \right)}}>0 and h ( x 0 ) = f ( x 0 ) e F ( x 0 ) = 0 h\left( {{x}_{0}} \right)=f\left( {{x}_{0}} \right)\cdot {{e}^{F\left( {{x}_{0}} \right)}}=0 .
By the extreme value theorem , h h has a minimum value which, being non positive, is evidently attained at some ξ ( 0 , 1 ) \xi \in \left( 0,\ 1 \right) .

By Fermat’s theorem for Extrema (Interior Extremum Theorem) it holds

h ( ξ ) = 0 ( f ( ξ ) + f 2 ( ξ ) ) e F ( ξ ) = 0 f ( ξ ) + f 2 ( ξ ) = 0 {h}'\left( \xi \right)=0\Rightarrow \left( {f}'\left( \xi \right)+{{f}^{2}}\left( \xi \right) \right)\cdot {{e}^{F\left( \xi \right)}}=0\Rightarrow {f}'\left( \xi \right)+{{f}^{2}}\left( \xi \right)=0

Therefore, whichever case we are given, there exists a ξ ( 0 , 1 ) \xi \in \left( 0,\ 1 \right) such that f ( ξ ) + f 2 ( ξ ) = 0 {f}'\left( \xi \right)+{{f}^{2}}\left( \xi \right)=0 .

Chris Lewis
Nov 14, 2020

Let a a be the smallest positive root of f ( x ) = 1 2 f(x)=\frac12 . From the given information, there is such an a a , and a 1 a\le1 . Also, by the intermediate value theorem, f ( x ) > 1 2 f(x)>\frac12 for x ( 0 , a ) x \in (0,a) .

Define g ( x ) = 1 f ( x ) g(x)=\frac{1}{f(x)} on the interval [ 0 , a ] [0,a] , so that g ( 0 ) = 1 g(0)=1 , g ( a ) = 2 g(a)=2 and g ( x ) g(x) inherits differentiability. Then, by the mean value theorem, there exists a ξ ( 0 , a ) \xi \in (0,a) such that g ( ξ ) = 1 g'(\xi)=1 .

But g ( ξ ) = f ( ξ ) f ( ξ ) 2 g'(\xi)=-\frac{f'(\xi)}{f(\xi)^2} so this implies that f ( ξ ) + f ( ξ ) 2 = 0 f'(\xi)+f(\xi)^2=0 as required.

Nice solution! I have thought of the same construction, but didn't know how to deal with the f(x)=0 case. Your solution has done it elegantly!

Alice Smith - 6 months, 4 weeks ago

Hi Chris! Using g ( x ) = 1 f ( x ) g\left( x \right)=\dfrac{1}{f\left( x \right)} is a nice idea, but when we apply mean value theorem on [ 0 , a ] \left[ 0,a \right] we get the existence of a ξ ( 0 , a ) \xi \in \left( 0,a \right) such that g ( ξ ) = 1 a {g}'\left( \xi \right)=\dfrac{1}{a} . Hence, In the case a < 1 a<1 , we have g ( ξ ) 1 {g}'\left( \xi \right)\ne 1 .

Thanos Petropoulos - 6 months, 4 weeks ago

Log in to reply

Thanks Thanos - good point! You probably guessed that I realised f ( x ) = 0 f(x)=0 was a problem a bit too late...

Chris Lewis - 6 months, 4 weeks ago
Tom Engelsman
Nov 13, 2020

If we solve this first-order, separable ODE for y = f ( ξ ) y = f(\xi) we get:

f ( ξ ) = f 2 ( ξ ) , f ( 0 ) = 1 , f ( 1 ) = 1 2 f'(\xi) = -f^{2}(\xi), f(0)=1, f(1)=\frac{1}{2} ,

or d y y 2 = d ξ \frac{dy}{y^2} = -d\xi ;

or 1 y = ξ + C -\frac{1}{y} = -\xi + C

which both of the above boundary conditions yield C = 1 C=-1 , or y = f ( ξ ) = 1 ξ + 1 . \boxed{y = f(\xi) = \frac{1}{\xi + 1}}. This function is indeed continuous and differentiable for all 0 ξ 1 0 \le \xi \le 1 \Rightarrow there will always be ξ ( 0 , 1 ) \xi \in (0,1) that satisfies the original ODE above.

Hello, can you explain your last step? You seem to have found a function that satisfies the constraint at every point, which is interesting in and of itself, but how does it follow that there will be a point satisfying the constraint with any differentiable function?

Adrian Klaeger - 7 months ago

Log in to reply

I think there's a way to get to a solution from this using an inequality version of this differential equation: aiming for a proof by contradiction, if there's no such ξ \xi , then either f ( x ) + f ( x ) 2 < 0 f'(x)+f(x)^2<0 on the entire interval, or f ( x ) + f ( x ) 2 > 0 f'(x)+f(x)^2>0 on the entire interval. Intuitively this could work by considering the vector field associated with the differential equation, and showing that the constraint forces the function to "miss" one of the boundary conditions, but I can't quite see how to formalise this neatly.

Looking at 1 f ( x ) \frac{1}{f(x)} seems to work - I've posted a solution based on that, but I'd be interested if you could get this idea to work too.

Chris Lewis - 6 months, 4 weeks ago

Since f ( x ) f(x) is differentiable for all real numbers, it must be continuous for all real numbers. Therefore, it satisfies the requirements for the Mean Value Theorem, meaning that there exists some c ( 0 , 1 ) c \in (0, 1) such that f ( c ) = f ( 1 ) f ( 0 ) 1 0 = 1 2 f'(c) = \frac{f(1) - f(0)}{1 - 0} = -\frac{1}{2} (since c c is in the interval ( 0 , 1 ) (0, 1) then we can just denote it as ξ \xi ). If we assume f 2 ( ξ ) + f ( ξ ) = 0 f ( ξ ) = f 2 ( ξ ) f^2(\xi) + f'(\xi) = 0 \rightarrow f'(\xi) = -f^2(\xi) , then f 2 ( ξ ) = 1 2 -f^2(\xi) = -\frac{1}{2} or f ( ξ ) = 2 2 f(\xi) = \frac{\sqrt2}{2} . Because f ( x ) f(x) is continuous on the interval ( 0 , 1 ) (0, 1) and has a range of ( 1 2 , 1 ) (\frac{1}{2}, 1) , regardless of our choice of f ( x ) f(x) , there exists a ξ \xi that satisifies f 2 ( ξ ) + f ( ξ ) = 0 f^2(\xi) + f'(\xi) = 0 . If the value we calculated for f ( ξ ) f(\xi) did not fall in the range of f ( x ) f(x) for the domain ( 0 , 1 ) (0, 1) , the property would not apply.

Hello, the MVT guarantees that there will be a point with that derivative. You then calculate the value the function would need to take at that point for the constraint to be satisfied, but there is no reason why it should have to take that exact value. For example, a function could begin by rising or falling rapidly, before then taking that derivative at some point, allowing the function to have any real value at that point. Note also that the range of the function is not restricted.

Adrian Klaeger - 7 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...