2
$\begingroup$

I have to find the supremum of the following function: $$f(x)=\frac{x}{x+1} \cdot \sin x$$, where $x \in (0,\infty)$

I think I know it is equal to $1$ but I can't prove it.

Where I'm stuck proving that $\sup f =1$:

Let $\sup f = y$

Let $\varepsilon>0$

We'll prove that for every $\varepsilon>0$ there is a value of $x$ such that $\frac{x}{x+1} \cdot \sin x$ > $y - \varepsilon$ - That is the formal definition in my text book.

Thanks!

  • 1
    What happens if $x\to-1$? – 2017-01-14
  • 0
    Please see my domain edit - $x \in (0,\infty)$ – 2017-01-14
  • 0
    Also, there's a mixing of $f$ and $g$ here. – 2017-01-14
  • 0
    @B.Goddard thank you, fixed that! – 2017-01-14
  • 0
    You're in the Open University aren't you? – 2017-01-14
  • 0
    @Theorem I am. So are you? – 2017-01-14
  • 0
    Indeed. :) I already asked about that question here, no need for finding the $\sup$! http://math.stackexchange.com/questions/2090790/show-gx-fracx-sin-xx1-has-no-maxima-in-0-infty – 2017-01-14
  • 0
    Thanks for the link! however I do find it easier with the sup, oddly :-) usually I hate the whole sup/inf thing. – 2017-01-14

4 Answers 4

3

Hint : $\sin(x)=1$ has arbitary large solutions and we have $$\lim_{x\rightarrow \infty} \frac{x}{x+1}=1$$

Moreover, $\frac{x}{x+1}$ is strictly increasing for $x>0$

  • 0
    All to easy. (+1) – 2017-01-14
  • 1
    thanks, all I have to prove is that sup(AB)=sup a* sup b? – 2017-01-14
  • 1
    @Alan: sup(AB) = sup(A)*sup(B) in general will not be correct. Consider for instance $A(x)=1-x$ and $B(x)=x$ over the range 0 – 2017-01-14
  • 0
    @Gamow thank, you're right, however in this case I believe I managed to prove it does work. – 2017-01-14
0

The sup of your function is $+\infty$; in fact, you have that $\lim _{x\to-1}f(x)=+\infty$ if $x$ tends to $-1$ from the right. With the domain restriction, we can observe that $f(x)< \frac x {x+1}$ that is $<1$ and tends to $1$ as $x\to+\infty$.

  • 1
    I imagine there were domain restrictions, so I'm waiting for the OP's feedback. – 2017-01-14
  • 0
    Please see my domain edit - $x \in (0,\infty)$ – 2017-01-14
0

Let $\epsilon>0$ be given. Choose $N$ so big that $1-\frac{1}{N+1} = \frac{N}{N+1} > 1 -\epsilon.$ Then for any integer $M$ such that $\alpha = \frac{(2M+1)\pi}{2} >N$, we have

$$1 - \frac{\alpha}{1+\alpha} \sin(\alpha) = 1- \frac{\alpha}{1+\alpha} > 1-\epsilon. $$

So the supremum is at least $1.$

0

(1) First we argue that $f(x)\le1$ for all $x\ge0$.
Indeed, $\sin x\le1$ implies $x\sin x\le x$ implies $x\sin x\le x+1$ implies $f(x)\le1$.

(2) Next we argue that $f$ comes arbitrarily close to $1$.
For this, consider $x_k=\frac{\pi}{2}+k\pi$ for an integer $k\ge0$. Since $\sin(x_k)=1$, we get $$f(x_k) = \frac{x_k}{(x_k+1)} = \frac{\frac{\pi}{2}+k\pi}{\frac{\pi}{2}+k\pi+1} \ge \frac{\frac{\pi}{2}+k\pi}{\frac{\pi}{2}+k\pi+ \frac{\pi}{2}} = \frac{2k+1}{2k+2}.$$ Now if we pick $k>(1-2\varepsilon)/\varepsilon$, we have $$ \frac{2k+1}{2k+2}>1-\varepsilon. $$ Hence for every $\varepsilon>0$, there indeed exists the desired value $x$ for which $f(x)>1-\varepsilon$.

(3) This means that the supremum is $1$.