I was wondering whether it is possible to prove that $f(x)=\frac{\sin x}{x}$ is strictly decreasing in $(0, \pi]$ and strictly increasing in $[-\pi , 0)$ without using any derivatives at all. If it is possible, how?
3 Answers
By the Weierstrass product: $$\frac{\sin x}{x}=\prod_{n\geq 1}\left(1-\frac{x^2}{n^2\pi^2}\right)$$ but for any $n\geq 1$ the function $f_n(x)=\left(1-\frac{x^2}{n^2\pi^2}\right)$ is non-negative and decreasing over $\left[0,\pi\right]$.
It follows that $f(x)=\frac{\sin x}{x}$ is non-negative and decreasing over $\left[0,\pi\right]$, and we just need to notice that $f(x)$ is an even function to prove the claim.
- 361,689
-
To Jack D'Aurizio, How can we prove $\displaystyle \frac{\sin x}{x}=\prod_{n\geq 0}\left(1-\frac{x^2}{n^2\pi^2}\right),$ Thanks – juantheron Aug 23 '15 at 13:41
-
1@juantheron: it is a very well-known result, it follows from Weierstrass factorization theorem. Without resorting to complex variables, such identity can be proved by exploiting the factorizations of Chebyshev polynomials of the first and second kind and by taking a limit. – Jack D'Aurizio Aug 23 '15 at 14:01
-
-
Let $y=x+h$ where $h>0$. We want to show:
$$\frac{\sin(x+h)}{x+h}-\frac{\sin(x)}{x}<0$$
for $x,y \in (0,\pi]$. So we compute:
$$\frac{\sin(x+h)}{x+h}-\frac{\sin(x)}{x}=\frac{\sin(x+h)x-\sin(x)(x+h)}{x(x+h)}$$
Because $x(x+h)>0$ is is sufficient to show that:
$$\sin(x+h)x-\sin(x)(x+h)<0$$
Next we use formula for $\sin$ of sum:
$$\sin(x)\cos(h)x+\sin(h)\cos(x)x-\sin(x)(x+h)<0$$
But:
$$\cos(h) \leq 1$$
So it suffices to show:
$$\sin(x)x+\sin(h)\cos(x)x-\sin(x)(x+h)<0$$
Next:
$$\sin(x)x+\sin(h)\cos(x)x-\sin(x)(x+h)=\sin(h)\cos(x)x-h\sin(x)$$
So it's equivalent:
$$\frac{\cos(x)x}{\sin(x)}<\frac{h}{\sin(h)}$$
But $h>\sin(h)$, so it is sifficent to show:
$$\cos(x)x-\sin(x)<0$$
(Note that using derivatives gives us the same inequality to prove!) We can prove that by any method (for example, compare $x$ and $\tan(x)$ on $(0,\frac{\pi}{2})$, on $(\frac{\pi}{2},\pi)$ it is obvious).
- 10,174
-
-
1@Jason, here: http://math.stackexchange.com/questions/98998/why-x-tanx-while-0x-frac-pi2 you can find some approaches. – agha Aug 23 '15 at 14:49
Observe that
$$\frac{\sin x}{x} = \frac{1}{x}\int_0^x \cos t\, dt.$$
This is the average of $\cos t$ over $[0,x].$ Now $\cos t$ is strictly decreasing on $[0,\pi].$ So if $0\le x <y \le \pi,$ the average over $[0,y]$ is smaller than the average over $[0,x],$ simply because we are adding in strictly smaller values to the mix.
- 107,943
You need to show that if $0<x<y\leq \pi$ then $y\sin x-x\sin y>0$.
– Thomas Andrews Aug 23 '15 at 12:26