1

$$F (x) = \begin{cases} \displaystyle \frac{\tan(x)}{x} & x \not=0 \\ 1 & x=0 \end{cases} $$

How do I prove that there is a local minimum at $x=0$?

Natalya
  • 49
  • From where did you get this question? – Aritra Das May 15 '16 at 18:56
  • I'm afraid he hasn't taught you much. Definitely not enough for this problem. – Aritra Das May 15 '16 at 20:37
  • They are some of the first things that are taught in calculus. I'm in school and I know those. It's "fundamental" for a reason. It ties calculus neatly together. – Aritra Das May 15 '16 at 20:40
  • I didn't learn it yet that's weird.. So anyways which solution should I use? I would need one that does not involve those theorems just trigonometric derivatives – Natalya May 15 '16 at 20:42
  • Firstly, you should accept William's answer. It is undoubtedly the best since it lists many methods and he suggests many links which are useful. Secondly, if you wish to write down a solution, you should use @user5713492 's answer's first part (before the edit). That is probably what you're after. I hope I could be of some help. Cheers. – Aritra Das May 15 '16 at 20:51

4 Answers4

4

The first derivative test seems easiest to apply here. For $x>0$, $$\frac d{dx}\frac{\tan x}x=-\frac1{x^2}\tan x+\frac1x\sec^2x=\frac{x-\sin x\cos x}{x^2\cos^2x}$$ By a geometric argument you were shown, around the time you learned the derivatives of the trigonometric functions, that for $0<x<\frac{\pi}2$, $\sin x\cos x<x<\tan x$. For this reason we know that $$\frac d{dx}\frac{\tan x}x=\frac{x-\sin x\cos x}{x^2\cos^2x}>0$$ for $0<x<\frac{\pi}2$. It follows that $-(x-\sin x\cos x)=(-x)-\sin(-x)\cos(-x)<0$ for $0<x<\frac{\pi}2$ so that $x-\sin x\cos x<0$ for $-\frac{\pi}2<x<0$, so $$\frac d{dx}\frac{\tan x}x=\frac{x-\sin x\cos x}{x^2\cos^2x}<0$$ for $-\frac{\pi}2<x<0$. Since $$\lim_{x\rightarrow0}\frac{\tan x}x=\lim_{x\rightarrow0}\left(\frac1{\cos x}\frac{\sin x}x\right)=(1)(1)=1=F(1)$$ Then $F(x)$ is continuous at $x=0$ and $F^{\prime}(x)<0$ for $-\frac{\pi}2<x<0$ and $F^{\prime}(x)>0$ for $0<x<\frac{\pi}2$ it follows that $F(x)$ has a relative minimum at $x=0$ by the first derivative test.

EDIT: Here is the geometric argument I was alluding to. Consider this figure: enter image description here

It may be rather hard to read my crude diagrams, but it is supposed to depict the unit circle centered at $O$. Sector $OQS$ of the circle has area $$A(OQS)=\frac{\theta}{2\pi}\pi(1)^2=\frac12\theta$$ Where we have assumed that the area of a sector is proportional to its central angle, here $\theta$, and used the mensuration formula for the area of a circle, $A=\pi r^2$. Then right triangle $ORS$ lies entirely within sector $OQS$ and its area is $$A(ORS)=\frac12\cos\theta\sin\theta$$ where we have use the mensuration formula $A=\frac12(\text{base})(\text{height})$ for the area of a triangle. Finally the sector $OQS$ itself lies entirely within right triangle $AQT$ with area $$A(OQT)=\frac12(1)\tan\theta=\frac12\tan\theta$$ Since one is contained within another, $$A(ORS)<A(OQS)<A(OQT)$$ Or $$\frac12\sin\theta\cos\theta<\frac12\theta<\frac12\tan\theta$$ I thought that this was the argument made to show how to differentiate trigonometric functions through $$\begin{align}\frac d{dx}\sin x&=\lim_{h\rightarrow0}\frac{\sin(x+h)-\sin x}h=\lim_{h\rightarrow0}\frac{\sin x\cos h+\cos x\sin h-\sin x}h\\ &=\lim_{h\rightarrow0}\frac{\sin x(1-2\sin^2(h/2))+\cos x\sin h-\sin x}h\\ &=\lim_{h\rightarrow0}\frac{\sin h}h\cos x-\lim_{h\rightarrow0}\frac{\sin(h/2)}{(h/2)}\sin(h/2)\sin x\\ &=(1)\cos x-(1)(0)\sin x=\cos x\end{align}$$ Because through the geometric argument you knew that $$\cos x<\frac{\sin x}x<\sec x$$ And since $$\lim_{h\rightarrow0}\cos x=\lim_{h\rightarrow0}\sec x=1$$ It follows by the squeeze theorem that $$\lim_{h\rightarrow0}\frac{\sin x}x=1$$

user5713492
  • 15,938
  • The $\frac{\pi}2$ restriction is not a problem because you only need an open interval including $x=0$ to evaluate limits and take derivatives and establish a local minimum or maximum. You can only hope for a local minimum because $$\lim_{x\rightarrow{\frac{\pi}{2}}^+}\frac{\tan x}x=-\infty$$ Just set your calculator in radians mode and start entering values of $x$ just a little bigger than $\frac{\pi}2$ to convince yourself of this. – user5713492 May 15 '16 at 19:56
2

Since you can rewrite $F$ as $$F(x)=\int_0^1{1\over\cos^2(\tau x)}\>d\tau\geq1=F(0)\qquad\left(-{\pi\over2}<x<{\pi\over2}\right)$$ the function $F$ takes its minimum in the interval $\>(-{\pi\over2},{\pi\over2})\ $ at $x=0$.

Chill2Macht
  • 20,920
1

(In what follows I am assuming that the domain here is $-\frac{\pi}{2} <x < \frac{\pi}{2}$)

For $x<0$, $\tan x < 0$, therefore $ \frac{\tan x}{x} >0$.

Likewise, for $x>0$, we have $\tan x > 0$, and again $\frac{\tan x}{x} > 0$.

Hence in order to show that the modified piecewise version of $\frac{\tan x}{x}$ has a minimum at $x=0$, it suffices to show that $|\tan x| > |x|$ for all $x \in (-\frac{\pi}{2},0) \cup (0, \frac{\pi}{2})$.

This is because this condition would imply that $\frac{\tan x}{x} >1$ for those values.

This is clear from the following graph:

enter image description here

(again, I am assuming that $x \in(-\frac{\pi}{2},0)\cup (0, \frac{\pi}{2})$)

since the graphs do not intersect. (Note: $\frac{\tan x}{x} >1 \iff \tan x > x$)

For a better graphical proof of the fact that $$\tan x > x, \quad x \in (0, \frac{\pi}{2}$$ see here: http://mathrefresher.blogspot.com/2006/08/sin-x-x-tan-x-for-x-in-02.html

or here: Why $x<\tan{x}$ while $0<x<\frac{\pi}{2}$?

or: https://www.quora.com/How-do-I-prove-this-displaystyle-frac-tan-x-x-gt-frac-x-sin-x-when-x-in-left-0-frac-pi-2-right

or: https://www.quora.com/How-do-I-prove-that-sin-x-x-tan-x-for-all-x-in-left-0-frac-pi-2-right

or: https://answers.yahoo.com/question/index?qid=1006050131519

or: https://answers.yahoo.com/question/index?qid=20110926004423AAR5UZM

Non-Graphical Proof #1

However, I imagine that an analytic proof for the fact that $|\tan x|>|x|$ for these points would be preferred.

By symmetry (i.e. the fact that both tangent and $x$ are odd), it suffices to show this for just $x \in (0, \frac{\pi}{2})$ (in other words, the proof for $x \in (-\frac{\pi}{2},0)$ will be completely analogous up to changes in sign).

This follows from the Taylor expansion of $\tan x$ around $x=0$:

$$\tan x = x + \frac{1}{3}x^3 + \frac{2}{15}x^5 + \dots = \displaystyle\sum\limits_{n=0}^{\infty} \frac{U_{2n+1}x^{2n+1}}{(2n+1)!}$$

(See: https://en.wikipedia.org/wiki/Trigonometric_functions#Tangent)

Hence, in particular, it follows that $\tan x > x + \frac{1}{3}x^3 > x$ on the interval in question $x \in (0, \frac{\pi}{2})$.

Completely analogously, $\tan x < x + \frac{1}{3}x^3 < x$ for $x \in (-\frac{\pi}{2},0)$.

Therefore, $|\tan x| > |x|$ for all $x \in (-\frac{\pi}{2},0) \cup (0, \frac{\pi}{2})$, it follows that in particular

$$\frac{\tan x}{x} > 1$$ for all $x \in (-\frac{\pi}{2},0) \cup (0, \frac{\pi}{2})$, and therefore the piecewise defined function in question has a minimum at $x=0$.

For proof without Taylor series:

Note that $x = \int_0^x 1 \text{d}t$ (Fundamental theorem of calculus, the fact that $x'=1$, and $x(0)=0$)

Meanwhile, because $\tan 0 = 0$, as well as the Fundamental Theorem of Calculus, $$\tan x = \int_0^x (\tan t)' \text{d}t = \int_0^x \sec^2 t \text{d}t = \int_0^x \displaystyle\frac{1}{\cos^2 t} \text{d}t$$

Since for $x \in (0, \frac{\pi}{2})$, we have

$$\cos x <1 \implies \cos^2 x < 1 \implies \displaystyle\frac{1}{\cos^2 x} > 1$$

it follows that (since $f>g \implies \int_0^x f \text{d}t > \int_0^x g \text{d}t$):

$$ \int_0^x \displaystyle\frac{1}{\cos^2 t} \text{d}t > \int_0^x 1 \text{d}t$$

which is equivalent to:

$$ \tan x > x,\ \text{for }x \in (0, \frac{\pi}{2})$$.

Using the fact that $\tan(-x)=-\tan x$, it follows immediately that $\tan x < X$ for $x \in (-\frac{\pi}{2},0)$.

Therefore $\left| \displaystyle\frac{\tan x}{x} \right| > 1$ for all $x \in (-\frac{\pi}{2}, 0) \cup (0, \frac{\pi}{2})$, as claimed.

Another One:

By definition of $\cos x$, we have that $$\text{for }x \in (0, \frac{\pi}{2}), 1 < \cos x.$$

Now we show, using this fact, that $$\sin x < x \quad \forall\ x \in (0, \frac{\pi}{2}).$$

Note that $\frac{d}{dx} x =1$ and $\frac{d}{dx} \sin x = \cos x$.

Therefore,

$$\frac{d}{dx} \sin x < \frac{d}{dx} x\ \ \quad \forall\ x \in (0, \frac{\pi}{2})$$

As a result,

$$\int_0^x \frac{d}{dt}\sin t \text{d}t < \int_0^x \frac{d}{dt} t \text{d}t \quad \forall \ x \in (0, \frac{\pi}{2})$$

Because $\frac{d}{dx} x =1$ and $\frac{d}{dx} \sin x = \cos x$, it follows from the fundamental theorem of calculus (sorry but you're going to have learn it eventually and it's not difficult) that

$$\sin x = \int_0^x \frac{d}{dt}\sin t \text{d}t = \int_0^x \cos t \ \text{d}t$$ and $$x= \int_0^x \frac{d}{dt} t\ \text{d}t= \int_0^x 1\ \text{d}t.$$

Therefore it finally follows that

$$\sin x < x \quad \forall\ x\in (0,\frac{\pi}{2})$$.

From this follows immediately that

$$\displaystyle\frac{\sin x}{x} < 1.$$

Now, to show that $$\frac{\sin x}{x} < \displaystyle\frac{\tan}{x}$$

we use the facts that:

$$\tan x := \displaystyle\frac{\sin x}{\cos x}$$

and

$$\cos x < 1 \quad \forall\ x\in (0,\frac{\pi}{2})$$

Hence we get:

$$\displaystyle\frac{\tan x}{x} = \displaystyle\frac{\sin x}{\cos x \cdot x} = \displaystyle\frac{\sin x}{x} \cdot \displaystyle\frac{1}{\cos x} > \displaystyle\frac{\sin x}{x}$$

Then just use any of the arguments in the other answers which rely on these facts being true.

In any case, if you want an answer that's better than "look at this graph, see it's true" (which I already gave by the way), you'll have to use integration and differentiation together (to the best of my knowledge) i.e. the fundamental theorem of calculus.

Chill2Macht
  • 20,920
  • 1
    Thanks so much for the reply William! For tan (x)/x the restriction is that x can't be 0. Also is there a way to solve it without using the taylor series? (not familiar with it) – Natalya May 15 '16 at 16:32
  • see my recent edit for a proof solely relying on the fundamental theorem of calculus -- the idea (in my opinion) is the same, although perhaps a little more indirect. – Chill2Macht May 15 '16 at 16:49
  • 1
    Well $x=0$ isn't a global minimum for values of $x$ outside of that range -- also $\displaystyle \frac{\tan x}{x}$ isn't defined for $k\pi + \frac{pi}{2}$ where $k \in \mathbb{Z}$ i.e. is an arbitrary integer. What I have shown above is more than sufficient for demonstrating that $x=0$ is a local minimum, since I have shown that there exists an open neighborhood $(-\frac{\pi}{2},\frac{\pi}{2})$ for which $x=0$ is a minimum; again though, $x=0$ clearly is not a global minimum $\displaystyle\frac{\tan(\pi)}{\pi}=0$ for example, so I don't understand what else you could want me to prove. – Chill2Macht May 15 '16 at 17:33
  • Since $x=0$ is not the minimum for $\displaystyle\frac{\tan x}{x}$ I can't prove that it is, only that it is a local minimum. Do you want a proof that doesn't use calculus? What are you familiar with? You tagged the post with "calculus" so I assumed that you knew at least the Fundamental Theorem of Calculus. – Chill2Macht May 15 '16 at 17:35
  • In other words, the statement as you phrased it is not correct, and you tagged the post "calculus", so I gave two proofs using calculus. If I have misunderstood what you are looking for, now would be an ideal time to elaborate. As for proofs without calculus, it would be the same argument, just less elegant and more confusing; basically the most insightful argument I could come up with off the top of my head is that look at the graph I posted. – Chill2Macht May 15 '16 at 17:39
  • 1
    @Natalya Actually it is. Read the question carefully. It says that $F(x)= \frac{\tan x}{x}$ only for $x$ in $(-\frac{\pi}{2},0)\cup(0, \frac{\pi}{2})$. And for $x=0$, $F(x) =1$. Anywhere else, the function is not defined. – Aritra Das May 15 '16 at 18:25
  • @Natalya $\tan x$ has no continuous extension for $\frac{\pi}{2} \pm k\pi$, $k\in \mathbb{Z}$, hence neither does $\displaystyle\frac{\tan}{x}$, so the function you gave originally $\frac{\tan x}{x}$ for $x\not=0$ and $1$ for $x=0$ is not a well-defined real-function. – Chill2Macht May 15 '16 at 18:54
  • @Natalya what William means to say through all that math (in his last comment) is that the function neither has a maximum, nor a minimum. – Aritra Das May 15 '16 at 18:56
  • Moreover, the function you originally wrote DOES NOT HAVE A GLOBAL MINIMUM. This is because it approaches $-\infty$ from one side for all $\frac{\pi}{2}\pm k\pi, \forall\ k\in \mathbb{Z}$, since $\cos x =0$ at those points and has different signs (relative to $\sin x$) to the immediate left and right of those points. – Chill2Macht May 15 '16 at 18:56
  • The function you gave originally @Natalya, has a LOCAL minimum at $x=0$, nothing else. Also, keep in mind that it is just the continuous extension of $\displaystyle\frac{\tan x}{x}$, since by L'Hospital's Rule, we have $$\underset{x \to 0}{\lim} \displaystyle\frac{\tan x}{x} = \underset{x \to 0}{\lim} \displaystyle\frac{\sin x}{x} \displaystyle \frac{1}{\cos x} = \underset{x \to 0}{\lim} \displaystyle\frac{\sin x}{x} \cdot 1 = 1$$ – Chill2Macht May 15 '16 at 18:58
  • I don't believe you are understanding my point -- the "restriction" I had suffers no loss of generality (i.e. it actually answers your original question), since in order to show that the function has a LOCAL minimum at $x=0$, we just need to show that there is an open interval (e.g. $(-\frac{\pi}{2},\frac{\pi}{2})$ ) for which $x=0$ is a minimum for the RESTRICTION of the function. – Chill2Macht May 15 '16 at 19:00
  • The Fundamental Theorem of calculus is (more or less) just the fact that differentiable functions are the antiderivative of their derivatives, and that the antiderivative can be written as an integral. https://en.wikipedia.org/wiki/Fundamental_theorem_of_calculus – Chill2Macht May 15 '16 at 19:02
  • I am not familiar with the fundamental theorem sorry @William . So could you please write the full answer without using it once again? Sorry for the trouble and thanks so much! – Natalya May 15 '16 at 19:06
  • @Natalya I'm sure you know the fundamental theorem, you just don't know it by name. It states that if (I'm simplifying a lot here, since this is your first time) $$\frac{d}{dx} f(x) = g(x) \implies \int_a^b g(x) dx = f(b)-f(a)$$ (things that I'm skipping are that the function should be defined in that interval, $[a, b]$, continuous, and differentiable.) – Aritra Das May 15 '16 at 19:22
  • @Natalya Things are getting rough around here, so would you mind coming over to chat? http://chat.stackexchange.com/rooms/39821/http-math-stackexchange-com-questions-1786269-how-to-find-the-minimum-value-for – Aritra Das May 15 '16 at 19:30
  • @Natalya I already gave the "proof" without using Taylor series or the fundamental theorem of calculus -- just look at the graph I included of $\tan x$ versus $x$ -- it should be eminently clear from looking at the graph that $\tan x > x, \quad x\in (0, \frac{\pi}{2})$. The only non-calculus proof of that fact is geometric, and I have already shown it to you, so I really don't understand what more you could possibly want in an answer to your question. – Chill2Macht May 15 '16 at 19:36
  • @Natalya come here: http://chat.stackexchange.com/rooms/39821/http-math-stackexchange-com-questions-1786269-how-to-find-the-minimum-value-for – Aritra Das May 15 '16 at 19:37
  • To whomever downvoted this: how does this answer not answer the question? The graph answers it without calculus, and then I give it at least two additional proofs of the same fact. If OP doesn't want to accept a proof based on a graph, and also isn't able to understand any proof which doesn't use a graph, that isn't my fault -- in any case, I put a lot of time and effort into this answer, it is entirely correct, and answers the OP's original question. – Chill2Macht May 15 '16 at 19:41
  • @Natalya Did you look at the graph which precedes any calculus argument? – Chill2Macht May 15 '16 at 19:42
  • @William yes I did but I still need the proof not only the graph and the proof had to be without the fundamental math theorem and not include this notation you just used that I am not familiar with. Sorry once again I apologize for the all the work. – Natalya May 15 '16 at 19:45
  • Here you go: http://mathrefresher.blogspot.com/2006/08/sin-x-x-tan-x-for-x-in-02.html – Chill2Macht May 15 '16 at 19:46
  • @William the same happened to mine and everyone else's answers. And I have an idea as to who it might be. – Aritra Das May 15 '16 at 19:54
  • @Natalya If you don't know either the fundamental theorem of calculus or L'Hospital's rule, it's a stretch to say you actually know derivatives or limits. The only part of the question that required any serious proof was $\tan x > x$ for $x \in (0, \frac{\pi}{2})$, and you clearly were only looking for a graphical/trigonometric proof of this fact, not one involving calculus (or two to three, which I gave). So the tag "calculus" was clearly inappropriate since you weren't looking for a calculus answer and don't even know the subject properly. – Chill2Macht May 15 '16 at 20:13
  • If you ask another question on the internet, please clarify beforehand the level of your knowledge and what type of answer you are looking for in order to avoid making anyone feel that they wasted their time answering your question. Unlike your calculus teacher, we don't know beforehand how much you do or don't know. – Chill2Macht May 15 '16 at 20:13
1

The function is not bounded. It does not have a maxima or a minima. As $x\to\frac{\pi}{2}^-$, $F(x)\to\infty$ and as $x\to\frac{\pi}{2}^+$, $F(x)\to-\infty$.


We already know that for positive $x\in (0, \frac{\pi}{2})$, $$\sin x < x < \tan x$$ Since $x\not=0$, we can divide by $x$ $$\frac{\sin x}{x} < 1 < \frac{\tan x}{x}$$ Hence, $\frac{\tan x}{x}$ is always greater than $1$ for positive $x$. Hence, $f(x)>1$ for all positive $x$.

Moreover, $$f(x) =\frac{\tan x}{x} = \frac{\tan (-x)}{(-x)} = f(-x)$$

Hence, for negative $x$ also, $f(x)>1$

Thus for both positive and negative $x$, $f(x)>1$. On the other hand, at $x=0$, your function is defined to be $1$.

Hence, the minimum occurs at $x=0$ and the minimum value is $1$.

Chill2Macht
  • 20,920
Aritra Das
  • 3,528
  • Do you have a proof for $$\sin x < x \ \text{for } x \in (0, \frac{\pi}{2})$$ that doesn't use the fundamental theorem of calculus? – Chill2Macht May 15 '16 at 19:02
  • 2
    @William draw a unit circle and compare the area of triangle and arc. –  May 15 '16 at 19:03
  • That's not a proof though. And if you wanted to prove your comparison of the triangle and the arc, you would have to use integration and -- the fundamental theorem of calculus. – Chill2Macht May 15 '16 at 19:04
  • Why isn't that a proof? – Aritra Das May 15 '16 at 19:11
  • And what would I want to prove? Why would I use integration? The formulae for area of triange and arcs are pretty well known and standard. Why would I want to prove them? – Aritra Das May 15 '16 at 19:12
  • And finally, even if I did use integration to find the area, why would I use the fundamental theorem? The fundamental theorem only links the idea of derivatives to that of integrals. I don't see why I would use it to just do a definite integral. – Aritra Das May 15 '16 at 19:15
  • @William Another thing, the way you reasoned that to prove the comparison between the triangle and the arc, I would have to use the fundamental theorem of calculus, I can also reason that to do the integral itself you would require that $\sin x < x$. The worst part in your logic is that I didn't require $\sin x < x$ in my proof at all. – Aritra Das May 15 '16 at 19:27
  • I assumed $\sin x < x$ was necessary for the assumption that $\tan x > x$ on the same interval, although upon further thought that does not necessarily seem to be the case. – Chill2Macht May 15 '16 at 19:31
  • @AritraDas The argument that one would need to know a priori that $\sin x < x$ in order to compute the necessary integrals is incorrect however. – Chill2Macht May 15 '16 at 19:32
  • Finally one last thing -- how does one prove $\tan x > x$ on that interval without using the fundamental theorem of calculus? I already gave the graph showing this in my answer, but for some reason it was insufficient for OP. – Chill2Macht May 15 '16 at 19:34
  • @William To compute the integral, you would need to do a substitution $x=\sin t$ and $dx = \cos t dt$ where you would use that the derivative of $\sin x$ is $\cos x$ to show which (from the definition of derivatives) you would require that $\lim_{x\to0} \frac{\sin x}{x} = 1$ to do which you would need that $\sin x \le x$ – Aritra Das May 15 '16 at 19:34
  • @William I think we're spending too much time on a petty issue. Clearly the OP doesn't know the fundamental theorem. Thus it's best to stray away from calculus. However since the point $x=0$ is a local minimum, calculus is necessary. Since it has to be avoided, the only answer possible is look at the graph. Do you agree? – Aritra Das May 15 '16 at 19:36
  • I agree. What I don't understand is that I already included the graph showing this in my original answer for OP several hours ago. The only place where calculus was used such that OP didn't understand was in showing that $\tan x > x$, but this was already clear from the graph; I just wanted to give extra arguments to make it extra convincing. – Chill2Macht May 15 '16 at 19:39
  • @William good then it's settled. It was nice talking with you. – Aritra Das May 15 '16 at 19:40