1
$\begingroup$

$$\frac{d}{dx}\operatorname{arcsec}(x)=\frac{1}{x\sqrt{x^2-1}}\tag{1}$$

$$\int{\frac{1}{x\sqrt{x^2-1}}dx}=\operatorname{arcsec}(x)\tag{2}$$

$$\frac{d}{dx}{\arcsin(x)}=\frac{1}{\sqrt{1-x^2}}\tag{3}$$

$$\int{\frac{1}{\sqrt{1-x^2}}dx=\arcsin(x)}\tag{4},\quad |x|<1$$

Questions:

  1. For $(3)$, the condition should be $|x|<1$, shouldn't it?
  2. For $(1)$ and $(2)$, the condition should be $|x|>1$, shouldn't it?
$\endgroup$
4
  • 4
    $\begingroup$ Yes.Where's the doubt? $\endgroup$ Commented Jan 20, 2022 at 5:01
  • 1
    $\begingroup$ Yes. $\operatorname{arcsec}$ is usually defined for $|x|\geq 1$ but is not differentiable at $x=\pm 1$. Similarly, $\arcsin$ is defined for $|x|\leq 1$ but is not differentiable at $x=\pm 1$. So those conditions arise naturally. $\endgroup$ Commented Jan 20, 2022 at 5:01
  • $\begingroup$ @Gary Thanks for your comment kind sir. Isn't it easier to write just $|x|>1$, instead of writing $|x|\geq1,x\neq\pm1$? Why did you write it like that sir, am I missing something? $\endgroup$ Commented Jan 20, 2022 at 5:10
  • 1
    $\begingroup$ Seems you did not understand what I wanted to say. If you consider the differentiability of a function, at which points do you consider it? Well, at points of the domain of the function. Look at for example $\operatorname{arcsec} x$. The domain of this function is $|x|\geq 1$. So you have to check those points. One finds that the one-sided derivatives do not exist at the endpoints, so the formula for the derivative of $\operatorname{arcsec} x$ can be valid at most on $|x|>1$, and it is valid on the whole of it. In summary, it is natural that the domain for $(1)$ is $|x|>1$. $\endgroup$ Commented Jan 20, 2022 at 5:14

1 Answer 1

2
$\begingroup$

$$\int{\frac{1}{\sqrt{1-x^2}}dx=\arcsin(x)}\tag{4},\quad |x|<1$$ $$\frac{d}{dx}{\arcsin(x)}=\frac{1}{\sqrt{1-x^2}}\tag{3}$$

  1. For $(3)$, the condition should be $|x|<1$, shouldn't it?

Notice that

  • $\displaystyle\frac{1}{\sqrt{1-x^2}}$ is defined on only $(-1,1);$
  • $\arcsin(x)$ is defined on only $[-1,1];$
  • the derivative of $\arcsin(x)$ is undefined when $x=\pm1.$

Thus, your suggested condition is already implicit in $(4)$'s integrand (but no harm explicating it), while in $(3),$ I think that it's good practice to explicitly specify either $x\ne\pm1$ or, as you suggest, $|x|<1.$

$$\frac{d}{dx}\operatorname{arcsec}(x)=\frac{1}{x\sqrt{x^2-1}}\tag{1}$$ $$\int{\frac{1}{x\sqrt{x^2-1}}dx}=\operatorname{arcsec}(x)\tag{2}$$

  1. For $(1)$ and $(2)$, the condition should be $|x|>1$, shouldn't it?

$(1)$ is actually incorrect (at least for negative $x$), because in fact, $$\frac{\mathrm d}{\mathrm dx}\operatorname{arcsec}(x)=\frac{1}{\color{red}|x\color{red}|\sqrt{x^2-1}}\quad\left(|x|>1\right).\tag{1c}$$

$(2)$ is also actually incorrect, because in fact, $$\int\frac{\mathrm dx}{x\sqrt{x^2-1}}= \operatorname{arcsec}\color{red}|x\color{red}|+C.\tag{2c}$$

As for your question about the conditions:

$\operatorname{arcsec}(x)$ is defined, but has no derivative, at $1$ and $-1,$ so we exclude these points from its domain in $(1c);$

$(2c)$ doesn't require the condition $|x|>1$ explicitly since its integrand is defined only there anyway.

$\endgroup$
7
  • 3
    $\begingroup$ Technically, $$\int\frac{\mathrm{d}x}{x\sqrt{x^2-1}}=\operatorname{arcsec}\,|x|+C$$ is incorrect too. What would be correct is $$\int\frac{\mathrm{d}x}{x\sqrt{x^2-1}}=\begin{cases}\operatorname{arcsec}(-x)+C_0&x\lt-1\\\operatorname{arcsec}(x)+C_1&x\gt1\end{cases}.$$ $\endgroup$ Commented Jan 20, 2022 at 17:08
  • 2
    $\begingroup$ @Angel I wouldn't say that my sly compression is incorrect, only that it simplifies the truth (haha) in the same way that $\displaystyle\int \frac1x\,\mathrm dx=\ln|x|+C$ does. This simplification is harmless, since in any particular usage, only one copy of $C_0$ or $C_1$ is used, since the integration domain anyway oughtn't include $[-1,1].$ $\endgroup$ Commented Jan 20, 2022 at 17:15
  • 2
    $\begingroup$ I would argue that the simplification $$\int\frac1{x}\,\mathrm{d}x=\ln\,|x|$$ is also incorrect, though. I agree with you that the integration domain should not include $[-1,1],$ but here, we are strictly speaking about the antiderivatives, so we should provide the complete family of antiderivatives. Otherwise, if it was alright to not provide the complete family, then why would we bother with the constants of antidifferentiation? Those are also irrelevant when doing integration over an interval. $\endgroup$ Commented Jan 20, 2022 at 17:24
  • 1
    $\begingroup$ I disagree completely. Your argument for $(2)$ is reliant on the assumption you made during $(1),$ which is that antiderivatives are defined over an interval, but strictly speaking, this is false. Antiderivatives, much like derivatives themselves, are defined for functions whose domain is an open set, and an open set can be a disconnected set that is the union of disjoint open intervals, and so without itself being a single interval. In this case, there is no OR disjunction to make, because you are defining the antiderivative over the entire domain, not a connected subset thereof. $\endgroup$ Commented Jan 21, 2022 at 13:15
  • 1
    $\begingroup$ So in that case, your answer remains technically incorrect. $\endgroup$ Commented Jan 21, 2022 at 13:16

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.