-1

I'm trying to solve the following integral: $$\int_0^1 \frac{1}{\sqrt{1+x^4}} \,dx$$

Here's a link to the original question:

https://photos.app.goo.gl/mfTe649pzgKHeUKy7

I've tried to substitute $x$ as $\tan t$ and then $2t$ as $u$. Eventually I end up with:

\begin{align*} \int_0^{\pi / 2} (\cos u)^{-1/2} \, du \end{align*}

Ѕᴀᴀᴅ
  • 34,263
  • Yes, I guess I can, but what have you tried? – primes.against.humanity Apr 06 '20 at 11:51
  • 2
    The question is unclear to me. What does it mean for an integral to belong to an interval? Do you mean that $\int_a^b f(x)dx\in [a,b]$? – Keen-ameteur Apr 06 '20 at 11:53
  • Thanks to whoever edited the question. I really don't don't know how to use this app correctly, sorry. – Kevin Andrew Apr 06 '20 at 11:58
  • @Keen-ameteur Thats how the question is worded in my text book. I was unable to post the picture of the question (an error was occurring) so I decided to write the question down. As far as I understand it, it's asking you to find the upper n lower limit after substitution but not subtract the 2. – Kevin Andrew Apr 06 '20 at 11:59
  • @primes.against.humanity Sorry about the question. It's been years since I used the site and I wasn't sure how to word the title correctly. I've tried multiple substitutions but none of them seem to work. – Kevin Andrew Apr 06 '20 at 12:00
  • Not down voting would help a lot. Thanks:) – Kevin Andrew Apr 06 '20 at 12:03
  • Perhaps there is some context before or after the question in the book clarifying the meaning of this statement then? – Keen-ameteur Apr 06 '20 at 12:07
  • @Keen-ameteur nope. Its an exercise. That question 11of 40. Finished all of except 5 or so. So can you please atleast show the integral? – Kevin Andrew Apr 06 '20 at 12:26
  • @primes.against.humanity if you're asking for a progress report, I don't have one. It's unnecessary for you to ask me that. If you dont intend to answer the question but only judge from your high hill, I request you to keep scrolling. Its okay if you feel this question is lame and if its real easy for you. I couldn't figure out a way to solve it after trying multiple times. Downvote n move along if u aren't gonna help. – Kevin Andrew Apr 06 '20 at 12:33
  • @KevinAndrew Wow. I did not mean for my comment to seem arrogant. I was trying to point out that the title of your question is more in the style of riddles posted on facebook, rather than actually describing what your question is about and living up to the standards of this site. Your question gives the impression that you haven't done anything to try to solve this problem by yourself. In that case, why would you expect anybody to solve it for you? If you have made an honest attempt at the problem, you should let us all know by describing what you tried and where you got stuck. – primes.against.humanity Apr 06 '20 at 13:15
  • @primes.against.humanity all my attempts led me to complex equations and I knew it was wrong. I could barely make a scratch without making the sum more complicated than it is. I only tried substitution and taking out x and x squared from the root n then substituting. But I couldn't do anything past that. – Kevin Andrew Apr 06 '20 at 13:24
  • @primes.against.humanity as for the title, I haven't seen the other posts on this site. I remembered that I could post questions on this site as this sum in particular had been bugging me for a couple of days. After checking a pcm doubts forum known as doubtnut(basically just for 11th n 12th graders as far as I know) n finding no solution I tot I'd post the question here. So do tell me what an appropriate title would be – Kevin Andrew Apr 06 '20 at 13:28
  • @KevinAndrew Alright. If you want the question to be re-opened, you should add some details and context to it and clarify what you are actually asking, as a lot of people (including myself) seem to be confused about that point. I wouldn't mind suggesting a new title, but since I don't really understand what the question is about, I don't feel I can. EDIT: It sounds as though a big part of your problem is to compute the integral. Why not make the question about this and call it "How to compute [integral]?" and include what you have tried to compute it. – primes.against.humanity Apr 06 '20 at 13:28
  • @primes.against.humanity is that enough? – Kevin Andrew Apr 06 '20 at 13:59
  • @KevinAndrew Much better. However, you want to compute an integral, not a sum. I edited your question to fix this and some other stuff. Hopefully your question will now be re-opened :) – primes.against.humanity Apr 06 '20 at 14:03
  • @primes.against.humanity thank you so much. Im sorry for lashing out earlier. How do I write the math eqn the way the other eqn is written? Someone edited it to look that way. I checked some other stackexchange answers n tried 1 method but I still couldn't figure it out. – Kevin Andrew Apr 06 '20 at 14:12
  • @KevinAndrew Don't worry about it! You can check out the following MathJax tutorial, which will show you how to write math that looks pretty $\mathbb{N}$eat. https://math.meta.stackexchange.com/questions/5020/mathjax-basic-tutorial-and-quick-reference – primes.against.humanity Apr 06 '20 at 14:16
  • @primes.against.humanity I read thru it. Excellent answer. Very detailed. Thank you. Is it based on some coding language? (I only know java and my knowledge even in that is quite limited). Also is there a moderator who will open closed questions or do I have to submit my question for review or something? And is there a way to follow someone on stackexchange? – Kevin Andrew Apr 06 '20 at 14:29
  • @KevinAndrew Yes, it's called LaTeX. I think that five people of 3000+ reputation need to vote to re-open closed questions, but I don't know if this is something to count on. I don't think you can follow anybody here, but I'm not exactly sure. I wrote a solution to your problem, but I can't post it as the question is closed. I can send you a mail, though, if you're interested. – primes.against.humanity Apr 06 '20 at 15:49
  • @primes.against.humanity that would be great. Is it safe to post your email ids online tho? – Kevin Andrew Apr 06 '20 at 15:52
  • @primes.against.humanity or should I just ask this same question again? Can I even? – Kevin Andrew Apr 06 '20 at 15:52
  • @primes.against.humanity https://math.stackexchange.com/q/3612625/768789 – Kevin Andrew Apr 06 '20 at 15:56
  • Your conversion using $x=\tan\frac{u}{2}$ is incorrect: it should obtain the unhelpful$$\int_0^1\frac{dx}{\sqrt{1+x^4}}=\frac12\int_0^{\pi/2}\frac{\sec^2\frac{u}{2}du}{\sqrt{1+\tan^4\frac{u}{2}}}.$$ – J.G. Apr 11 '20 at 07:43

3 Answers3

1

On the interval $[0,1]$ $$1\geq1-x^8$$ Therefore $$\frac{1}{1+x^4}\geq1-x^4=\left(1+x^2\right)\left(1-x^2\right)\geq1-x^2$$ Therefore $$\frac{1}{\sqrt{1+x^4}}\geq\sqrt{1-x^2}$$ Therefore $$\int_0^1\frac{1}{\sqrt{1+x^4}}dx\geq\int_0^1\sqrt{1-x^2}dx=\frac{\pi}{4}>\frac34$$ Also, it is clear the integral in question is less than $1$, since $\frac{1}{\sqrt{1+x^4}}\leq1$, and therefore $$\int_0^1\frac{1}{\sqrt{1+x^4}}dx\leq\int_0^11\,dx=1$$

So $\int_0^1\frac{1}{\sqrt{1+x^4}}dx$ is in $\left[\frac34,1\right]$. There is no smaller interval among the options listed in the multiple choice version of the question that is linked.

2'5 9'2
  • 54,717
0

You don't need to actually evaluate the integral to answer your question.

It is clear that $\frac{1}{\sqrt{1+x^4}} ≤ 1$ since $x^4 ≥ 0 \Rightarrow 1+x^4 ≥ 1$.

For all real $x$ in the interval $[0,1]$, $1 + x^4 ≤ 1+x^2$, so $\frac{1}{\sqrt{1+x^4}} ≥ \frac{1}{\sqrt{1+x^2}}.$

We can use the fact that $\int \frac{1}{\sqrt{1+x^2}} \ \mathrm d x = \sinh^{-1} x $, so: $$\int_0^1 \frac{1}{\sqrt{1+x^2}} \mathrm d x = \left[\sinh^{-1} x\right]_0^1 = \sinh^{-1} 1 \approx 0.881.$$

Therefore the value of the integral lies in $\left[\frac{3}{4}, 1 \right]$, so option $4$ is correct.


Here is a proof that $\int \frac{1}{\sqrt{1+x^2}} \ \mathrm d x = \sinh^{-1} x $:

Substitute $x = \tan u, \mathrm d x = \sec^2 u \ \mathrm d u$:

$$\int \frac{1}{\sqrt{1+\tan^2 u}} \ \sec^2 u \ \mathrm d u$$ $$=\int \sec u \ \mathrm d u$$ $$=\ln(\tan u + \sec u)$$

Now draw a right triangle with legs $1, \tan u$ and hypotenuse $\sqrt{1+\tan^2 u}$. We find that $\sec u = \frac{\sqrt{1+\tan^2 u}}{1} = \sqrt{1+\tan^2 u} = \sqrt{1+x^2}$.

We thus have: $$=\ln(x + \sqrt{1+x^2})$$ $$=\sinh^{-1} x$$

according to the definition of $\sinh^{-1} x$. This can also be shown by letting $y = \sinh x = \frac{e^x+e^{-x}}{2}$, solving for $x$ using the quadratic formula then rearranging (the proof of the last step is widely available online such as Yahoo Answers).

Toby Mak
  • 16,827
0

Moving on from your last integral, there exists a function:

$$\beta(x,y) := 2\int_0^{\pi/2} \sin^{2x-1}(x)\cos^{2y-1}dx$$

So, with the exponent over the cosine being -1/2 and the exponent over the sine being technically 0, we can conclude $x$ must be $\frac{1}{4}$ and $y$ must be $\frac{1}{2}$. Thus, if we let your integral be $I$,

$$I = \frac{1}{2}\beta(\frac{1}{4},\frac{1}{2})$$

An identity exists such that:

$$\beta(x,y) = \frac{\Gamma(x)\Gamma(y)}{\Gamma(x+y)}$$

Therefore,

$$I = \frac{\Gamma\bigl(\frac{1}{4}\bigr)\Gamma\bigl(\frac{1}{2}\bigr)}{2\Gamma\bigl(\frac{3}{4}\bigr)}$$

Note though, I typed the integral with the reciprocal of the square root of the cuartic into wolfram alpha and it did not match that answer. So maybe you made an error in the substitution or I’m missing something in your work?