5
$\begingroup$

I need to verify the value of the following integral $$ 4n(n-1)\int_0^1 \frac{1}{8t^3}\left[\frac{(2t-t^2)^{n+1}}{(n+1)}-\frac{t^{2n+2}}{n+1}-t^4\{\frac{(2t-t^2)^{n-1}}{n-1}-\frac{t^{2n-2}}{n-1} \} \right] dt.$$ The integrand (factor of $4n(n-1)$) included) is the pdf of certain random variable for $n\geq 3$ and hence I expect it be 1. If somebody can kindly put it into some computer algebra system like MATHEMATICA, I would be most obliged. I do not have access to any CAS software.

PS:-I do not know of any free CAS software. If there is any somebody may please share

$\endgroup$
  • $\begingroup$ Your integral equals $1$. See the proof below. $\endgroup$ – GH from MO Sep 8 at 6:20
  • 2
    $\begingroup$ I suppose that the round bracket in $t^4\{\frac{(2t-t^2)^{n-1}}{n-1})$ is redundant (the one on the right). But I did not want to edit this, since it is better if you clarify what is the intended function. I am referring to the current revision. $\endgroup$ – Martin Sleziak Sep 8 at 6:24
  • $\begingroup$ @MartinSleziak: I omitted the closing round bracket (which has no opening counterpart), and proved that the resulting expression equals $1$. See below. $\endgroup$ – GH from MO Sep 8 at 6:31
  • 1
    $\begingroup$ If it is a pdf, this itself proves that the integral equals 1, "expect" is underestimating. $\endgroup$ – Fedor Petrov Sep 8 at 8:14
  • 1
    $\begingroup$ @FedorPetrov oh! I didnt see it. Then wasnt it too trivial? $\endgroup$ – vidyarthi Sep 8 at 12:06
23
$\begingroup$

The integral can be rewritten as \begin{align*} I&=\frac{n(n-1)}{2}\int_0^1\frac{t^{n-2}(2-t)^{n+1}-t^{2n-1}}{n+1}-\frac{t^n(2-t)^{n-1}-t^{2n-1}}{n-1}\,dt\\[6pt] &=\frac{1}{2n+2}+\frac{n(n-1)}{2}\int_0^1\frac{t^{n-2}(2-t)^{n+1}}{n+1}-\frac{t^n(2-t)^{n-1}}{n-1}\,dt. \end{align*} Integrating by parts, we obtain $$\int_0^1\frac{t^{n-2}(2-t)^{n+1}}{n+1}\,dt=\frac{1}{n^2-1}+\int_0^1\frac{t^{n-1}(2-t)^n}{n-1}\,dt.$$ Therefore, \begin{align*} I&=\frac{1}{2}+\frac{n}{2}\int_0^1t^{n-1}(2-t)^n-t^n(2-t)^{n-1}\,dt\\[6pt] &=\frac{1}{2}+\frac{1}{2}\int_0^1(t^n(2-t)^n)'\,dt=\frac{1}{2}+\frac{1}{2}=1. \end{align*}

P.S. You can use SageMath and WolframAlpha for symbolic calculations. Both are free.

| cite | improve this answer | |
$\endgroup$
  • 2
    $\begingroup$ It seems that yesterday I inadvertently put a -1 to your answer. the only way to correct is making a trivial edit. sorry! $\endgroup$ – Pietro Majer Sep 9 at 8:52
  • $\begingroup$ @PietroMajer: This is very kind of you, thanks! $\endgroup$ – GH from MO Sep 9 at 15:11
6
$\begingroup$

It seems your conjecture is true. Mathematica gives the result $$ (1 + 4^n (-1 + n) n \mbox{Beta} [1/2, -1 + n, 2 + n] - 4^n n (1 + n) \mbox{Beta} [1/2, 1 + n, n])/(2 (1 + n)) $$ in terms of the incomplete Beta function, and putting in random integers $\geq 3$ always yields 1 (I haven't managed to get Mathematica to spit that out as a general result for arbitrary $n$).

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ I proved the conjecture in my response. $\endgroup$ – GH from MO Sep 8 at 6:33
  • 1
    $\begingroup$ @GHfromMO - yes, very nice, you have my upvote. We posted simultaneously. I considered deleting my answer, but decided against it, since at least I answered the OP's literal request :-D $\endgroup$ – Michael Engelhardt Sep 8 at 14:43
4
$\begingroup$

You can use CoCalc. For instance, type integral(x^2,x) and get 1/3*x^3

It also permits symbolic parameters.

Input:

f(x,n)=x^2+n

integral(f(x,n),x)

Output:

1/3*x^3+n*x

| cite | improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.