1

Reading Schwartz's quantum field theory book p.220, I came across certain integral : he calculated commutator for a scalar field as

$$[\phi(x) , \phi(y)] = \frac{-i}{2\pi^{2}}\int^{\infty}_{0} q^{2}dq\frac{\operatorname{sin}(\sqrt{q^{2}+m^{2}}t)}{\sqrt{q^{2}+m^2}} \frac{\operatorname{sin}(qr)}{qr} =: iD(t,r)$$.

Whatever the physical content, let's focus on this integration. We are faced with the problem of representing the above integral in terms of the Besel function. He(the author) wrote that "For $m\neq 0$, we can find an exact expression for $D(t,r)$ in terms of the Bessel functions $\mathcal{J_0}(x)$ :

$$D(t,r)=\frac{1}{4\pi r} \frac{\partial}{\partial r} \begin{cases} \mathcal{J}_0(m\sqrt{t^2-r^2)}, & t >r \\ 0, & r >t >-r \\ -\mathcal{J}_0(m\sqrt{t^2-r^2)}, & t<-r \\ \end{cases} $$."

And how such expression can be deduced? I can't do computation at all. In fact I found a same question someone posted but I don't know how to continue calculation to next step : Rewrite integral in terms of Bessel functions. I think that I am begginer to integration involving trigonometric functions-combined as somewhat complex manner- and its expression as a Bessel function. Can anyone help?

My first attempt is as follows. In the linked question, I understood upto next step : i.e., Let $q=m\sinh u$ ,

Then $dq=(m\cosh u )du$. So,

$$-D(t,r) = \int_{0}^{\infty}\dfrac{q^2}{2\pi^2}\dfrac{\sin(\sqrt{q^2+m^2}t)}{\sqrt{q^2+m^2}}\dfrac{\sin (qr)}{qr}~dq$$

$$=\int_{0}^{\infty}\dfrac{m^2\sinh^2u}{2\pi^2}\dfrac{\sin\left(t\sqrt{m^2\sinh^2u+m^2}\right)}{\sqrt{m^2\sinh^2u+m^2}}\dfrac{\sin(rm\sinh u)}{rm\sinh u}~(m\cosh u)du$$

$$=\dfrac{m}{2\pi^2r}\int_{0}^{\infty}\sin(mr\sinh u)\sin(mt\cosh u)\sinh u~du$$

$$=\dfrac{m}{4\pi^2r}\int_{0}^{\infty}\cos(mr\sinh u-mt\cosh u)\sinh u~du-\dfrac{m}{4\pi^2r}\int_{0}^{\infty}\cos(mr\sinh u+mt\cosh u)\sinh u~du$$.

On the other hand, note that the Bessel function can be represented as :

$$\mathcal{J}_0(x) = \frac{2}{\pi}\int_{0}^{\infty}\operatorname{sin}(x\operatorname{cosh}t)dt $$ for $ x>0$. ( c.f. 10.9.9 in https://dlmf.nist.gov/10.9#i )

Let $t>r$. Then

$$ \frac{\partial}{\partial r}\mathcal{J}_0(m\sqrt{t^2-r^2})= \frac{2}{\pi}\int_{0}^{\infty}\frac{\partial}{\partial r}\operatorname{sin}(m\sqrt{t^2-r^2}\operatorname{cosh}t)dt$$

$$ = \frac{2}{\pi}\int_{0}^{\infty}\operatorname{cos}(m\sqrt{t^2-r^2}\operatorname{cosh}t) \frac{\partial}{\partial r}(m\sqrt{t^2-r^2}\operatorname{cosh}t) dt $$ $$ = \frac{2}{\pi}\int_{0}^{\infty}\operatorname{cos}(m\sqrt{t^2-r^2}\operatorname{cosh}t)(m\operatorname{cosh}t)\frac{1}{\sqrt{t^2-r^2}}(-2r) dt $$

(Correct calculation?) Then how can we connect two calculations about objects $D(t,r)$ and $\frac{1}{4\pi r} \frac{\partial}{\partial r}\mathcal{J}_0(m\sqrt{t^2-r^2})$ ? I don't know how to proceed to next step of calculation. Can anyone help?

Plantation
  • 2,417
  • 1
    See the question here. – Random Variable Mar 01 '23 at 02:37
  • O.K. Thank you. I'm reffering your link. And until now, I can't completely apply the argument in the link to our situation similarly. I edited my question to include present progress. Perhaps, can you review my edited question and give me an advice? – Plantation Mar 01 '23 at 04:48
  • 1
    I was suggesting differentiating the other integral with respect to the parameter $b$. But the integral you're asking about is not even convergent. The author is assigning a value to a divergent integral. – Random Variable Mar 01 '23 at 05:14
  • @Random Variable : Can you express the term "suggesting differentiation of the 'other' integral with respect to the parameter $b$ more concretely and in detail? What that points out? The parameter $b$ is $r$ in our situation ? As 'the author', you mean Schwartz ? – Plantation Mar 01 '23 at 05:33
  • @Random Variable : This integration looks quite difficult :) – Plantation Mar 01 '23 at 05:35
  • 1
    Yes, I mean Schwartz. Schwartz seems to basically be saying that $$ \begin{align} \frac{\partial}{\partial r} \int_{0}^{\infty} \mathrm d q , \frac{\sin (t \sqrt{q^2 + m^2})}{\sqrt{q^2 + m^2}} , \cos (qr) &= \int_{0}^{\infty} \frac{\partial}{\partial r} , \mathrm d q , \frac{\sin (t \sqrt{q^2 + m^2})}{\sqrt{q^2 + m^2}} , \cos (qr) \ &= -\int_{0}^{\infty} \mathrm d q , \frac{q\sin (p \sqrt{q^2 + m^2})}{\sqrt{q^2 + m^2}} , \sin (qr) . \end{align}$$ The integral on the right does not converge in the traditional sense. – Random Variable Mar 01 '23 at 06:20
  • O.K. Thanks for pointing out~. Perhaps, does that integration always diverges regardless of conditions $t>r$, $r>t>-r$, $t<-r$? And..why did the author assign the value $-\mathcal{J}_0(m\sqrt{t^2-r^2})$ to $D(t,r)$ at $t<-r$? Can we show that similarly as your comment? – Plantation Mar 01 '23 at 06:56
  • 1
    It would converge if $t=0$ because then the integrand is zero. With regard to your second question, if $t$ is negative, then $\sin(t\sqrt{q^{2}+m^{2}}) = - \sin(t \sqrt{q^{2}+m^{2}})$. – Random Variable Mar 01 '23 at 07:50
  • @Random Variable : O.K. Thanks for comment.~ It's valuable. :) – Plantation Mar 01 '23 at 08:14

0 Answers0