5
$\begingroup$

I've noticed it is not in Erdelyi or Gradshteyn, although the version with the sine replaced by a cosine is in Erdelyi (page 26 eq. 33). I've tried using the substitution $x=a \sinh z$ to avoid a square root.
I would also like to know the Fourier sine transform of the sine and cosine form.

$\endgroup$
4
  • $\begingroup$ Is bsqrt(a^2+x^2) $=b\sqrt{a^2+x^2}$? $\endgroup$ Commented Dec 18, 2020 at 20:15
  • 1
    $\begingroup$ I've tried to fix your formatting (you should use LaTeX), but please check that I got your integrand correctly stated. $\endgroup$ Commented Dec 18, 2020 at 20:33
  • $\begingroup$ Yes, thank you.It is now more legible.The ist authour is Erdelyi $\endgroup$
    – H.Davies
    Commented Dec 18, 2020 at 20:42
  • $\begingroup$ the Fourier sine transform is zero, isn't it? $\endgroup$ Commented Dec 18, 2020 at 21:49

1 Answer 1

4
$\begingroup$

This is an integral in Oberhettinger's Table of Fourier Transforms (page 27, with $r=\sqrt{a^2+x^2}$). It is not a very helpful result, but it does indicate why a fully closed-form expression will not be forthcoming (you would need the indefinite integral of a Bessel function of argument $\sqrt{1-x^2}$).

There is a typo in the formula, a numerical check does not match. For $y<b$ a square root is missing in the argument of the Bessel $Y_0$ function, the following expressions do pass a numerical check:

$$\int_{0}^\infty \frac{\sin\left(b\sqrt{a^2+x^2}\right)}{a^2+x^2}\cos yx\,dx =\frac{e^{-a y} \,\text{Ei}(a y)-e^ {ay} \,\text{Ei}(-a y)}{2 a}$$ $$\qquad\qquad-\frac{\pi}{2} \int_y^b Y_0\left(a \sqrt{t^2-y^2}\right) \, dt,\;\;0\leq y<b,$$ $$\int_{0}^\infty \frac{\sin\left(b\sqrt{a^2+x^2}\right)}{a^2+x^2}\cos yx\,dx =\int_0^b K_0\left(a \sqrt{y^2-t^2}\right) \, dt,\;\;y>b.$$

Notes:
• As a further check, note that the derivative with respect to $b$ gives formula 3.876.2 of Gradshteyn.
• The corresponding formulas in Erdelyi contain several typo's. These have been corrected in Gradshteyn, the formulas in question have a label $^6$.
• The integral also implies the identity $$\int_0^1 K_0\left(\sqrt{1-t^2}\right)\,dt=\tfrac{1}{2e} \text{Ei}(1)-\tfrac{e}{2} \text{Ei}(-1).$$

$\endgroup$
1
  • 1
    $\begingroup$ Your reply (and notes) very welcome.I had hoped the answer had the simplicity of the cosine version.Nevertheless, my mind is relieved.(Apparently Mathematica haven't seen this result yet!) $\endgroup$
    – H.Davies
    Commented Dec 19, 2020 at 15:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .