I want to show that the derivative of integral formulation of zeroth order Bessel function of second kind $K_{0}(z)$, with respect to argument $z$, is equal to minus the integral formulation of first order Bessel function of second kind $K_{1}(z)$, i.e.

\begin{equation} \frac{d}{dz} K_{0}(z) = -K_{1}(z) \end{equation}

I use that \begin{equation} K_{0}(z) = \int\limits_{0}^{\infty} \frac{\cos{(zt)}}{\sqrt{t^2 + 1}}\,dt \end{equation} and \begin{equation} K_{1}(z) = z\int\limits_{0}^{\infty} \frac{\cos{(t)}}{(t^2 + z^2)^{3/2}}\,dt \end{equation} from http://mathworld.wolfram.com/ModifiedBesselFunctionoftheSecondKind.html.

So far I have only got that \begin{equation} \frac{d}{dz} K_{0}(z) = \int\limits_{0}^{\infty} \frac{-\sin{(zt)}}{\sqrt{t^2 + 1}}t\,dt = \int\limits_{0}^{\infty} \frac{zt\cos{(zt)}-\sin{(zt)}}{z^{2}(t^2 + 1)^{3/2}}t\,dt, \end{equation} but from here I do not know which path to pursue. Any guidance would be most welcome.

  • 1
    $\begingroup$ You're looking to get $\cos(t)$ inside the trig functions, so try a substitution $u=zt$ $\endgroup$ – B. Mehta May 19 '17 at 13:13
  • $\begingroup$ Also, where did that second integral come from? $\endgroup$ – B. Mehta May 19 '17 at 13:14
  • $\begingroup$ Via integrations by parts. The other part is \begin{equation} \frac{zt\cos{(zt)}-\sin{(zt)}}{z^2t(t^2+1)^{1/2}} \end{equation} which is goes to zero for t going to zero or infinity. $\endgroup$ – Raibyo May 19 '17 at 13:23

If $s=zt$, the $K_0$ integral becomes $$ \int_0^{\infty} \frac{\cos{s}}{z\sqrt{1+s^2/z^2}} \, ds = \int_0^{\infty} \frac{\cos{s}}{\sqrt{z^2+s^2}} \, ds $$ for $z$ nonnegative (fine since we expect a branch cut anyway). Now you can just differentiate with respect to $z$ to get the $K_1$ formula, since $$ \frac{\partial}{\partial z} \frac{1}{\sqrt{z^2+s^2}} = -\frac{z}{(z^2+s^2)^{3/2}}. $$

  • $\begingroup$ With that change of variables, don't I get $$\int_0^{\infty} \frac{\cos{s(z)}}{\sqrt{z^2+s(z)^2}} \, ds $$, which gives me another expression when differentiating with respect to z? $\endgroup$ – Raibyo May 19 '17 at 13:29
  • $\begingroup$ I don't think so. $t=s/z$, so $dt=ds/z$, and the original integral was $\int_0^{\infty} \frac{\cos{zt}}{\sqrt{1+t^2}} \, dt$? $\endgroup$ – Chappers May 19 '17 at 14:31

Use this definition

$$ K_{\alpha }(x)=\int _{0}^{\infty }\exp(-x\cosh t)\cosh(\alpha t)\,dt $$

So that

$$ \frac{d}{dx}K_0(x) = \int _{0}^{\infty }\frac{d}{dt}\exp(-x\cosh t)\,dt = -\int _{0}^{\infty }\exp(-x\cosh t)\cosh(t)\,dt = -K_1(x) $$

  • $\begingroup$ I guess you mean to differentiate w.r.t. x, but thanks! I did this earlier, so the problem is really to go between the expressions I have stated. $\endgroup$ – Raibyo May 19 '17 at 13:31

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.