[Math] Integral solution of a differential equation

definite integralsordinary differential equations

I'm having difficulty verifying that $$K(x)=\int_{1}^{\infty}\frac{e^{-xt}}{\sqrt{t^{2}-1}}\, dt$$

satisfies the differential equation $$f^{\prime\prime}\left(x\right)+\frac{1}{x}f^{\prime}\left(x\right)-f\left(x\right)=0$$

I've tried evaluating each of the terms in the differential equation using $K(x)$ in place of $f(x)$ and then summing them as in the differential equation to see if it equaled zero, but when I do that, I always wind up with the exact same sum that I started with. Can anyone help?

Best Answer

Let $g(t,x) = \exp(-t x) \left(1-t^2\right)^{-1/2}$. Note that $$ \frac{\partial^2}{\partial x^2} g(t,x) + \frac{1}{x} \frac{\partial}{\partial x} g(t,x) - g(t,x) \stackrel{\text{evaluate}}{=} \exp(-t x) \frac{-t-x(1-t^2)}{\sqrt{1-t^2}} = \frac{\partial}{\partial t} \left( \exp(-t x) \sqrt{1-t^2} \right) $$ Now $f(x) = \int_1^\infty g(t,x) \mathrm{d}t$ giving: $$ f^{\prime\prime}(x) + \frac{1}{x} f^\prime(x) - f(x) = \left. \exp(-t x) \sqrt{1-t^2} \right|_{t \to 1^+}^{t \to \infty} $$ The right-hand-side vanishes for $\Re(x)>0$, which is incidentally the same condition when $\int_1^\infty g(t,x)\,\mathrm{d}t$ converges, thus establishing the claim.

By the way, function $K(x)$ defined by the integral is the modified Bessel function of the second kind of order $0$.

Related Question