4

I am looking to find an explicit integral formula for $\operatorname{Bi}(x)$ via solving Airy's equation:

$${\mathrm d^2y\over\mathrm dx^2}-xy=0\tag1$$

Currently I am able to solve for $\operatorname{Ai}(x)$ using the means of Fourier transform:

Let $F(\omega)$ denote the Fourier transform of $y(x)$. If we take Fourier transform on both side of (1), then

$$ -\omega^2F(\omega)-iF'(\omega)=0\tag2 $$

Solving (2) yields $$ F(\omega)=F(0)e^{i\omega^3/3}\tag3 $$

Take inverse Fourier transform on (3) gives

$$ y={F(0)\over2\pi}\int_{-\infty}^\infty e^{i(\omega x+\omega^3/3)}\mathrm d\omega $$

Due to the fact that $\int_{-\infty}^\infty f(x)\mathrm dx=\int_0^\infty[f(x)+f(-x)]\mathrm dx$, we are able to get rid of the complex exponential:

$$ y={F(0)\over\pi}\int_0^\infty\cos\left(\omega x+{\omega^3\over3}\right)\mathrm d\omega\tag4 $$

and (4) is identical to $y=F(0)\operatorname{Ai}(x)$. Fourier transform is only valid for square-integrable funcntion, so the particular solution $\operatorname{Ai}(x)$ is only the square-integrable branch of the more general solution:

$$ y=C_1\operatorname{Ai}(x)+C_2\operatorname{Bi}(x) $$

However, I would like to derive an explicit integral formula for $\operatorname{Bi}(x)$, so I plug $y=f(x)\operatorname{Ai}(x)$ back into (1) and obtain

$$ f''(x)\operatorname{Ai}(x)+2f'(x)\operatorname{Ai}'(x)+f(x)\operatorname{Ai}''(x)-xf(x)\operatorname{Ai}(x)=0 \\ f(x)[\operatorname{Ai}''(x)-x\operatorname{Ai}(x)]+f''(x)\operatorname{Ai}(x)+f'(x)\operatorname{Ai}'(x)=0 $$ $$ f''(x)\operatorname{Ai}(x)+2f'(x)\operatorname{Ai}'(x)=0\tag5 $$

Via some basic algebraic operation, I arrived at

$$ {f''(x)\over f'(x)}=-2{\operatorname{Ai}'(x)\over\operatorname{Ai}(x)}\tag6 $$

Integrate and exponentiate on both side of (6) gives

$$ f'(x)={C\over\operatorname{Ai}^2(x)} $$

and eventually I got stuck on

$$ f(x)=C\int{\mathrm dx\over\operatorname{Ai}^2(x)} $$

Although this integral became a simple business if we made use of the Wronskian of $\operatorname{Ai}(x)$ and $\operatorname{Bi}(x)$, but I wonder if it is possible for me to obtain an explicit integral formula $\operatorname{Bi}(x)$ via this integral. Alternatively, could somebody provide some other ways for me to find this integral formula

TravorLZH
  • 7,786
  • 1
    Have you already looked at the DLMF and references therein? (Pay particular attention to the notes in the ⓘ icons next to the formulae and the start of each section.) – J. M. ain't a mathematician Sep 11 '20 at 08:08
  • @J.M.isn'tamathematician Thanks for your comment, I looked up your link and even read a few pages of the book Airy Functions and Applications to Physics it referenced, but I only found that the author is manipulating the definition of $\operatorname{Ai}(x)$ to obtain $\operatorname{Bi}(x)$. I wonder if it is possible to derive $\operatorname{Bi}(x)$ more intuitively. – TravorLZH Sep 11 '20 at 12:02

0 Answers0