$\newcommand{\R}{\mathbf{R}} \newcommand{\C}{\mathbf{C}} \newcommand{\A}{\mathcal{A}} \newcommand{\cF}{\mathcal{F}} \newcommand{\SPAN}{\text{span}} \newcommand{\B}{\mathcal{B}} \newcommand{\calL}{\mathcal{L}} \renewcommand{\u}{\mathbf{u}} \newcommand{\uu}{\mathbf{u}} \newcommand{\e}{\mathbf{e}} \newcommand{\vv}{\mathbf{v}} \newcommand{\w}{\mathbf{w}} \newcommand{\ww}{\mathbf{w}} \newcommand{\x}{\mathbf{x}} \newcommand{\xx}{\mathbf{x}} \newcommand{\y}{\mathbf{y}} \newcommand{\yy}{\mathbf{y}} \newcommand{\Cbar}{\overline{\mathbf{C}}} \newcommand{\Dbar}{\overline{\mathbf{D}}} \newcommand{\X}{\mathbf{X}} \newcommand{\Y}{\mathbf{Y}}$ \newcommand{\Xbar}{\widehat{\mathbf{X}}} \newcommand{\Ybar}{\widehat{\mathbf{Y}}} \newcommand{\zz}{\mathbf{z}} \renewcommand{\a}{\mathbf{a}} \renewcommand{\aa}{\mathbf{a}} \newcommand{\bb}{\mathbf{b}} \newcommand{\cc}{\mathbf{c}} \newcommand{\ee}{\mathbf{e}} \newcommand{\hh}{\mathbf{h}} \newcommand{\m}{\mathbf{m}} \newcommand{\0}{\mathbf{0}} \newcommand{\ve}[1]{\mathbf{#1}} \newcommand{\col}[1]{\ifmmode\begin{bmatrix}#1\end{bmatrix}\else $\begin{bmatrix}#1\end{bmatrix}$\fi} \newcommand{\scol}[1]{\left[\begin{smallmatrix}#1\end{smallmatrix}\right]} \newcommand{\rref}{\operatorname{rref}} \newcommand{\hide}[1]{{}} \newcommand{\proj}{\operatorname{\mathbf{Proj}}} \newcommand{\Span}{\operatorname{span}} \newcommand{\pd}[2]{\frac{\partial #1}{\partial #2}} \newcommand{\pdt}[2]{\tfrac{\partial #1}{\partial #2}} \newcommand{\pdd}[2]{\dfrac{\partial #1}{\partial #2}} \newcommand{\svdots}{\raisebox{3pt}{$\scalebox{.75}{\vdots}$}} \newcommand{\sddots}{\raisebox{3pt}{$\scalebox{.75}{$\ddots$}$}} \DeclareMathOperator{\Aut}{Aut} \DeclareMathOperator{\Char}{char} \DeclareMathOperator{\Cl}{Cl} \DeclareMathOperator{\codim}{codim} \DeclareMathOperator{\coker}{coker} \DeclareMathOperator{\disc}{disc} \DeclareMathOperator{\dist}{dist} \DeclareMathOperator{\Div}{Div} \DeclareMathOperator{\End}{End} \DeclareMathOperator{\Eth}{Eth} \DeclareMathOperator{\Frac}{Frac} \DeclareMathOperator{\Free}{Free} %\DeclareMathOperator{\frob}{frob} %\DeclareMathOperator{\Gal}{Gal} %\DeclareMathOperator{\genus}{genus} %\DeclareMathOperator{\Hecke}{Hecke} \DeclareMathOperator{\Hom}{Hom} %\DeclareMathOperator{\id}{id} %\DeclareMathOperator{\im}{im} \DeclareMathOperator{\lcm}{lcm} \DeclareMathOperator{\Mat}{Mat} \DeclareMathOperator{\modulo}{\medspace mod} \DeclareMathOperator{\Norm}{N} %\DeclareMathOperator{\nullity}{nullity} \DeclareMathOperator{\ord}{ord} \DeclareMathOperator{\Pic}{Pic} %\DeclareMathOperator{\rank}{rank} \DeclareMathOperator{\red}{red} \DeclareMathOperator{\res}{res} \DeclareMathOperator{\sgn}{sgn} %\DeclareMathOperator{\Span}{span} \DeclareMathOperator{\Spec}{Spec} \DeclareMathOperator{\Split}{Split} \DeclareMathOperator{\Sturm}{Sturm} \DeclareMathOperator{\Supp}{Supp} \DeclareMathOperator{\Tate}{Tate} \DeclareMathOperator{\tors}{tors} %\DeclareMathOperator{\tr}{tr} \DeclareMathOperator{\val}{val} \DeclareMathOperator{\Weil}{Weil} \DeclareMathOperator{\sech}{sech} \newcommand{\adjacent}{\leftrightarrow} \DeclareMathOperator{\GL}{GL} \DeclareMathOperator{\SL}{SL} \DeclareMathOperator{\PGL}{PGL} \DeclareMathOperator{\PSL}{PSL} \DeclareMathOperator{\SO}{SO} \newcommand{\cm}{\text{,}} %\newcommand{\pd}{\text{.}} \newcommand{\n}{\noindent} \newcommand{\Omicron}{\mathrm{O}} \newcommand{\Zeta}{\mathrm{Z}} \renewcommand{\div}{\mathop{\mathrm{div}}} \renewcommand{\Im}{\mathop{\mathrm{Im}}} \renewcommand{\Re}{\mathop{\mathrm{Re}}} \renewcommand{\ss}{\mathop{\mathrm{ss}}} \newcommand{\elliptic}{\mathop{\mathrm{ell}}} \newcommand{\new}{\mathop{\mathrm{new}}} \newcommand{\old}{\mathop{\mathrm{old}}} \newcommand{\Bs}{\boldsymbol} %\newcommand{\ds}{\displaystyle} %\newcommand{\f}{\mathfrak} \newcommand{\s}{\mathcal} %\newcommand{\A}{\mathbb{A}} %\newcommand{\C}{\mathbb{C}} \newcommand{\F}{\mathbb{F}} \newcommand{\Fpbar}{\bar{\mathbb{\F}}_p} \newcommand{\G}{\mathbb{G}} \newcommand{\Gm}{\mathbb{G}_{\mathrm{m}}} \newcommand{\N}{\mathbb{N}} \renewcommand{\P}{\mathbb{P}} \newcommand{\Q}{\mathbb{Q}} %\newcommand{\R}{\mathbb{R}} %\newcommand{\R}{\mathbf{R}} \newcommand{\T}{\mathbb{T}} \newcommand{\V}{\mathcal{V}} \newcommand{\Z}{\mathbb{Z}} \newcommand{\E}{\mathbf{E}} \renewcommand{\H}{\mathrm{H}} \newcommand{\M}{\mathbf{M}} \renewcommand{\S}{\mathbf{S}} \newcommand{\var}{\mathbf{Var}} \newcommand{\eps}{\varepsilon} \newcommand{\erf}{\operatorname{erf}} \newcommand{\rar}{\rightarrow} \newcommand{\lar}{\leftarrow} \newcommand{\hrar}{\hookrightarrow} \renewcommand{\iff}{\Longleftrightarrow} \newcommand{\xrar}{\xrightarrow} \newcommand{\rrar}{\longrightarrow} \newcommand{\mt}{\mapsto} \newcommand{\mmt}{\longmapsto} \newcommand{\angles}[1]{\langle #1\rangle} \newcommand{\ceiling}[1]{\lceil #1\rceil} \newcommand{\floor}[1]{\lfloor #1\rfloor} \newcommand{\set}[2]{\{\,#1\,\,|\,\,#2\,\}} \renewcommand{\emph}{\it} \renewcommand{\em}{\emph} $\newcommand{\pd}[2]{\frac{\partial #1}{\partial #2}}$

Chapter 18

Visualizing linear ODE’s with non-constant coefficients.

Bessel equations.

As introduced in the textbook, the Bessel equation is given by $$t^2 y'' + t y' + (t^2 - n^2) y = 0$$ (with $n = 0, 1, 2, \dots$).

Let's first focus on the Bessel equation with $n=0$: \begin{equation}\label{bessel0} t^2 y'' + t y' + t^2 y = 0,\,\,\,\mbox{ or equivalently } \,\,\, y'' + \frac{1}{t}y' + y = 0. \end{equation}

Using power series, we find a solution of the following form, which is called the Bessel function $J_0(t)$: $$J_0(t) = \sum_{k=0}^{\infty} \frac{(-1)^k}{2^{2k} (k!)^2} t^{2k} = 1 - \frac{1}{4}t^2 + \frac{1}{64}t^4 - \frac{1}{2304} t^6 + \dots .$$

It can be shown that the scalar multiples $c \, J_0(t)$ are the only solutions of $t^2 y'' + t y' + t^2 y = 0$ on $(0, +\infty)$ that remain bounded as $t \to 0^+$. We have $c\,J_0(t) \to c$ as $t \to 0^+$.

It is tempting to think that since $J_0(t)$ satisfies the ODE $y'' + (1/t)y' + y = 0$ (as on the right in the Bessel equation and for large $t$ perhaps the middle term $(1/t)y'$ is "negligible", the behavior of $J_0(t)$ for large $t$ should be approximated by some nonzero solution to $y'' + y = 0$; i.e., a nonzero linear combination of $\sin t$ and $\cos t$. But that guess is wrong! Such linear combinations $c_1 \cos t + c_2 \sin t$ are phase-shifted cosine waves $A \cos(t - \phi)$ up to a change in amplitude, and in particular they do not decay as $t \to +\infty$, whereas it can be shown that $J_0(t)$ decays like $1/\sqrt{t}$, and more precisely \begin{equation}\label{J0bigt} J_0(t) \approx \sqrt{\frac{2}{\pi t}} \cos(t - \pi/4) \end{equation} as $t \to +\infty$ (with error bounded by a constant multiple of $1/t^{3/2}$).

There is an analogous solution to $t^2 y'' + ty' + (t^2 - n^2) y = 0$ for positive integers $n$, denoted $J_n(t)$ and called the Bessel function of the "first kind" of order $n$. It is defined by a convergent power series $$J_n(t) = t^n \sum_{k=0}^{\infty} \frac{(-1)^k}{2^{2k+n}k!(k+n)!}t^{2k}$$ and satisfies an analogue of decaying for big $t$, with precise form depending on whether $n$ is even or odd: $$J_{2m}(t) \approx (-1)^m \sqrt{\frac{2}{\pi t}} \cos(t - \pi/4),$$ $$J_{2m+1}(t) \approx (-1)^m \sqrt{\frac{2}{\pi t}} \sin(t - \pi/4),$$ as $t \to +\infty$ (with error bounded by a constant multiple of $1/t^{3/2}$).

Let $J_{n,2r+n}(t) := t^n\sum_{k=0}^{r} \sum_{k=0}^{\infty} \frac{(-1)^k}{2^{2k+n}k!(k+n)!}t^{2k}$ be the sum up to the $(2r+n)$-th power term in the expansion of $J_n$.

Below we visualize the appriximations of the Bessel function $J_n$ both by $J_{n,2r+n}(t)$ and by the big $t$ asymptotic behavior ($(-1)^m\sqrt{\frac{2}{\pi t}} \cos(t - \pi/4)$ for even $n = 2m$ and $(-1)^m \sqrt{\frac{2}{\pi t}} \sin(t - \pi/4)$ for odd $n= 2m+1$). The red curve is this trigonometric function that approximates $J_n(t)$ for large $t$. You can change the visibility of a curve by clicking on its legend.

Order


Truncation $J_{n,2r+n}(t)$

Graph