368 chapter? Two- and Three-Dimensional Boundary Value Problems in problems which include the axis r = 0 in the range over which the solution is to apply. The solution to the 2 equation (7) when 7? = —7*is Z = Cjsin re + Cheos 72 18) ‘Summarizing, ether of the following forms satisfies Laplace's equation in the two cylindrical coordinates r and =: (7, 2) = [CTA + CANQTAILCs sinh Te + Cy cosh Ts] (19) G(r, 2) = [Ciel rr) + CAKo(AIIIC} sin re + Ci c0s 72] 20) ‘As was the case with the rectangular harmonics, the two forms are not really different since (19) includes (20) if Tis allowed to become imaginary, but the two separate ways of writing the solution are useful, as will be demonstrated in later examples. The case ‘with no assumed symmetries is discussed in the following section. 7.14 BESSEL FUNCTIONS In Sec. 7.13 an example of a Bessel function was shown as solution ofthe differential equation 7.13(8) which describes the radial variations in Laplace's equation for axially symmetric fields where a product solution is assumed. This i just one of a whole family of functions which are solutions of the general Bessel differential equation. Bessel Functions with Real Arguments For certain problems, as, for example, the solution for field between the two halves of longitudinally split eyliner, it may ‘be necessary to retain the @ variations in the equation. The solution may be assumed in product form again, RZ, where R is @ funtion ofr alone, F,of alone, and Z of alone, Z has solutions in hyperbolic functions as before, and F may also be satisfied by sinusoids: Z = CcoshTs + Dsioh Te w Fg = Ecos vb + F sin vs @ ‘The differential equation for & is then slightly diferent from the zero-order Bessel equation obtained previously: ee rs Itis apparent at once that Eq, 7.13(8) is special case ofthis more general equation, that is, v= OA series solution tothe general equation cartied through asin Sec. 7.13 ‘shows that the function defined by the series =} pray Goes 2 miP(v + m+ 1) 0 e ® is a solution to the equation. T(w + m+ 1)isthe gamma function of (v + m + 1) and, for vintegral, is equivalent to the factorial of (v + m). Also for v nonintegral, values of this gamma function are744 Bessel Functions 369 o ‘Fo, 7.14 (a) Bessel functions ofthe fist kind. (b) Bessel functions ofthe second kind, tabulated, If is an integer n, 5 uray Jy = 5 Crary = Ba Itcan be shown that J_, = (—1)"V,. A few of these functions are plotted in Fig. 7.144. ‘Similarly, a second independent solution’? to the equation is, cos va, (Tr) fa) oy Narn = © "8 Ity is nonintegral. J_, snot nearly related fo J,.and itt then proper to use elher J. ‘orN, as he second solution: fos Integral. N, must be used. Equation (6/8 indeterminaie {or integral but Is subject fo evaluation by usual methods.370 Chapter Two- and Three-Dimensional Boundary Value Problems — 1)'N,. AS may be noted in Fig. 7.14 these are infinite atthe origin. A. complete solution to (3) may be written R= ALATr) + BNAT) o ‘The constant v is known asthe order ofthe equation. J, i then called a Bessel function of first kind, oder v; N, is a Bessel function of second kind, order v. Of most interest for this chapter are cases in which v = n, an integer. It is useful to keep in mind that, in the physical problem considered here, vis the numberof radians ofthe sinusoidal variation ofthe potential per radian of angle about the axis. ‘The functions J,(v) and N,(v) are tabulated in the references.'*'* Some care should be observed in using these references, for there is a wide variation in notation for the second solution, and not all the functions used are equivalent, since they differ in the Values of arbitrary constans Selected forthe series. The N,(v) is chosen here because itis the form most common in current mathematical physics and also the form most ‘commonly tabulated, Of course, iis quite proper to use any one ofthe second solutions throughout a given problem, since all the differences will be absorbed in the arbitrary constants ofthe problem, and the same final numerical result will be obtained; but i is necessary tobe consistent in the use of only one of these throughout any given analysis. Itis of interest to observe the similarity between (3) and the simple harmonic equa- tion, the solution of which are sinusoids. The difference between these two differential equations lies inthe term (1/7) dR /dr) which produces its major effec as r —> 0. Note that for regions far removed from the axis as, fr example, near the outer edge of Fig. 1.19a, the region bounded by surfaces of a cylindrical coordinate system approximates 4 cube. For these reasons, it may be expected that, away from the origin, the Bessel functions ae similar to sinusoids. That this is true may be seen in Figs. 7.14a and b, For large values of the arguments, the Bessel functions approach sinusoids with mag- nitude decreasing a the square root of radius, as will be seen inthe asymptotic forms, Eqs. 7.15(1) and 7.15(2). Hankel Functions It is sometimes convenient to take solutions to the simple har- ‘monic equation in the form of complex exponentials rather than sinusoids. That is, the solution of os Kz=0 ® canbe writen as = Act + Bem © 4. Jahnke, F.Emae, and F. sch. Table of Higher Functions, éth ed. revised by F.Lésch, ‘MeGrowrHil, New York, 1960. 18M, Abramowitz and I. Stogun (Eds), Handbook of Mathematical Functions, Dover, Now York, 1964.146 Bessel Functions 371 where e2M* = cos Kz + j sin Kz (10) ‘Since the complex exponentials are linear combinations of cosine and sine functions, ‘we may also write the general solution of (8) as Z = Ate + BY sin Ke or other combinations. Similarly, itis convenient to define new Bessel functions which are linear combi- nations of the J,(7+) and N,(Tr) functions. By direct analogy with the definition (10) of the complex exponential, we write HOD = LATA) + INT, ay HPO = IATA) ~ jNATA, a2) “These are called Hankel functions of the first and second kinds, respectively. Since they both contain the function N,(T7), they are both singular at r = 0. Negative and positive orders are related by HOLT) = HOC) HT) = HPT For large values of the argument, these can be approximated by complex exponentials, with magnitude decreasing as square root of radius. For example, sapeay = [EZ crrmcves ‘This asymptotic form suggests that Hankel functions may be useful in wave propagation ‘problems asthe complex exponential i in plane-wave propagation. Its also sometimes ‘convenient to use Hankel functions as alternate independent solutions in static problems. ‘Complete solutions of (3) may be written in a variety of ways using combinations of ‘Bessel and Hankel functions. Bessel and Hankef Functions of Imaginary Arguments If 7 is imaginary, T = jr, and (3) becomes (* + eye =0 ey ‘The solution to (3) is valid here if Tis replaced by jr in the definitions of J,(Tr) and N,(T>. In this case N,(jm) is complex and so requires two numbers for each value of the argument, whereas j-"J,(jm) is always a purely real number. It is convenient to replace N,(j77) by a Hankel function. The quantity ;*~'H{( jr) is also purely real and 0 requires tabulation of only one value for each value of the argument. If» is not an integer, j°J (jr) is independent of j~"J,{j2r) and may be used as a second solution.372 chapter Two- and Three-Dimenslonsl Boundary Value Problems ‘Thus, for nonintegral v two possible complete solutions are. R = Agim) + Blin) aa and R = Agim) + BHM) (1s) ‘where powers of jare included in the constants. For v = 7, an integer, the two solutions in (14) are not independent but (15) is still a valid solution, It is common practice to denote these solutions as Taf) = FT 20) a6) Kyo) = Fr HPD) an where v = 7. ‘As is noted in Sec. 7.15 some ofthe formulas relating Bessel functions and Hankel functions must be changed for these modified Bessel functions. Special cases of these functions were seen as (ar) and Ke(77) in Sec. 7.13 forthe axially symmetric field. ‘The forms of /,(zr) and K,(a) for v = 0, 1 are shown in Fig. 7.14c. As is suggested by these curves, the asymptotic forms ofthe modified Bessel functions are related t0 growing and decaying real exponentials, as willbe seen in Eqs. 7.15(S) and 7.15(6). It is also clear from the figure that Kr) is singular atthe origin. Flo. 7.140 Modified Bessel functions.