11. Integrals of Bessel Functions YUDELLL. LUKE’ Contents Page . . . . . . . . . . 480 480 482 483 485 488 4aa 490 492 Table 11.2 Integrals of Bessd Functions . . . . . . . . . . . . . 494 Mathematical Properties . . . . . . . . . . . . . 11.1. Simple Istegrals of Bessel Functions . . . . 11.2. Repeated Integrals of J,,(z) and &(z) - . . . 11.3. Reduction Formulaa for Indefinite Integrals 11.4. Definite Integrals . . . . . . . . . . . . Numerical Methoda . . . . . . . . . . . . . . . 11.5. Use and Extension of the Tables . . . . . References . . . . . . . . . . . . . . . . . . . Table 11.1 Integrals of Bessel Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The author acknowledges the assistance of Geraldine Coombs, Betty Kahn, Marilyn Kemp, Betty Ruhlman, and Anna Lee Samuels for checking formulas and developing numerical examples, only a portion of which could be accommodated here. Midwest Research Institute. (Prepared under contract with the National Bureau of Standards.) 479 11. Integrals of Bessel Functions Mathematical Properties 11.1. Simple Integrals of Beesel Functions 11.1.8 F 0 ( t ) d t =;cz,(z) p,(t)dt +z1 fl{ -LO(z)Z1(z) +Ll(Z)Z"(d1 .Z,(z)=AI, (2)+Be'**K,(z),v=0'1 11.1.1 A and B are constants. E&) and L,(z) are Struve functions (see chapter 12). 11.1.9 (Wb+v+l)>O) y (Euler's constant)=.57721 56649 . . . 11.1.3 So'J,.(t)dl=S'J.(t)dt-2 0 % k=O Jzs+l(z) In this and all other integrals of 11.1, z is real and positive although all the results remain valid for extended portions of the complex plane unless stated to the contrary. 11.1.4 So'J2m+1(t)dt=1-J0(z)-2 5 J&) 11.1.10 Recurrence Relations k=1 l-f= s': f &(t)dt=i 11.1.11 0 Jo(t)dt+i Y,(t)dt 481 INTEGRALS OF BE8SEL FUNCMONB 11.1.14 2'e-z~Z0(t)dt-(2r)-* 2a$-* 11.1.18 k-0 d8s.O' ~ o ( tdt) = where the at are defined as in 11.1.12. 11.1.15 dezl"Ko(t)dt-(g) r t OD k 0 1 2 3 4 5 6 Polynomial Approximatione f 81x1.. " 11.1.19 1 7 .00100 89872 .OOO63 66169 .OOO39 92825 .Ooo27 55037 .OOO12 70039 .00002 68482 2 3 4 5 6 7 11.1.17 .00178 70944 .OOO67 40148 .OOO41 00676 .Ooo25 43955 .OOO11 07299 .00002 26238 For #(z), see 6.3. 11.1.20 8 9 5 - a9e-'JZ lo(t) dt=$ -0 dt(x/8)-t+c(x) 0 I 11.1.21 le(z)l52XlO-' k 0 1 2 3 4 5 6 dr ,39894 .03117 .00591 .00559 --.01148 .01774 -. 00739 23 34 91 56 58 40 95 Approximation 11.1.16 is from A. J. M. Hitchcock. Polynomial approximations to Bessel functions of order zero and one and to related functions, Math. Tables Aids Comp. 11, 86-88 (1957) (with permission). 2 (-)*et (x/7)-t+ t (z) le(~)l 52x10-7 (-)'aF' where the at are defined m in 11.1.12. P 6 k-0 k-0 11.1.16 71x5.. I 11.1.22 11.1.23 ek 1.25331 0.11190 .02576 .00933 .00417 .00163 .00033 414 289 646 994 454 271 934 INTEQRALS OF BESSEL FUNCTION13 482 Aeymptotic Expamiom 11.1.30 45x5 Ic(z)I 56X lo-' where k 0 1 2 3 4 5 6 11.1.25 11.1.31 1.25331 0.50913 .32191 .26214 .20601 .11103 .02724 41 39 84 46 26 96 00 5 5 2 5 Q) where 13 co=l, c1=8 11.1.27 k 0 1 2 3 4 5 6 7 8 9 10 where c k is defined as in ll.l.!Z?. Polynomial Approximations 11.1.29 55x5 fk 0.39893 . 13320 -. 04938 1.47800 -8.65560 28.12214 -48.05241 40.39473 -11.90943 -3.51950 2.19454 14 55 43 44 13 78 15 40 95 09 64 Repeated Intqgrals of J,(z) k bk 0 1 2 3 4 5 6 7 8 9 1. 0 0. 15999 2815 . 10161 9385 . 13081 1585 .20740 4022 .28330 0508 .27902 9488 . 17891 5710 .06622 8328 .01070 2234 1. 0 0.81998 6629 11.2.3 10702 2336 483 INTEGRALS OF BESSEL FUNCTIONS 11.2.5 Recurrence Relatiom 11.2.6 mim 11.2.10 (9?220, 11.2.11 w>o, wz>o, r=O) 11.3.7 g.,,(z)=- 2v+l Ki,(z) =- (WZ20, Wr>O) 11.2.12 U 11.2.13 KL+,(O)- ij r(r+3) -r (3) (r+ 1) 11.2.14 rKir+l(z)= -zKir(z) + (T- l)Kir-l(z) +ZKi,-a(z) 11.3. Reduction Formulas for Indefinite Integrals Let 11.3.1 i + 2 ” - ~ v -1)- r 11.3.11 g,,,.(z)=f’ e-P‘trZ,(t)dt where Z&) represents any of the Bessel functions of the first three kinds or the modified Beasel functions. The parameters a and b appearing in the reduction formulae are associated with the particular type of Bessel function as delineated in the following table. 11.3.12 484 INTEQRALS OF BESSEL FUNCTIONS 11.3.13 11.3.26 s,’ t-’l,+l(t)dt=z-’l,(z)-2,r 1 (v+l) 11.3.27 s,’t.~.-,(t)dt=--z.~,(2)+2’-1r(~) 11.3.28 (gV>o) 1- t -.K,+,(t)dt =2- .K.(Z) I n d a t e Integrab of Productcl of h l FUIME~~OM Let Wp(z) and 9,(z) denote any two cylinder functions of orders p and v respectively. 11.3.29 11.3.33 11.3.35 11.3.24 s,’t.Y,-,(t)dt=z.Y, (z)+-2’r(v) U = (L@v>O) 2 Ji(Z) k-n+l *seepage XI. INTEGRALS OF BESSEL FUNCTIONS 485 2. There must exist numbers kl and not zero) so that for all n 11.4.4 klAnV*i (La)-B%(AaU) 4 (both =O In connection with these formulae, see 11.3.29. If a=O, the above is,valid provided B=O. This case is covered by the following result. 11.4.5 So'tJ,(%t)J,(%t)dt==O (mZn, u>-1) =3[J;(41' (m=n, b=O, ) I - > Y (m=n, b ZO, 2 - 1) Y . . . are the positive zeros of ccJ,(z)+bz..Z(z)=O, where a and b are red constants. ( ~ 1 ,a¶, 11.4.6 m>o, 11.3.41 *>-I) Definite Inregrab Over a Finite Range 11.4. Definite Integrals Orthogonality Pmpertiea of1- Let %(z) be a cylinder function of order u. In particular, let f 11.4.7 Functions 11.4.8 J2=(2Z sin t)dt=2 R s,(z) Lr l' J0(22sin t) cos 2ntdt=~J:(z) I 11.4.1 %(z) -AJ,(z) +BP,(z) where A and B are real constants. Then 11.4.2 11.4.9 Y0(2zsin t) cos 2ntdt=i Jn(z)Y,(z) 11.4.10 11.4.11 provided the following two conditions hold: 1. A,,isarealzeroof 11.4.3 h,AV*,(Ab)-hl%(hb) =o f J,(z sin*t)J,(z cos4 t) c8c 2tdt 486 INTEGRALS OF BESSEL FUNCTIONS Infinite Integrals 11.4.23 I n y r a b of the Form 11.4.12 =0 (@'> 1) where T,,(w) is the Chebyshev polynomial of the first kind (see chapter 22). (gP<;t a(r+v)>o) 11.4.25 11.4.13 $-;t-1e-"'Jn pCI<;t W(r+v)>O) 11.4.14 11.4.17 (-i)"(l-~Z)+u~~~(@)(~*<l) =0(wZ>1) where UJw) is the Chebyshev polynomial of the second kind (see chapter 22). 11.4.26 11.4.15 SO0 --2in (t)dt sin bt K,(t)dt= arc sinh b (l+b')+ J,(t)dt=l (I 41<) (Bv>-1) wbere r(a,z) is the incomplete gamma function (see chapter 6). 11.4.18 Integra& of the Form 11.4.28 11.4.19 2 r L= c."WZ,[&)dt (B(CI+v)>O, ga'>O) where the notation M(a, b, z) stands for the confluent hypergeometric function (see chapter 13). 11.4.21 JmYo(t)dt=O 11.4.29 11.4622 (Wu>-l, @a*>O) 487 11.4.30 11.4.31 e-""Y,,(bt)dt=-- +: K, d e 2u (5) sec m] h v-l [I.(2) tan (19ul<r1 a w 11.4.31 11.4.32 (Wu>- 1,9a*>O) som J,(at) sin bt dt= :] sin [ p arc sin (a2-b2)+ Weber-Schafheitlin Type Integrals 11.4.33 b'r (u-p) (O,<b<a) 488 INTEGRALS OF BESSEL FUNCTIONS Hankel-Nicholson Tope Integrda I 11.4.44 (a>o, 11.4.47 Bz>O,-l<BY<29P+;) 11.4.45 b OD J,(at)dt =- t’(t*+ZZ) (a>O, BZ>O, !r [I,(az)-L,(az)] 22.+1 (a>o, B2>0, .%>-ij5, 9?v>-1) 11.449 11.4.46 Numerical Methods 11.5. Use and Extension of the Tables For moderate values of 2, use 11.1.2 and 11.1.711.1.10 as appropriate. For z suf6ciently large, use the asymptotic expansions or the polynomial approximations 11.1.11-11.1.18. Example 1. Compute l’M Jo(t)dt to 5D. Using 11.1.2 and interpolating in Tables 9.1 and 9.2, we have This value is readily checked using 2 ~ 3 . 1and h=-.05. Now IJo(z)l51 for all 2 and IJ,,(z)l <2-4, rill for all 2. In Table 11.1, we can always choose Ihl5.05. Thus if all terms of O(h4) and higher are neglected, then a bound for the absolute error is 2+h4/48<.3.10-6 for all z if Ihl <.05. Similarly, the absolute error for quadratic interpolation does not exceed h3(24+2)/24< .2- lo--? Example 3. Interpolation of Simpson’srule. We have 13.M Jo(t)dt=2[.32019 09 +.31783 69 +.04611 52 +. 00283 19 +.00009 72+ .OOOOO 211 = 1.37415 Example 2. Compute l.’’ Jo(t)dt to 5D by interpolation of Table 11.1 using Taylor’sformula. We have h‘ +3[Jz(z)-Jo(z)I+gg [3J1(4--Ja(dI+ ha and with IhII .05, it follows that - - lRl<.9 10-10 Then with z=3.0 and h=.05, Thus if 2=3.0 and h=.05 ~~J0(t)dt=1.387567+(.05)(-.260052) -(.00125)(.339059) + (.OOOOlO) (.746143) = 1.37415 ~‘06Jo(t)dt=1.38756 72520+-( 05) [-.26005 19549 6 +4(-.26841 13883)-.27653 495991 =1.37414 86481 489 INTEQRALS OF BESSEL FUNCTION8 which is correct to 10D. The above procedure gives high accuracy though it may be necessary to interpolate twice in Jo(z) to compute J o ( 3 1:+- and Jo(r+h). A similar technique based on the trapezoidal rule is less accurate, but at most only one interpolation of Jo(z)is required. Example 4. The recurrence formula gives fi=2 (f1 Similarly, Using 11.1.7, we have Jo(t)dt=3(-.260052) = 1.38757 +? [(.574306)(.339059) -(1.020110)(- .26OO52)1 = 1.69809 10 f3=1.20909 66,f4=.62451 73,f5=.25448 17 When r>>r, iary function Cornputel Jo(t)dt a n d l YdtW to 5D using the representation in terms of Struve functions and the tables in chapters 9 and 12. For 2=3, from Tablea 9.1 and 12.1 J1= ,339059 Jo=-. 260052 Yo= .376850 Y1= .324674 I&= .574306 &=l. 020110 4-f-1) it is convenient to use the auxil- Si(%)= (T- l)!~-'+'fr(;~) This satisfies the recurrence relation a2gr+i(x)=$gr91 (4= s' 0 (T- l)2gr-1(z) + (r- 1) (~-2)gr-2(2) r 1 3 JONdt, 92(4=91(3)-J1(4 93 (4= Vg2(4-91(4+Z J O O ) 1/39 Example 6. Compute g,(z) to 5D for z=10 and r=0(1)6. We have for z=lO, Jo=-.24593 58, J17.04347 27, g1=1.06701 13 Thus g2331.02353 86, g3=.98827 49 Similarly, and the forward recurrence formula gives Using 11.1.8 and Tables 9.8 and 12.1, one can For tables of 2-77(1:), see [11.16]. compute1 Io(t)dt a n d l Ko(t)dt. Jm Jo(;)dt,S," Yo(i)dt,1 V o ( t ) - 1 l d l , l t Ko(t)dt t For moderate values of z, use 11.1.19-11.1.23. For z sufficiently large, use the asymptotic expansions or the polynomial approximations 11.1.24-11.1.31. For moderate values of z and r, use 11.2.4. If r=l, see Example 1. For moderate values of 2, use tha recurrence formula 11.2.5. If 1: is large and z ~ rsee , tho discussion below. Example 5. Compute j r , &)=jr(z) to 5D for 2=2 and r=0(1)5 using 11.2.6. We have =r (r)+S fr-,(z) j-1(1:> = -Jl(4, f o b ) =JO(z1, f~(z> Jo (t)dt 0 and the termson this last line are tabulated. Thus for 2=2, f-l=--.57672 Repeated Integrals of &(x) For moderate values of z, use the recurrence formula 11.2.14 for all T. Example 7. Compute Ki,(z) to 5D for 2=2 and r=0(1)5. We have rKir+1(z)=-2Kir(z)+(r-1)Kir-1(z)+~Kir-2(z) m Repeated Integrals of J&) d r + l(z)=zfrCz)- (r-1)jr-i g4=.96867 36, g,=.94114 12, g,=.90474 64 48,fo=.22389 08,fl=1.42577 03 KLl(z)=Kl(z), Kio(z)=KO(%), Ki,(z)=S Ko(t)dt and the functions on this last line are tabulated Thus for ,2=2, K0=.11389 39, K1=.13986 59, Kil=.09712 06 and Ki2=-2Kil+2K1=.08549 06 Similarly, Kia=.07696 36, Ki4=.07043 17, Kis=.06525 22 If x/r is not large the formula can still be used provided that the starting values are sufliciently accurate to offset the growth of rounding error. For tables of Kit(%),see [11.11]. 490 INTEGRALS QF BESSEL FUNCTIONS Apart from roundaff error, the value of r needed to achieve a stated accuracy for given x and m can be determined a priori. Let Now fo(d = [l -zK1(41/z =SK0(t)dt,j1(2) 0 Then the latter following from 11.3.27 with v = l . In 11.3.5, put a = l , b=-1, p=O and v=O. Let p=m. Then Qr-2k= Qr jn(4 =[(m -1>Ynl-2(4 -2KI (4 -4m-1>~0(41/2 (m>l> Using tabular values of joand f l , one can compute in succession j2, js,. . . provided that m/x is not large. Example 8. Computej,,,(z)to 5 D for 2 = 5 and m=0(1)6. We have, retaining two additional decimals &= -00369 11 Thus jo=l. 56738 74 K1=. 00404 46 fi=. 19595 54 (~-1)~(r-3)~.. . (r-2k+1)2 I [ ~ K(5) I+dr- 1)Ko(z)I/(r- 1) since for 2 fixed,jr(Z) is positive and decreases as r increases. Example 9. Computej,,,(z)to 5D for 2=3 and m=0(2)10. We have &=.03473 95 K1=.04015 64 If r= 16, ~6<.86*1O-’ q0<1.4-1O-~ Taking g16=0,we compute the following values of g14) g12, . . ., go by recurrence. Also recorded are the required values off,,, to 5D. j2=.05791 27,j,=.01458 93,j6=.00685 36 Similarly starting withjl,we can computej3andjs. If m>z, employ the recurrence formula in backward form and write m 14 12 10 8 6 4 2 0 fm-2(z>=[22fm(2)+~~~(~>+~(m--1)Ko(~)l/(m--1)2 In the latter expression, replace j,,, by gn. Fix 2. Take r>m and assume gr=O. Compute gr-2, gr-4, etc. Then lim gr-w(z)=j,,,(z), m=r-2k fn 9., .00855 .01061 .01325 .01751 .02548 .04447 . 11936 1. 53994 42 09 05 39 09 31 90 71 .01325 .01751 .02548 .04447 . 11937 1.53995 For tables of j,,,(z),see [11.21]. r-m References Tests [11.1] H. Bateman and R. C. Archibald, A guide to tables of Bessel functions, Math. Tables Aids a m p . 1, 247-252 (1943). See also Supplements I, 11, IV, same journal, 1,403-404 (1943); 2,59 (1946); 2, 190 (1946), respectively. (11.21 A. Erd6lyi et al., Higher transcendental functions, vol. 2, ch. 7 (McGraw-Hill Book Co.,Inc., New York, N.Y., 1953). [11.3] A. ErdQyi et al., Tables of integral transforms, vola. 1, 2 (McGraw-Hill Book Co., Inc., New York, N.Y., 1954). [11.4] W. Grirbner and N. Hofreiter, Integraltafel, I1 Teil (Springer-Verlag, Wien and Innsbruck, Austria, 1949-1950). [11.5] L. V. King, On the convection of heat from small cylinders in a stream of fluid, Trans. Roy. Soo. London 214A, 373-432 (1914). [11.6] Y. L. Luke, Some notes on integrals involving Bessel functions, J. Math. Phys. 29, 27-30 (1960). [11.7] Y. L. Luke, An associated Bessel function, J. Math. Phys. 31, 131-138 (1952). [11.8] F. Oberhettinger, On some expansions for Bessel integral functions, J. Research NBS 59, 197-201 (1957) RP 2786. [11.9] G. Petiau, La thQrie des fonctions de Bessel (Centre National de la Recherche Scientifique, Paris, France, 1955). [11.10] G. N. Watson, A treatise on the theory of Bessel functions, 2d ed. (Cambridge Univ. Press, Cambridge, England, 1958). Tables [11.11] W. G. Bickley ana J. Nayler, A short table of the functions Kin@), from n = l to n=16. Philos. Mag.7,20,343-347 (1935). Kil(z)=LmKo(t)&, Ki,(z) =Jm Ki,,-I(t)&, (.1)2, 3, QD. n=1(1)16, z=0(.06).2 491 INTEGRALS OF BESSEL FUNCTIONS [11.12] V. R. Bursian and V. Pock, Table of the functions Akad. Nauk, Leningrad, Inst. Fiz. Mat., Trudy (Travaux) 2, 6-10 (1931). JmKo(t)dl, L- z=O(.1)12, 7D; e= Zo(t)dt, z=O(.1)6, Ko(t)dt, z=O(.l)l6, 7D; 7D; e-= Zo(t)dt, z= 0 (.1)16, 7D. [11.13] E. A. Chistova, Tablitay funktsii Besselya ot deistvitel’ nogo argumenta i integralov ot nikh (Izdat. Akad. Nauk SSSR., MOSCOW, U.S.S.R., n=O, 1; z=0(.001)15(.01)100, 7 0 . Also tabulated are auxiliary expressions to facilitate interpolation near the origin. [11.14] A. J. M. Hitchcock, Polynomial approximations to Bessel functions of order zero and one and to related functions, Math. Tables Aids a m p . 11, 86-88 (1957). Polynomial approximations for i‘Jo(t)dtand KKo(t)dt. [ll.l5] C. W. Horton, A short table’of Struve functions and of some integrals involving Bessel and Struve functions, J. Math. Phys. 29, 66-68 (1960). C,(z) 4D; D,(z) =$ tJ,(t)dt, n= 1(1)4,2=O(.I)10, =J’t”E.(t)dt,n=0(1)4, ~=0(.1)10, 4D, where E,@) is Struve’s function; see chapter 12. [ll.l6] J. C. Jaeger, Repeated integrals of Bessel functions and the theory of transienta in filter circuita, J. Math. Phys. 27, 210-219 (1948). f~(z)= l’ Jo(t)dl, f,(z)=Jzf,l(t)dt, 2-%(z), r=1(1)7, [11.18] H. L. Knudsen, Bidrag til teorien for antennesystemer med he1 eller delvis rotations-symmetri. I (Kommission Has Teknisk Forlag, Copenhagen, Denmark, 1953). J”J,(t)dt, 0(.01)10, SD. Also f n=0(1)8, z= J.(t)ecodl, a=t,a=z-t. (11.191 Y. 5.Luke and D. Ufford, Tables of the function &(z) =6’Ko(t)dt. Math. Tables Aids Comp. UMT 129. Z(z)= -[r+h(~/2)1Ai(~) + Aa(z), Ai@), A&). z=0(.01).5(.05)1, 8D. [11.20] C. Mack and M. Castle, Tables of s,”Zo(z)& and JamKo(z)&,Roy. Soc. Unpublished Math. Table File No. 6. a=0(.02)2(.1)4, QD. [11.21] G. M. Muller, Table of the function Kj,(z) =z-ns,’unKo(u)du, Office of Technical Services, U.S. Department of Commerce, Washington, D.C. (1954). n=0(1)31, ~=0(.01)2(.02)5, 89. (11.221 National Bureau of Standards, Tables of functions and zeros of functions, Applied Math. Series 37 (U.S. Government Printing Office, Washington, D.C., 1954). (1) pp. 21-31: s,’Jo(t)dt, z=O(.01)10, 10D. (2) pp. 33-39: ~=0(.1)10(1)22, 10D; Y~(t)dt, Jo(t)dt/t, F(z)=Jm Jo(t)dl/t +In (z/2), z=0(.1)3, 10D; F(”(z)/n!, z= lO(l)22, n=O(1)13, 12D. [11.23] National Physical Laboratory, Integrals of Bessel functions, Roy. Soc. Unpublished Math. Table [11.24] M. Rothman, Table of PO(Z)& for 0(.1)20(1)26, Quart. J. Mech. Appl. Math. 2, 212-21.7 (1949). Z=0(1)24, 8D. Also +,(z) =~mJ~[2(~t)“]Jn(t)~p 85-98. Jo(t)dt for large z, [11.26] P. W. Schmidt, Tables of an@),*i(z), n=1(1)7, ~=0(1)24, 4D. [11.17] L. N. Karmaeina and E. A. Chistova, Tablitay J. Math. Phys. 34, 169-172 (1955). 2=10(.2)40, funktaii Besselya ot mnimogo argumenta i 6D. integralov ot nikh (Izdat. Akad. Nauk SSSR., (11.261 G. N. Watson, A treatise on the theory of Bessel MOSCOW,U.S.S.R., 1958). e-+Zo(z), e-+Zl(z), functions, 2d ed. (Cambridge Univ. Press, Cambridge, England, 1958). Table VIII, p. 752: e=Ko(z),ezK1(z), e=, e-=f Zo(t)dl, e = cKo(t)dt, i p ~ ( t ) d t , ; f Y ~ ( t ) d t , 2=0(.02)1, 7D, with z=0(.001)5(.005)15(.01)100, 7D except for e’ the first 16 maxima and minima of the integrals which is 78. Also tabulated are auxiliary expresto 7D. sions to facilitate interpolation near the origin. 6’