A Some Special Functions and Their Properties
The main purpose of this appendix is to introduce several special functions and to state their basic properties that are most frequently used in the theory and applica- tions of ordinary and partial differential equations. The subject is, of course, too vast to be treated adequately in so short a space, so that only the more important results will be stated. For a fuller discussion of these topics and of further properties of these functions the reader is referred to the standard treatises on the subject.
A-1 Gamma, Beta, and Error Functions
The gamma function (also called the factorial function) is defined by a definite inte- gral in which a variable appears as a parameter ∞ Γ (x)= e−ttx−1 dt, x > 0. (A-1.1) 0 The integral (A-1.1) is uniformly convergent for all x in [a, b] where 0 0. Integrating (A-1.1) by parts, we obtain the fundamental property of Γ (x) ∞ − −t x−1 ∞ − −t x−2 Γ (x)= e t 0 +(x 1) e t dt 0 =(x − 1)Γ (x − 1) for x − 1 > 0. Then we replace x by x +1to obtain the fundamental result Γ (x +1)=xΓ(x). (A-1.2) In particular, when x = n is a positive integer, we make repeated use of (A-1.2) to obtain Γ (n +1)=nΓ (n)=n(n − 1)Γ (n − 1) = ··· = n(n − 1)(n − 2) ···3 · 2 · 1Γ (1) = n!, (A-1.3) where Γ (1) = 1.
L. Debnath, Nonlinear Partial Differential Equations for Scientists and Engineers, DOI 10.1007/978-0-8176-8265-1, c Springer Science+Business Media, LLC 2012 690 A Some Special Functions and Their Properties
We put t = u2 in (A-1.1) to obtain ∞ Γ (x)=2 exp −u2 u2x−1 du, x > 0. (A-1.4) 0 1 Letting x = 2 , we find √ 1 ∞ π √ Γ =2 exp −u2 du =2 = π. (A-1.5) 2 0 2 Using (A-1.2), we deduce √ 3 1 1 π Γ = Γ = . (A-1.6) 2 2 2 2
5 7 2n+1 Similarly, we can obtain the values of Γ ( 2 ),Γ( 2 ),...,Γ( 2 ). The gamma function can also be defined for negative values of x by the rewritten form of (A-1.2)as Γ (x +1) Γ (x)= ,x=0 , −1, −2,.... (A-1.7) x For example, 1 Γ ( 1 ) 1 √ Γ − = 2 = −2 Γ = −2 π, (A-1.8) 2 − 1 2 2 3 Γ (− 1 ) 4√ Γ − = 2 = π. (A-1.9) − 3 2 2 3 We differentiate (A-1.1) with respect to x to obtain d ∞ d e−t Γ (x)=Γ (x)= tx dt dx dx t 0 ∞ d e−t ∞ = exp(x log t) dt = tx−1(log t)e−t dt. (A-1.10) 0 dx t 0 At x =1, this gives ∞ Γ (1) = e−t log tdt= −γ, (A-1.11) 0 where γ is called the Euler constant and has the value 0.5772. The graph of the gamma function is shown in Figure A.1. The volume, Vn, and the surface area, Sn, of a sphere of radius r in n-dimensional space Rn are given by { 1 }n n { 1 }n n−1 Γ ( 2 ) r 2 Γ ( 2 ) r Vn = n ,Sn = n . Γ ( 2 +1) Γ ( 2 )
dVn Thus, dr = Sn. A-1 Gamma, Beta, and Error Functions 691
Fig. A.1 The gamma function.
2 4 3 In particular, when n =2, 3,..., we get V2 = πr , S2 =2πr; V3 = 3 πr , 2 S3 =4πr ;etc. Using (A-1.2) and (A-1.5), we obtain the following results: πmr2m 2πmr2m−1 V = ,S= , 2m m! 2m (m − 1)! 2(2π)mr2m+1 22m+1m!πmr2m V = ,S = . 2m+1 1.3.5 ···(2m +1) 2m+1 (2m)!
Legendre Duplication Formula
Several useful properties of the gamma function are recorded below for reference without proof. We begin with 1 √ 22x−1Γ (x)Γ x + = πΓ(2x). (A-1.12) 2 In particular, when x = n (n =0, 1, 2,...), √ 1 π (2n)! Γ n + = . (A-1.13) 2 22n n! The following properties also hold for Γ (x):
Γ (x)Γ (1 − x)=π cosec πx, x is a noninteger, (A-1.14) ∞ Γ (x)=px exp(−pt) tx−1 dt, (A-1.15) 0 ∞ Γ (x)= exp xt − et dt. (A-1.16) −∞ 692 A Some Special Functions and Their Properties √ x+ 1 Γ (x +1)∼ 2π exp(−x)x 2 for large x, (A-1.17) √ n+ 1 n! ∼ 2π exp(−n)x 2 for large n. (A-1.18)
The latter formulas are known as Stirling approximation of Γ (x +1)for large x and of n! for large n. The incomplete gamma function, γ(x, a), is defined by the integral x γ(a, x)= e−tta−1 dt, a > 0. (A-1.19) 0 The complementary incomplete gamma function, Γ (a, x), is defined by the integral ∞ Γ (a, x)= e−t ta−1 dt, a > 0. (A-1.20) x Thus, it follows that γ(a, x)+Γ (a, x)=Γ (a). (A-1.21) The beta function, denoted by B(x, y), is defined by the integral t B(x, y)= tx−1(1 − t)y−1 dt, x > 0,y>0. (A-1.22) 0 The beta function B(x, y) is symmetric with respect to its arguments x and y, that is, B(x, y)=B(y, x). (A-1.23) This follows from (A-1.22) by the change of variable 1 − t = u, that is, 1 B(x, y)= uy−1(1 − u)x−1 du = B(y, x). 0 If we make the change of variable t = u/(1 + u) in (A-1.22), we obtain another integral representation of the beta function ∞ ∞ B(x, y)= ux−1(1 + u)−(x+y) du = uy−1(1 + u)−(x+y) du. (A-1.24) 0 0 Putting t = cos2 θ in (A-1.22), we derive π/2 B(x, y)=2 cos2x−1 θ sin2y−1 θdθ. (A-1.25) 0 Several important results are recorded below for ready reference without proof: 1 1 B(1, 1) = 1,B, = π, (A-1.26) 2 2 x − 1 B(x, y)= B(x − 1,y), (A-1.27) x + y − 1 A-1 Gamma, Beta, and Error Functions 693
Fig. A.2 The error function and the complementary error function.
Γ (x)Γ (y) B(x, y)= , (A-1.28) Γ (x + y) 1+x 1 − x πx B , = π sec , 0 The error function, erf(x), is defined by the integral 2 x erf(x)=√ exp −t2 dt, −∞ erf(−x)=− erf(x), (A-1.31) d 2 erf(x) = √ exp −x2 , (A-1.32) dx π erf(0) = 0, erf(∞)=1. (A-1.33) The complementary error function, erfc(x), is defined by the integral 2 ∞ erfc(x)=√ exp −t2 dt. (A-1.34) π x Clearly, it follows that erfc(x)=1− erf(x), (A-1.35) erfc(0) = 1, erfc(∞)=0. (A-1.36) The graphs of erf(x) and erfc(x) are shown in Figure A.2. 1 erfc(x) ∼ √ exp −x2 for large x. (A-1.37) x π Closely associated with the error function are the Fresnel integrals, which are defined by 694 A Some Special Functions and Their Properties Fig. A.3 The Fresnel integrals C(x) and S(x).