ebook img

Gamma Function PDF

29 Pages·2012·0.41 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Gamma Function

Chapter 10 The Gamma Function (Factorial Function) The gamma function appears in physical problems of all kinds, such as the normalizationofCoulombwavefunctionsandthecomputationofprobabilities instatisticalmechanics.Itsimportancestemsfromitsusefulnessindeveloping other functions that have direct physical application. The gamma function, therefore, is included here. A discussion of the numerical evaluation of the gammafunctionappearsinSection10.3.Closelyrelatedfunctions,suchasthe errorintegral,arepresentedinSection10.4. 10.1 DefinitionsandSimpleProperties At least three different, convenient definitions of the gamma function are in commonuse.Ourfirsttaskistostatethesedefinitions,todevelopsomesimple, directconsequences,andtoshowtheequivalenceofthethreeforms. InfiniteLimit(Euler) Thefirstdefinition,duetoEuler,is 1·2·3···n (cid:2)(z)≡ lim nz, z=(cid:7) 0,−1,−2,−3,.... (10.1) n→∞ z(z+1)(z+2)···(z+n) Thisdefinitionof(cid:2)(z)isusefulindevelopingtheWeierstrassinfinite-product form of (cid:2)(z) [Eq. (10.17)] and in obtaining the derivative of ln(cid:2)(z) (Sec- tion10.2).Hereandelsewhereinthischapter,zmaybeeitherrealorcomplex. 523 524 Chapter10 TheGammaFunction(FactorialFunction) Replacingzwithz+1,wehave 1·2·3···n (cid:2)(z+1)= lim nz+1 n→∞(z+1)(z+2)(z+3)···(z+n+1) nz 1·2·3···n = lim · nz n→∞ z+n+1 z(z+1)(z+2)···(z+n) = z(cid:2)(z). (10.2) Thisisthebasicfunctionalrelationforthegammafunction.Itshouldbenoted thatitisadifferenceequation.Thegammafunctionisoneofageneralclass of functions that do not satisfy any differential equation with rational coef- ficients. Specifically, the gamma function is one of the very few functions of mathematical physics that does not satisfy any of the ordinary differential equations(ODEs)commontophysics.Infact,itdoesnotsatisfyanyusefulor practicaldifferentialequation. Also,fromthedefinition 1·2·3···n (cid:2)(1)= lim n=1. (10.3) n→∞1·2·3···n(n+1) Now,applicationofEq.(10.2)gives (cid:2)(2)=1, (cid:2)(3)=2(cid:2)(2)=2,... (10.4) (cid:2)(n)=1·2·3···(n−1)=(n−1)! We see that the gamma function interpolates the factorials by a continuous functionthatreturnsthefactorialsatintegerarguments. DefiniteIntegral(Euler) Aseconddefinition,alsofrequentlycalledtheEulerintegral,is (cid:2) ∞ (cid:2)(z)≡ e−ttz−1dt, (cid:10)(z)>0. (10.5) 0 Therestrictiononzisnecessarytoavoiddivergenceoftheintegralatt = 0. Whenthegammafunctiondoesappearinphysicalproblems,itisofteninthis formorsomevariation,suchas (cid:2) ∞ (cid:2)(z)=2 e−t2t2z−1dt, (cid:10)(z)>0 (10.6) 0 or (cid:2) (cid:3) (cid:4) (cid:5)(cid:6) 1 1 z−1 (cid:2)(z)= ln dt, (cid:10)(z)>0. t 0 10.1 DefinitionsandSimpleProperties 525 EXAMPLE10.1.1 The Euler Integral Interpolates the Factorials The Euler integral for positive integer argument, z = n+1 with n> 0, yields the factorials. Using integrationbypartsrepeatedlywefind (cid:2) (cid:7) (cid:2) (cid:2) ∞ (cid:7)∞ ∞ ∞ e−ttndt =−tne−t(cid:7)(cid:7) +n e−ttn−1dt =n e−ttn−1dt 0 (cid:3) 0 (cid:7) 0 (cid:2) 0 (cid:6) (cid:7)∞ ∞ =n −tn−1e−t(cid:7)(cid:7) +(n−1) e−ttn−2dt (cid:2) 0 0 ∞ =n(n−1) e−ttn−2dt =···=n! 0 (cid:8) because ∞e−tdt =1.Thus,theEulerintegralalsointerpolatesthefactorials. 0 (cid:2) When z = 1, Eq. (10.6) contains the Gauss error integral, and we have the 2 interestingresult (cid:9) (cid:10) √ (cid:2) 1 = π. (10.7) 2 Thevalue(cid:2)(1)canbederiveddirectlyfromthesquareofEq.(10.6)forz= 1 2 2 byintroducingplanepolarcoordinates(x2+y2 =ρ2,dxdy=ρdρdϕ)inthe productofintegrals (cid:4) (cid:5) (cid:2) (cid:2) 1 2 ∞ ∞ (cid:2) =4 e−x2−y2dxdy 2 0 0 (cid:2) (cid:2) (cid:7) π/2 ∞ (cid:7)∞ =4 e−ρ2ρdρdϕ =−πe−ρ2(cid:7)(cid:7) =π. ϕ=0 ρ=0 0 GeneralizationsofEq.(10.7),theGaussianintegrals,are (cid:2) ∞ s! x2s+1exp(−ax2)dx= , (10.8) 2as+1 0 (cid:2) (cid:11) ∞ (s− 1)! (2s−1)!! π x2sexp(−ax2)dx= 2 = , (10.9) 2as+1/2 2s+1as a 0 whichareofmajorimportanceinstatisticalmechanics.Aproofislefttothe readerinExercise10.1.11.ThedoublefactorialnotationisexplainedinExer- cise5.2.12andEq.(10.33b). To show the equivalence of the two definitions, Eqs. (10.1) and (10.5), considerthefunctionoftwovariables (cid:2) (cid:4) (cid:5) n t n F(z,n)= 1− tz−1dt, (cid:10)(z)>0, (10.10) n 0 withnapositiveinteger.Since (cid:4) (cid:5) t n lim 1− ≡e−t, (10.11) n→∞ n 526 Chapter10 TheGammaFunction(FactorialFunction) fromthedefinitionoftheexponential (cid:2) ∞ lim F(z,n)= F(z,∞)= e−ttz−1dt ≡(cid:2)(z) (10.12) n→∞ 0 byEq.(10.5). ReturningtoF(z,n),weevaluateitinsuccessiveintegrationsbyparts.For convenienceletu=t/n.Then (cid:2) 1 F(z,n)=nz (1−u)nuz−1du. (10.13) 0 Integratingbyparts,weobtainfor(cid:10)(z)>0, (cid:7) (cid:2) F(z,n) =(1−u)nuz(cid:7)(cid:7)(cid:7)1+ n 1(1−u)n−1uzdu. (10.14) nz z z 0 0 Repeating this, with the integrated part vanishing at both end points each time,wefinallyget (cid:2) n(n−1)···1 1 F(z,n)=nz uz+n−1du z(z+1)···(z+n−1) 0 1·2·3···n = nz. (10.15) z(z+1)(z+2)···(z+n) ThisisidenticaltotheexpressionontherightsideofEq.(10.1).Hence, lim F(z,n)= F(z,∞)≡(cid:2)(z) (10.16) n→∞ byEq.(10.1),completingtheproof. Usingthefunctionalequation(10.2)wecanextend(cid:2)(z)frompositiveto negativearguments.Forexample,startingwithEq.(10.2)wedefine (cid:4) (cid:5) (cid:4) (cid:5) 1 1 1 √ (cid:2) − =− (cid:2) =−2 π, 2 0.5 2 andEq.(10.2)forz→0impliesthat(cid:2)(z→0)→∞.Moreover,z(cid:2)(z)→1for z → 0 [Eq. (10.2)] shows that (cid:2)(z)has a simple pole at the origin. Similarly, wefindsimplepolesof(cid:2)(z)atallnegativeintegers. InfiniteProduct(Weierstrass) Thethirddefinition(Weierstrass’sform)is (cid:4) (cid:5) (cid:12)∞ 1 z ≡zeγz 1+ e−z/n, (10.17) (cid:2)(z) n n=1 whereγ istheEuler–Mascheroniconstant[Eq.(5.27)] (cid:13) (cid:15) (cid:14)n 1 γ = lim −lnn =0.5772156···. (10.18) n→∞ m m=1 10.1 DefinitionsandSimpleProperties 527 This infinite product form may be used to develop the reflection identity, Eq.(10.24a),andappliedintheexercises,suchasExercise10.1.17.Thisform canbederivedfromtheoriginaldefinition[Eq.(10.1)]byrewritingitas (cid:4) (cid:5) 1·2·3···n 1 (cid:12)n z −1 (cid:2)(z)= lim nz= lim 1+ nz. (10.19) n→∞ z(z+1)···(z+n) n→∞ z m m=1 InvertingEq.(10.19)andusing n−z=e−zlnn, (10.20) weobtain 1 (cid:12)n (cid:16) z(cid:17) =z lim e(−lnn)z 1+ . (10.21) (cid:2)(z) n→∞ m m=1 Multiplyinganddividingby (cid:3)(cid:4) (cid:5) (cid:6) 1 1 1 (cid:12)n exp 1+ + +···+ z = ez/m, (10.22) 2 3 n m=1 weget (cid:18) (cid:3)(cid:4) (cid:5) (cid:6)(cid:19) 1 1 1 1 = z lim exp 1+ + +···+ −lnn z (cid:2)(z) n→∞ 2 3 n (cid:20) (cid:21) (cid:12)n (cid:16) z(cid:17) × lim 1+ e−z/m . (10.23) n→∞ m m=1 As shown in Section 5.2, the infinite series in the exponent converges and definesγ,theEuler–Mascheroniconstant.Hence,Eq.(10.17)follows. The Weierstrass infinite product definition of (cid:2)(z) leads directly to an importantidentity, π (cid:2)(z)(cid:2)(1−z)= , (10.24a) sinzπ usingtheinfiniteproductformulasfor(cid:2)(z),(cid:2)(1−z),andsinz[Eq.(7.60)]. Alternatively,wecanstartfromtheproductofEulerintegrals (cid:2) (cid:2) ∞ ∞ (cid:2)(z+1)(cid:2)(1−z)= sze−sds t−ze−tdt (cid:2)0 0 (cid:2) ∞ dv ∞ πz = vz e−uudu= , (v+1)2 sinπz 0 0 transforming from the variables s,t to u = s+t,v = s/t, as suggested by combiningtheexponentialsandthepowersintheintegrands.TheJacobianis (cid:7) (cid:7) J =−(cid:7)(cid:7)(cid:7)11 −1 s (cid:7)(cid:7)(cid:7)= st+2 t = (v+u1)2, t t2 (cid:8) where (v +1)t = u. The integral ∞e−uudu = 1, whereas that over v may 0 be derived by contour integration, giving πz . Similarly, one can establish sinπz 528 Chapter10 TheGammaFunction(FactorialFunction) Legendre’sduplicationformula (cid:9) (cid:10) √ (cid:2)(1+z)(cid:2) z+ 1 =2−2z π(cid:2)(2z+1). 2 Settingz= 1 inEq.(10.24a),weobtain 2 (cid:9) (cid:10) √ (cid:2) 1 = π, (10.24b) 2 (takingthepositivesquareroot)inagreementwithEqs.(10.7)and(10.9). The Weierstrass definition shows immediately that (cid:2)(z)has simple poles atz= 0,−1,−2,−3,...,andthat[(cid:2)(z)]−1 hasnopolesinthefinitecomplex plane,whichmeansthat(cid:2)(z)hasnozeros.Thisbehaviormayalsobeseenin Eq.(10.24a),inwhichwenotethatπ/(sinπz)isneverequaltozero. Actually, the infinite product definition of (cid:2)(z) may be derived from the Weierstrass factorization theorem with the specification that [(cid:2)(z)]−1 have simplezerosatz=0,−1,−2,−3,....TheEuler–Mascheroniconstantisfixed byrequiring(cid:2)(1)= 1.Seealsotheproductexpansionsofentirefunctionsin Chapter7. In mathematical probability theory the gamma distribution (probability density)isgivenby   1 xα−1e−x/β, x>0 f(x)= βα(cid:2)(α) (10.25a)  0, x≤0. Theconstant[βα(cid:2)(α)]−1 isincludedsothatthetotal(integrated)probability will be unity. For x → E, kinetic energy, α → 3 and β → kT, Eq. (10.25a) 2 yieldstheclassicalMaxwell–Boltzmannstatistics. BiographicalData Weierstrass,KarlTheodorWilhelm. Weierstrass,aGermanmathemati- cian,wasbornin1815inOstenfelde,Germany,anddiedin1897inBerlin. Heobtainedadegreeinmathematicsin1841andstudiedAbel’sandJacobi’s workonellipticalfunctionsandextendedtheirworkonanalyticfunctions whilelivingasaschoolteacher.Afterbeingrecognizedasthefatherofmod- ernanalysis,hebecameaprofessorattheUniversityofBerlinandamember oftheAcademyofSciencesin1856. FactorialNotation So far, this discussion has been presented in terms of the classical notation. As pointed out by Jeffreys and others, the −1 of the z−1 exponent in our seconddefinition[Eq.(10.5)]isacontinualnuisance.Accordingly,Eq.(10.5) isrewrittenas (cid:2) ∞ e−ttzdt ≡z!, (cid:10)(z)>1, (10.25b) 0 10.1 DefinitionsandSimpleProperties 529 todefine(cid:26)afactorialfunctionz!.Occasionally,wemayevenencounterGauss’s notation, (z),forthefactorialfunction (cid:12) (z)=z!. (10.26) The(cid:2)notationisduetoLegendre.ThefactorialfunctionofEq.(10.25a)is, ofcourse,relatedtothegammafunctionby (cid:2)(z)=(z−1)!, or (cid:2)(z+1)=z!. (10.27) Ifz=n,apositiveinteger[Eq.(10.4)]showsthat z!=n!=1·2·3···n, (10.28) thefamiliarfactorial.However,itshouldbenotedthatsincez!isnowdefined byEq.(10.25b)[orequivalentlybyEq.(10.27)]the factorial function is no longer limited to positive integral values of the argument (Fig. 10.1). Thedifferencerelation[Eq.(10.2)]becomes z! (z−1)!= . (10.29) z Thisshowsimmediatelythat 0!=1 (10.30) Figure10.1 TheFactorial x! Function---Extension toNegative Arguments 5 4 3 2 1 x –5 –3 –1 1 2 3 4 5 –1 –2 –3 –4 –5 530 Chapter10 TheGammaFunction(FactorialFunction) Figure10.2 TheFactorial 6 Functionandthe 5 FirstTwo Derivativesofln(x!) 4 x! 3 2 d2 ln (x!) ddxln (x!) dx2 1 d2 ln (x!) x! dx2 0 d ln (x!) –1.0 dx –1.0 0 1.0 2.0 3.0 4.0 x and n!=±∞ forn,anegativeinteger. (10.31) Intermsofthefactorial,Eq.(10.24a)becomes πz z!(−z)!= (10.32) sinπz because 1 (cid:2)(z)(cid:2)(1−z)=(z−1)!(1−z−1)!=(z−1)!(−z)!= z!(−z)!. z By restricting ourselves to the real values of the argument, we find that x! definesthecurveshowninFig.10.2.TheminimumofthecurveinFig.10.1is x!=(0.46163···)!=0.88560···. (10.33a) DoubleFactorialNotation In many problems of mathematical physics, particularly in connection with Legendrepolynomials(Chapter11),weencounterproductsoftheoddpositive integersandproductsoftheevenpositiveintegers.Forconvenience,theseare givenspeciallabelsasdoublefactorials: 1·3·5···(2n+1)=(2n+1)!! (10.33b) 2·4·6···(2n)=(2n)!!. Clearly,thesearerelatedtotheregularfactorialfunctionsby (2n+1)! (2n)!!=2nn! and (2n+1)!!= . (10.33c) 2nn! 10.1 DefinitionsandSimpleProperties 531 IntegralRepresentation An integral representation that is useful in developing asymptotic series for theBesselfunctionsis (cid:2) e−zzνdz=(e2πiν −1)ν!, (10.34) C whereCisthecontourshowninFig.10.3.Thiscontourintegralrepresentation is only useful when ν is not an integer, z = 0 then being a branch point. Equation (10.34) may be verified for ν > −1 by deforming the contour as shown in Fig. 10.4. The integral from ∞ to (cid:14) > 0 in Fig. 10.4 yields −(ν!), placingthephaseofzat0.Theintegralfrom(cid:14) to∞(inthefourthquadrant) then yields e2πiνν!, the phase of z having increased to 2π. Since the circle of radius (cid:14) around the origin contributes nothing as (cid:14) → 0, when ν > −1, Eq.(10.34)follows. Itisoftenconvenienttoputthisresultintoamoresymmetricalform (cid:2) e−z(−z)νdz=2isin(νπ)ν!, (10.35) C multiplyingbothsidesofEq.(10.34)by(−1)ν =e−πiν. ThisanalysisestablishesEqs.(10.34)and(10.35)forν >−1.Itisrelatively simpletoextendtherangetoincludeallnonintegralν.First,wenotethatthe integralexistsforν < −1, aslongaswestayawayfromtheorigin.Second, integratingbypartswefindthatEq.(10.35)yieldsthefamiliardifferencere- lation [Eq. (10.29)]. If we take the difference relation to define the factorial Figure10.3 y FactorialFunction Contour C x +∞ Figure10.4 y TheContourof Fig.10.3Deformed Cut line e x +∞ 0 532 Chapter10 TheGammaFunction(FactorialFunction) functionofν <−1,thenEqs.(10.34)and(10.35)areverifiedforallν (except negativeintegers). EXERCISES 10.1.1 Derivetherecurrencerelations (cid:2)(z+1)=z(cid:2)(z) fromtheEulerintegral[Eq.(10.5)] (cid:2) ∞ (cid:2)(z)= e−ttz−1dt. 0 10.1.2 In a power series solution for the Legendre functions of the second kind,weencountertheexpression (n+1)(n+2)(n+3)···(n+2s−1)(n+2s) , 2·4·6·8···(2s−2)(2s)·(2n+3)(2n+5)(2n+7)···(2n+2s+1) in which s is a positive integer. Rewrite this expression in terms of factorials. 10.1.3 Showthat (s−n)! (−1)n−s(2n−2s)! = , (2s−2n)! (n−s)! wheresandnareintegerswiths<n.Thisresultcanbeusedtoavoid negativefactorialssuchasintheseriesrepresentationsofthespher- ical Neumann functions and the Legendre functions of the second kind. 10.1.4 Showthat(cid:2)(z)maybewritten (cid:2) ∞ (cid:2)(z)=2 e−t2t2z−1dz, (cid:10)(z)>0, 0 (cid:2) (cid:3) (cid:4) (cid:5)(cid:6) 1 1 z−1 (cid:2)(z)= ln dt, (cid:10)(z)>0. t 0 10.1.5 In a Maxwellian distribution the fraction of particles with speed betweenvandv+dvis (cid:4) (cid:5) (cid:4) (cid:5) dN m 3/2 mv2 =4π exp − v2dv, N 2πkT 2kT where N isthetotalnumberofparti(cid:8)cles.Theaverageorexpectation valueofvnisdefinedas(cid:15)vn(cid:16)= N−1 vndN.Showthat (cid:4) (cid:5) (cid:4) (cid:5)(cid:27) 2kT n/2 n+1 1 (cid:15)vn(cid:16)= ! !. m 2 2

Description:
This is the basic functional relation for the gamma function. It should be noted that it is a . Repeating this, with the integrated part vanishing at both end points each time, we finally get Handbook of Mathematical. Functions with
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.