Professional Documents
Culture Documents
Special Functions
Special Functions
> (5 -- 1) n=O oo 1 q — jj _ rP7"9 —: lim s(t lim > (— +.) lim s@) n=0 by Abel’s continuity theorem for power series. From the series — log(1 — +) = yt" /n, and Simpson’s dissection with w = e?"'/9, we get q—t s(t) = —1?~4 log(1 — #7) + S> wo”? log(1 — a") n=0 — 77 q-l — (¢t?-? — 1) log(i —t ~”P log(1 — w"t). = = ( )log(i —1) + Sw”? log(1 — w"2) a=l 1 = —t?~? log \ Let t — I” to get ql w(p/q) = —y —logqg + 5) wo”? log(1 — &"). n=] Replace p by g — p and add the two expressions to obtain q-l (2) +4 (4-2) = —2y — 2loggq +2) c0s( "2 ) log(1 — w”). q n=l The left side is real, so it is equal to the real part of the right side. Thus q-l — 2 2 v(2)+¥(4? P) =-2y—210g4+ S cos 7p log( 22008 =*). q q = q q (1.2.17) But w(x) -—wd-x)= * log (Qa) — x) = —m cotrx.1.3 The Hurwitz and Riemann Zeta Functions 15 So v(p/q) ~ Wl — (p/q)) = —x cotmp/q. (1.2.18) Add this identity to (1.2.17) to get ic 2 2 v(2) =-y ~~ cot? — tog + = S~ cos ae log(2— 208 ="), q 2 4q 2° q q (1.2.19) But cos 21 (q —n)/q = cos2mn/q, so the sum can be cut in half, going from 1 to |q¢/2], where |x| denotes the greatest integer in x. Thus Pp 7 mp we 27 np 2mn v(2) = -y — =cot— —logg+ >> Cos lop (2-2 cos =" ) q 2 4 q q n=] 1@/2] iT 2mn mn = ~y ~ 5 cot" —logg +2 S "cos 2 P top(2sin™"). a q q q n=] 13 The Hurwitz and Riemann Zeta Functions The half series ~ 1 xX,s)= — forx>0, 1.3.1 g(x, s) d Gaay x (1.3.1) called the Hurwitz zeta function, is of great interest. We have seen its connection with the gamma function for positive integer values of s in the previous section. Here we view the series essentially as a function of s and give a very brief discussion of how the gamma function comes into the picture. The case x = | is called the Riemann zeta function and is denoted by f(s). It plays a very important role in the theory of the distribution of primes. The series converges for Res > 1 and defines an analytic function in that region. It has a continuation to the whole complex plane with a simple pole at s = 1. The analytic continuation of ¢(s) up to Res > 0 is not difficult to obtain. Write the series for ¢(s) as a Stieltjes integral involving |x]. Thus for Res > 1 “asf Lx Jdx ] s+] ] x 1 ~|xl—x =1+— 45 / yea dx. oO oats [Ea ns | x’ xs + n=] The last integral converges absolutely for Res > 0 and we have the required16 1 The Gamma and Beta Functions continuation. The pole at s = 1 has residue 1 and, moreover, , I _ ~ Ix) —x tim {e(s) - oh = 1+ [ 2 dx lim (+f “5s as) now 1 Xx I tim. (- mT wen =y. (1.3.2 m=] \| I The best way to obtain analytic continuation to the rest of the plane is from the functional relation for the zeta function. We state the result here, since the gamma function is also involved. There are several different proofs of this result and we give a nice one due to Hardy [1922], as well as some others, in the exercises. In Chapter 10 we give yet another proof. Theorem 1.3.1 For all complex s, nD (s/2)¢(s) = 2 OAT — s)/2)CU — 5). (1.3.3) If s < 0, then 1 —s > 1 and the right side provides the value of ¢(s). This relation was demonstrated by Euler for integer values of s as well as fors = 1/2and s = 3/2. He had proofs for integer values of s, using Abel means. An interesting historical discussion is contained in Hardy [1949, pp. 23-26]. The importance of (s) as a function of a complex variable in studying the distribution of primes was first recognized by Riemann [1859]. The last section contained the result (- 1)-122k-1 — hehe 2k éQ2k) = Qk)! Bun. The following corollary is then easy to prove. Corollary 1.3.2 —] ¢(1 — 2k) = T-Bar, £0) = -5 and ¢(—2k)=0 fork =1,2,3,.... (1.3.4) Corollary 1.3.3 1 c'(0) = 5 log 2z. (1.3.5) Proof. From the functional equation and the fact that re) 5)1.3 The Hurwitz and Riemann Zeta Functions 17 we have _ T'(s/2) _ _ _— stT/2 ATS _ cls) sm Ps — DE). (13.6) Now (1.3.2) implies that (s — 1)¢(s) = 1 +. y(s — 1) 4+ A(s — 1) +---. So take the logarithmic derivative of (1.3.6) to get sd-s)_, _1,(2 1 f2es)_v teas Do ci=sy 2 (3) (>) I+y(s—-Ite Set s = 1 and use Gauss’s result in Theorem 1.2.7 with p = 1 and q = 2. This proves the corollary. There is a generalization of the last corollary to the Hurwitz zeta function ¢ (x, s). A functional equation for this function exists, which would define it for all complex s, but we need only the continuation up to some point to the left of Re s = 0. This can be done by using the function £(s). Start with the identity CO C(x, 8) — (C(s) —sxt(s +I) =a + Son [0 +x/ny% — (1 = sx/n)]. n=l The sum on the right converges for Re s > —1, and because ¢(s) is defined for all s, we have the continuation of f(x, 5) toRes > —1. The following theorem is due to Lerch. Theorem 1.3.4 ag (x, 8) P@) =] . 1.3.7 ( as )., ve V 20 ( Proof. The derivative of the equation ¢(x + 1,5) = €(x, s) — x * with respect tos at s = 0 gives (Beet) _ (S) — log x. (1.3.8) Os s=0 as s=0 For Res > 1, a°r(x, 5) ~ 1 ee ] ras 5a = (8 + du Geax SO d?> (actx,s) _ 1 i= as )_- case C89)18 1 The Gamma and Beta Functions Now (1.3.8) and (1.3.9) together with (1.2.14) of Theorem 1.2.5 imply that (aS) = C + log P(x). as s=0 To determine that the constant C = —} log 21, set x = 1 and use Corollary 1.3.3 This completes the proof of Lerch’s theorem. W@ Fora reference to Lerch’s paper and also for a slightly different proof of Theoren 1.3.4, see Weil [1976, p. 60]. 1.4 Stirling’s Asymptotic Formula De Moivre [1730] found that n! behaves like Cn"*+!/*e—" for large n, where C is some constant. Stirling [1730] determined C to be /27; de Moivre then used ¢ result of Stirling to give a proof of this claim. See Tweddle [1988, pp. 9-19]. Thi: formula is extremely useful and it is very likely that the reader has seen application: of it. In this section we give an asymptotic formula for (x) for Re x large, wher Im x is fixed. First note that log D(x ++ 1) = Soy_, log(k + x) + log P(x + 1) We then employ the idea that an integral often gives the dominant part of the surr of a series so that if the integral is subtracted from the series the resulting quantity is of a lower order of magnitude than the original series. (We have already used this idea in Equation (1.3.2) of the preceding section.) In Appendix D we prove the Euler—Maclaurin summation formula, a very precise form of this idea when the function being integrated is smooth. Two fuller accounts of the Euler—Maclaurin summation formula are given by Hardy [1949, pp. 318-348] and by Olver [1974, pp. 279-289]. Theorem 1.4.1 V(x) ~ V2mx "ee as Rex > 00. Proof. Denote the right side of the equation n—-1 log V(x +n) = S - logtk +x)+log T(x + 1) k=] by c,, so that Cat — Cn = loge +n). By the analogy between the derivative and the finite difference we consider c, to be approximately the integral of log(x + n) and set Ch = (n+ x) log(in4+ x) -(n+x)4+d,.1.4 Stirling’s Asymptotic Formula 19 Substitute this in the previous equation to obtain log(x +n) = (n+1+x)logn+1+x)—-(@+x)logn4+x)4+d,41—-d, —1. Thus 1 dna) —d, =1—(n+x4+ 1) og (1 + —) n+x 1 1 1 1 1 n+x 2n+tx) 3(n+x)3 = + + — Wn+x) 6(n+x)? Proceeding as before, take =1- (4x0! 1 dn = €n — 5 log(n + x), and substitute in the previous equation to get 1 1 1 1 ntl — €n = = 1 1 + — | — —— + ——— +: Ent — € 5 og( +] 2n +x) + 6m ax I i ~ 12(n+x)2 +0(—5). Now n-1 n—1 1 l én —@0 = > (en — &%) = >> st (ata) (1.4.1) k=0 k=0 therefore, lim, o0(€n — €0) = K(x) exists. Set K(x)+ +O = K(x) + —————_ ——— ], “ 12(n + x) (n+x) where K(x) = Ki(x) + eo. The term (n + x)7! comes from completing the sum in (1.4.1) to infinity and approximating the added sum by an integral. So we can write Ch = (n+x) logan +x) -(n+x) —- 5 login + x) + log C(x) + | + o(—). 12(n + x) (n+ x)? where K (x) = log C(x). This implies that - weet 1 46f 1 P(x+n)=C(x)(n+x) exp| "TOF atlas) (1.4.2)20 1 The Gamma and Beta Functions We claim C(x) is independent of x. By the definition of the gamma function _ Pats) y. TR). @a yx PO) TO) lim ———n = —— lim n = -_—— = ] noo D(n + y) T'(y) n> (Y)n iy) VG) Now, from (1.4.2) and (1.4.3) we can conclude that _~U(n+x) C(x) fim (: Ven _ C(x) C(O) n (1.4.3) l= li a nce” Fin) C(O) 2300 Thus C(x) is a constant and T(x) ~ Cx* '?e-* as Rex > ov. To find C, use Wallis’s formula: 2" (nl)? 1 = lim — Va = lim (Qn)! Jn 220 (2p 2n+] g—2n+0(1) l = lim ———_——_ : => n—> 00 C(Qn)y"t2e-*t9G) Jn Cc J2- This gives C = 27 and proves the theorem. Observe that the proof gives the first term of an error estimate. ®@ We next state a more general result and deduce some interesting consequences. A proof is given in Appendix D. For this we need a definition. The Bernoulli polynomials B, (x) are defined by tet oO " = 3 Bux) =. (1.4.4) = n! e —1 The Bernoulli numbers are given by B, (0) = B, forn > 1, Theorem 1.4.2. For a complex number x not equal to zero or a negative real number, m 1 1 Bo; 1 log (x) = 5 log 2m + (x-5) logx —x+ 5° 1 j=l (27 —1)2j x2/-1 1 f* Bont — [#) “in, wD “ The value of log x is the branch with log x real when x is real and positive. The expansion of log '(x) in (1.4.5) is an asymptotic series since the integral is easily seen to be O(x~”""*!) for larg x| < 1 — 6,5 > 0.1.4 Stirling’s Asymptotic Formula 21 From this theorem the following corollary is immediately obtained. Corollary 1.4.3 For é > Oand |argx| < a — 4, T(x) ~ Joaxt Pe as |x| > 00. Corollary 1.4.4 Whenx =a-+ib,a, 00, then [F(a + ib)| = V2 [ble 711 + OCB), where the constant implied by O depends only on a; and ap. Proof. Take |b| > 1,a > 0. It is easy to check that the Bernoulli polynomial By — Bz(t) =t — 27. Thus 4|B2 — B2(t)| < 5\fC1 — | < } forO0, arctan — + arctan — ={ a, 5 ifb <0 This gives xn a -barctan® = -6] 2 —F +0(3 )| 1 =-Flbl+a+o(— Putting all this together gives 1 1 1 log |I'(a + ib)| = (« - ;) log |b| — a + 5 log2m + o(a5).22 1 The Gamma and Beta Functions The condition a > 0 is removed by a finite number of uses of the functional equation (1.1.6) and the corollary follows. Observe that the proof only uses a = o(|b|) rather than a bounded. @ Corollary 1.4.5 For \jargx| <2 —8,5 > 0, By; 1 W(x) = lex gaat 0( se): Corollary 1.4.4 shows that (a + ib) decays exponentially in the imaginary direction. This can be anticipated from the reflection formula, for 1, 1 W r(; + w)\r(; 7 ib) = coshrb’ or 1 2 Qt . _— ~ —n|b| r(5 +0) ~ xb 4 p-ab 2ne 7 as b> +00, or 1 r (5 + is) ~ Smee 7/2 ag bh > oo. Similarly, rGb)r(—ib) = ——. = __* ee “bsinnbi ble — eb) and [Tb] ~ V2n |b} e712 as bh > 00. Since I(x) increases rapidly on the positive real axis and decreases rapidly in the imaginary direction, there should be curves going to infinity on which a normalized version of (x) has a nondegenerate limit. Indeed, there are. See Exercise 18. 1.5 Gauss’s Multiplication Formula for [(mx) 1 (2)2 = 27" (5) ¢ : ) together with the definition of the gamma function leads immediately to Legendre’s duplication formula contained in the next theorem. The factorization Theorem 1.5.1 | 2a—1 1 Pear @ =2 ror(a+5). (1.5.1)1.5 Gauss’s Multiplication Formula for [ (mx) 23 This proof suggests that one should consider the more general case: the factor- ization of (@)mn, Where m is a positive integer. This gives Guass’s formula. Theorem 1.5.2 Tima) Qn)" _— m™2 7 (ay (« 4 =) 2 (« + “—). (1.5.2) Proof. The same argument almost gives (1.5.2). What it gives is (1.5.2) but with mld 1\ m-— 1 (27)? m2 replacedby V[ —]...0| ——— ] = P. (1.5.3) m m To show that (1.5.3) is true, we show that p2 _ (Qxy"7} By the reflection formula So it is enough to prove nm . an _ (m—lr gm-l gin sino... sin ED =m. m m m Start with the factorization x" — 1 m—1 x—-— 1} = | [@ - expkxi/m)). k=] Let x — 1 to obtain m—] m= [Ia — exp(2ki/m)) k=1 m1. 0. 20 (m —1)x =? sin — sin —— +--+ sin —————_-. m m m This proves (1.5.3). Remark 1.5.1 A different proof of (1.5.1) or (1.5.2) that uses the asymptotic formula for (x) and the elementary property "(x + 1) = xI'(x) is also possible. In fact it is easily verifed that P¢x)U (x + 1/2) Pd/2)0 2x) satisfies the relation g(x + 1) = g(x). Stirling’s formula implies that g(x) ~ | as x — oo so that lim,-.o g(x +n) = 1 when n is an integer. Since g(x +n) = g(x) we can conclude that g(x) = 1. A similar proof may be given for Gauss’s formula. This is teft to the reader. g(x) = 224 1 The Gamma and Beta Functions An elegant proof of the multiplication formula using the integral definition of the gamma function is due to Liouville [1855]. We reproduce it here. The product of the gamma functions on the right side of (1.5.2) is exe) fora) co [ eo tly xf las | ex dys | ew xan i)/m)— ‘dtm -[ [- fre en Orrbmat bi) ga La (1/1 OHO DIMO-| By dx +++ AX. Introduce a change of variables: zm xy = ——— ,X2 = X2,...,Xm = Xp. Xo --++Xp The Jacobian is easily seen to be MZ NQNZ°+° Xp and the integral can be written [Lol boone: rota) ot /m)— 1 atm I)/m)—1 mz” 7 x { ———— X> . ————dzd x» --- dX,» Xo +++ Xp XoxX4 +++ Xp Set t = x2 +23 4+-++ + Xm +2 /(x2x3-++X»), and rewrite the integral as mf [- ofre magma dy MMT IMDAT | OM DIMAV 7 dey. dx. (1.5.4) First compute eal 00 ml ; i= /[ af eT dandy dy. 0 0 a Clearly, m—1 aft zn mam! fo [ “Ts (j/m)— oa dx2-++dXm dz Xj41 2+ Xm . Now introduce a change of variables, XQ = 2" / (XAG ++ Xm), AZ = ABs 0s Lp = Am,1.5 Gauss’s Multiplication Formula for F Gx) 25 and th = x3 txg te tXy $y 7 / (03 +++ Xm x1). The Jacobian is J = =—-—_ ; X7X3°°*Xm-1 and a is given by di oO oO zm (1/m)-1 men f eae e "|| —_— dz 0 0 XjX3 +--+ Xm m-1 (j/m)—i dx,dx3 tte dxXm Tee 2” [xy =2 oO al m—1| =m fo fe TL si ass din 0 0 i=? IF —mI il Therefore, r=Ce™. To find C, set z = O in the integral for J as well as in the above equation and equate to get 2 —1 r(+)r(2) (==) =C. m m m By (1.5.3), C = (20) DPm-? and I = (27) "VP m-'/?e—™, Substitution in (1.5.4) gives Pia)l (a+ 1/m)---Tia+(m— 1)/m) = miZanye-rr f ene mal ge 0 — m\/2-™9 (97 )"F T (ma), which is Gauss’s formula. Remark 1.5.2. We pointed out earlier that 1 / T(x) is a half of sin zx. In this sense the duplication formula is the analog of the double angle formula ] sin2ax = 2sinaxsinz (: + 3) This is usually written as sin27x = 2sinaxcoszx and so is thought of as a special case of the addition for sin(x + y). The gamma function does not have an addition formula.26 1 The Gamma and Beta Functions 1.6 Integral Representations for Log I'(x) and (x) In (1.2.13), we obtained {1 1 = x-1 —~wl)= — = a vO) ~ lm =H) ETD Fh from the product for 1/ T(x). We start this section by rederiving it from the beta integral. Note that, for x > 1, CO yp abd se" l-e _ _ x-2 _ _— (x vf tr log(1 — tdt ki hath k=0 by term-by-term integration, which is valid because of uniform convergence in [O, 1 — €]. Now lete > 0. By Abel’s continuity theorem for power series, —(x-1) [ t~* log(1 —tdt = » a) 0 ay (K+ I(x +k) We can introduce log(1 — fr) in the beta integral fy t*-2(1 — t)dt by taking the derivative with respect to y. In that case, a Try +1) a 1 (x - v= | m1 —1’dt= dy Jo dy T(x+y) or ] (x — » f (1 — 1) log(1 — n)dt 0 _T@rot+Dr@ty) -Pero+ bdr +y) - T(x + y)? The case y = 0 gives the necessary result. The differentiation is justified since the integrands involved are continuous. Some care should also be taken of the fact that the integrals are improper. The details are easy and left to the reader. The next theorem gives the integral representations of (x) due to Dirichlet and Gauss. Theorem 1.6.1 For Rex > 0, fel. 1 _ (i) w(x) = [ ; (« daa i" sz) (Dirichlet), (ii) w(x) = [ (< — poe) (Gauss). 0 x l—e~:1.6 Integral Representations for Log (x) and w(x) 27 Proof. (i)Evaluate the integral [;° f; e~'“dtdz in two different ways by changing the order of integration to get the formula OO .-Z 4-SZ [ ff dz = logs. (1.6.1) 0 Zz Similarly, the double integral e342) [ [os —_—_——_—__._—_dsdz xz when first integrated with respect to z yields (by (1.6.1)) CO d oO —s x1 _ —s x1 _ yv | e's logs ds = + | e °s* 'ds = T(x). 0 dx Jo If we integrate the double integral with respect to s we get P(x) “it ext d » | -( ay) ‘ Equate the last two expressions to get Dirichlet’s formula. (ii) Gauss’s formula is obtained from Dirichlet’s by a change of variables: OO poz oO dz in ( [Lae [ ao) 50+ 5 zit+z)* co e ( [Sac [0 a log(1+6) l1—e re eF oo et elk cry [See [oa SE 6—0t iog(1+8) tf l-—e- ro — — dz, [LE Zz l-e es) 3 1 8 < —dz = log ——_—— — 0 as 8— 0. [ons z log(1 + 8) (x) lim 630+ since log(i+é) e7 / —dz 5 Zz This proves (ii). The integrated form of the last theorem is given in the next result. Theorem 1.6.2. For Rex > 0, , _{* + _ G+) 049) at ” lr) = f Ga, - log(1 + 1) )¢28 1 The Gamma and Beta Functions and (ii) log (x) = [ ( — let — =) ar 0 l—e" i Proof. The integrals in Theorem 1.6.1 are uniformly convergent for Rex > 8 > 0, so we can integrate from | to x under the sign of integration. The integrals in Theorem 1.6.2 are the corresponding integrated forms. A change of variables u = e~ in (ii) gives Mya! du logs) = [ ——— -—x+4+]1]—. (1.6.2) 0 l—u logu There are two other integrals for log F(x) due to Binet that are of interest. These are given in the next theorem. A proof of one of them is sketched and the other is left as an exercise. See Exercise 43. Theorem 1.6.3 For Rex > 0, () logr@)=(x-+)1 + Jog? [ ett Ne, i og TPE AE OBA NAL Gy OBAT TH | 577 to] ; and © arctan(t /x) etn —] l 1 (it) log P(x) = («-5) logx —x + 5 logan +2 0 Proof. Gauss’s formula in Theorem 1.6.1 together with Equation (1.6.1) give d 1 a | 1) = Slog P(x +1) = — +logx — <8 agp, Wat = 7 logl(e + 1) = 5 + logx [ (5 r+os)e Integrate from | to x, changing the order of integration to get log P(x +1) 2\y +4 [ vai, tt jet ne, 0 = _ _— -—- —_—_____(f. Bias BT DP OBE TET IF Na tay f Use log lx + 1) = log F(x) + log x to rewrite the above formula as log P(x) My +14 [0 poe Ne al Oo = ——|logx — —~~- — BEDE NE PBR — » \2 t et) where r= fo Poi, od yet), (1.6.3) dp M2 tt eats rn! ~ Stirling’s formula applied above gives J = 1 — (1/2) log 27. The second Binet formula can be used to derive the asymptotic expansion for Jog I(x) contained in Corollary 1.4.5.1.7 Kummer’s Fourier Expansion of Log ['(x) 29 Expand | /(e?”! — 1) by the geometric series and integrate term by term to see that co 72k-1 — TQkK)E 2k) (—1yk-1 B2k (1.64) 9 emt (2m) Ak The last equality comes from Theorem 1.2.4. Now, I¢?e0 0 “ 142 gives, after integration, t if if (—Ayr ert ly" ft etdz arctan/2) =~ 33 5s Oy 7 eel Te f Pte Substitute this in Binet’s formula (ii) and use (1.6.4) to arrive at By 2j(2j — xs! n DO t H ool [ ( [ Saat). xen 0 o x +2 ect? — | For jargx| < 5 —€,€ > 0, it can be seen that Il < esc ze for all z > 0. This implies that the last term involving the integral is Op —;). So we have the asymptotic series but only for Jargx| < 5 — € instead of |argx| < m7 —€. Whittaker and Watson [1940, §13.6] show how to extend the range of validity. It is also possible to derive an asymptotic formula for log (x) from Binet’s first formula. See Wang and Guo [1989, §3.12]. For references to the works of Gauss, Dirichlet, Binet, and others, see Whittaker and Watson [1940, pp. 235-259]. l (x) = ] ay on +5 Paes x gx—-x ig 270 . 1.7 Kummer’s Fourier Expansion of Log T(x) Kummer [1847] discovered the following theorem: Theorem 1.7.1 ForO 1, Now use integral (1.6.2) for log T(x) in (1.7.3) so that D; = [ [ leu) e+ sin 2kaxdudx o Jo l—u log u l 1 ] [ sin 2kaxdx = 0, [ x sin2kaxdx = ———., and (1 — u)2ka u((log u)* + 4k22) The first two integrals are easy to solve and the third is the imaginary part of I - | er lloguttkri) 4. I uot u Jo u logu+2k7i I [ u*—! sin 2kaxdx = 01,7 Kummer’s Fourier Expansion of Log ['(x) 31 Therefore, Dp, = [ —2kn ,)\ au a o \u((ogu)? + 4k22*) 2kw J logu’ —2knt D, = to [ — eW hat at Qkn Jo \L4+? t Take & = | and we have D, _ i * il — eon at 20 6 ] + fz r Moreover, x = 1 in Dirichlet’s formula (Theorem 1.6.1) gives y 1 [ff _, 1 dt eS e — ——— —, 2x 2m Jo l+t/t where y is Euler’s constant. Therefore, 1 fSet—e 1 f*f 1 1 \dt peat Pent (4-4) 2n =a Jo t 2x Jo \14h l4t/t By (1.6.1), the first integral is log 27 and a change of variables from ¢ to 1/t shows that the second integral is 0. Thus or, with uw = e D, = 4+ 1092 =-—- + —lo . 1 On on Oe To find D,, observe that 1 oO ,—2nt —2kat ] kD, — D, = =f ff at = —logk, 20 0 t 2m where the integral is once again evaluated by (1.6.1). Thus I D, = ——(y +log2km), k=1,2,3,.... 2k The Fourier expansion is then ] —. cos 2rkx log T(x) = 5 og 2m + S i k=] k=l 1 li + log? > Sinden — 0 en a ne qh TOetm 2k ; 1 3 logk . 3p — —— sin . - ; Xx k=l Apply (1.7.1) and (1.7.2) to get the result. Kummer’s expansion for log (T(x) / 27.) and Theorem 1.3.4 have applications in number theory. Usually they give different ways of deriving the same result. This suggests that the Hurwitz zeta function itself has a Fourier expansion from32 1 The Gamma and Beta Functions which Kummer’s result can be obtained. Such a result exists and is simply the functional equation for the Hurwitz function: 27r(l—s)[ . 1 Sxcos2mnx 1 sin2mrx E(x, s) = ——_—_ (Qn yi 5 {sin drs So SIME cosas J INA ‘ (1.7.4) The functional equation for the Riemann zeta function is a particular case of this when x = 1. See Exercises 24 and 25 for a proof of (1.7.4) and another derivation of Kummer’s formula. 1.8 Integrals of Dirichlet and Volumes of Ellipsoids Dirichlet found a multidimensional extension of the beta integral which is use- ful in computing volumes. We follow Liouville’s exposition of Dirichlet’s work. Liouville’s [1839] presentation was inspired by the double integral evaluation of the beta function by Jacobi and Poisson. Theorem 1.8.1 If V isaregiondefinedbyx; > 0, i= 1,2,...,n;andS~ x; <1, then for Rea; > 0, [- [tris wl yteMdyy dx, = = F(t Ser Proof. The proof is by induction. The formula is clearly true for n = 1. Assume it is true form = k. Then for a (k + 1)-dimensional V ai—1 xem 1 41-1 J- fx ES dx\dx>-- -AX 4 1 1-x, l—x] ~ 4) —-- xy i 1 ay a Oe41— A f -f Kp Ky TY ‘dxgy1-+-dx} 0 0 0 1- Xp TX] xt —1 —l eet -(1— xy — +++ — xy) dxpdxy_} + - dx; - Now set x, = (1 — x; —--+ — xx_1)f to get 1 pi- Laxy xg ] x x] Xk-2 wn 1 oy) eee xy . “Xp y +1 Jo Jo 0 -(l—xj—--: =, yeeten pl — tt)" dtdx,_ 1--dx; _ Te) Or +1) + 1) [ [ x -f mrp Mel (Ox + Oey + D + O41) + 1) L x" i xy “ ko ‘a (1 —~— Xp att — xy) xy, s+ dx).1.8 Integrals of Dirichlet and Volumes of Ellipsoids 33 Compare this with the integral to which the change of variables was applied and use induction to get P@ Per +1) e+ esi) (Tar P@A)P (oe + oe) ail (ty + e41 + 1) r(1+ ditt a) This reduces to the expression in the theorem. Corollary 1.8.2 If V is the region enclosed by x; > 0 and S“(x;/a;)”' < 1, then a-to-t ayer _ LN Cai" / pi) Pi / Pid) [fs X5 xen dx Om = T+ Sei/py) . Proof. Apply the change of variables, y; = (x;/a;)"', i = 1,...,n. Then Oxi _ t Xj ay; Pi Yi and the Jacobian is ] 1X2 + Xn PiPr-** Pn ViYar** Yn The integral becomes Q a aay ai Sef fy CU /POTT | y@n/ P| dy dyy ++ -dVq P) P2- where V is defined by y; > O and \> y, < 1. The corollary now follows from the theorem. [[_, #Patt/pp rd+>) l/p) Jn particular the volume of the n-dimensional ellipsoid S°(x; jai)’ < lis Corollary 1.8.3 The volume enclosed by )> (x; /a;)"' <1, x; > Ois m"aiay ++ py rd+n/2) ~ Proof. For the first part of the corollary take a; = 1. For the particular case take P; = 2 and use the fact that (3) = 4./7. Corollary 1.8.4 If V is given by x; > Oand $)(+)?' 0, t, < 2%; /a;)”' 0, a1 Xi = 1, then _ _ Il V'(a;) bes ab eld od — / [x Xn X] Xn-1 rea) This is a surface integral rather than a volume integral, but it can be evaluated directly by induction or from Corollary 1.8.2. It is also a special case of Theorem 1.8.5 when f(u) is taken to be the delta function at u = 1. This function is not continuous, but it can be approximated by continuous functions. 1.9 The Bohr—Mollerup Theorem The problem posed by Euler was to find acontinuous function of x > Othat equaled n! atx =n, an integer. Clearly, the gamma function is not the unique solution to this problem. The condition of convexity (defined below) is not enough, but the fact that the gamma function occurs so frequently gives some indication that it must be unique in some sense. The correct conditions for uniqueness were found by Bohr and Molierup (1922]. In fact, the notion of logarithmic convexity was extracted from their work by Artin [1964] (the original German edition appeared in 1931) whose treatment we follow here. Definition 1.9.1 A real valued function f on (a, b) is convex if Fax + (—Ajy) < Af) + -ASO) forx,y € (a,b) and0 0 and (i) fl) = 1, (ii) F(x +1) = xf (x), and (iii) f is logarithmically convex, then f(x) = T(x) for x > 0. Proof, Suppose n is a positive integer and 0 < x < 1. By conditions (i) and (ii) it is sufficient to prove the theorem for such x. Consider the intervals [n,n + 1], [In +1,n +14], and [x + 1,n +2]. Apply (1.9.1) to see that the difference quotient of log f(x) on these intervals is increasing. Thus f(in+1) 1 f(a+1+x) f(v+2) —__—. < — log ——————_ < log ———_. f{n) x f+) fin +1) Simplify this by conditions (i) and (ii) to get (x+n)(x +n —1)---xf (x) ni log xlogn < log | < x log(n + 1). Rearrange the inequalities as follows: x(x + 1)-+-(x +n) og ———_——_—_—— nin* 1 0<]1 + log f(x) = xlog(1+ 7). n Therefore, nin* => li = P , I(x) ned, x(x +1)---(e +n) (x) and the theorem is proved. This theorem can be made the basis for the development of the theory of the gamma and beta functions. As examples, we show how to derive the formulas oS T(x) -| er ldt, x>0, 0 and } rawr [ ed —1 dt = PON) x>0QO and y>0O. (1.9.2) 0 P(x + y) We require Holder’s inequality, a proof of which is sketched in Exercise 6. We state the inequality here for the reader’s convenience. If f and g are measurable nonnegative functions on (a, b), so that the integrals on the right in (1.9.3) are finite, and p and q positive real numbers such that 1/p + 1/q = 1, then b b I/p b 1/4 | fgdx < ( / ftax) ( | gidx) , (1.9.3) It is clear that we need to check only condition (iii) for log P(x). This condition can be written as Tax + By) 0,B>0 and a+f=l1. (1.9.4)36 1 The Gamma and Beta Functions Now observe that oO : Tax + By) = [ (ete ")*e te Par 0 and apply Hélder’s inequality with w = 1/p and 8 = 1 /q to get (1.9.4). To prove (1.9.2) consider the function Pa + y) B(x, y) Py) Once again we require the functional relation (1.1.14) for B(x, y). This is needed to prove that f(x +1) = xf (x). Itis evident that f(1) = Land we need only check the convexity of log f(x). The proof again uses Hilder’s inequality in exactly the same way as for the gamma function. We state another uniqueness theorem, the proof of which is left to the reader. f(x) = Theorem 1.9.4 If f(x) is defined for x > 0 and satisfies (i) f(1) = 1, (it) f(x +1) = xf (x), and (iii) lim, _, 05 f (x +n)/[n* f(n)] = 1, then f(x) = P(x). For other uniqueness theorems the reader may consult Artin [1964] or Anastassiadis [1964]. See Exercises 26—30 at the end of the chapter. Finally, we note that Ahern and Rudin [1996] have shown that log | (x + éy)| is a convex function of x in Rex > 1/2. See Exercise 55. 1.10 Gauss and Jacobi Sums The integral representation of the gamma function is V(x) I ert = e-tr—. ct 0 t Here dt/t should be regarded as the invariant measure on the multiplicative group (0, 00}, since d(ct) _ dt ct i To find the finite field analog one should, therefore, look at the integrand e~“r*. The functions e~" and #* can be viewed as solutions of certain functional relations. This point of view suggests the following analogs. Theorem 1.10.1 Suppose f is a homomorphism from the additive group of real numbers R to the multiplicative group of nonzero complex numbers C*, that is, f:RO-C1.10 Gauss and Jacobi Sums 37 and farty) = f@)FO). (1.10.1) If f is differentiable with f'(0) = c 4 0, then f(x) = e™. Remark 1.10.1 Wehave assumed that f(x) 4 0 for any x but, in fact, the relation g(x + y) = g(x)g(y), where g : R > C, implies that if g is zero at one point it vanishes everywhere. Proof. First observe that f(0+0) = f(0)* by (1.10.1). So f (0) = 1, since f (0) cannot be 0. Now, by the definition of the derivative, Fx +t) — Fe) _ mn ff) — fx) f f(x) = lim li t—>0 t0 t ~ fox) tim LO=LO t>0 t = cf (x). So f(x)=e". H Remark 1.10.2 In the above theorem it is enough to assume that f is continuous or just integrable. To see this, choose a y € R such that fj f(r)dt 4 0. Then FR) J f@dt = fo fatode = f° F@drt. So _ fr? f@at — f f@dt This equation implies that if f is integrable, then it must be continuous and hence differentiable. F() Corollary 1.10.2 | Suppose g is a homomorphism from the multiplicative group of positive reals R* to C*, that is, g(xy) = g(x)a(y). (1.10.2) Then g(x) = x° for some c. Proof. Consider the map f = goexp: R > C*, where exp(x) = e*. Then f satisfies (1.10.1) and g(e*) = e. This implies the result. A finite field has p” elements, where p is prime and » is a positive integer. For simplicity we take n = 1, so the field is isomorphic to Z(p), the integers modulo p. The analog of f in (1.10.1) is a homomorphism w: Zp) > C*. Since Z(p) is acyclic group of order p generated by 1 we need only specify w(1). Also, y(1)? = w (0) = 1 and we can choose any of the pth roots of unity as the38 1 The Gamma and Beta Functions value of y(1). We therefore have p different homomorphisms w(x) =e? ™*/P GF =0,1,...,p—1. (1.10.3) These are called the additive characters of the field. In a similar way the multi- plicative characters are the p — 1 characters defined by the homomorphisms from Z(p)* to C*. Here Z(p)* = Z(p) — {0}. Since Z(p)* is a cyclic group of order p — 1, we have an isomorphism Z(p)* = Z(p — 1). The p — 1 characters on Z(p)* can be defined by means of this isomorphism and (1.10.3). We denote a multiplicative character by either x or 7, unless otherwise stated. It is now clear how to define the “gamma” function for a finite field. Definition 1.10.3 For an additive character W; and multiplicative character y; we define the Gauss sums g;(x:), 7 =0,1,..., p — 1 by the formula p-l g(x) = D> xi) vj), (1.10.4) x=0 where we extend the domain of x; by setting x;(0) = 0. It is sufficient to consider g(x) := g(x), for when j 4 0, 8) = S- X(x)wjyx) = S- x Cx yer™t/P = xx) d— x@err/? =x) Doxa? = x(/)g(X)- (1.10.5) This formula corresponds to {, e/*x5-ldy = T(s)/j5, where j is a nonzero complex number with positive real part. When j = 0 in (1.10.4) the sum is S~, Xx), which can be shown to be zero when x (x) # | for at least one value of x. Theorem 1.10.4 For a character x, _ fo ifx #id, dX) = 1 1 ify cid. (1.10.6) Remark 1.10.3 The identity character is the one that takes the value 1 at each point in Z(p)*. Proof. The result is obvious for x = id. If x # id, there is a y € Z(p)* such that x (y) 4 1. Then xO) 0 x) = 30 x@y) = 3° x),1.10 Gauss and Jacobi Sums 39 which implies the theorem. There is a dual to (1.10.6) given by the following theorem: Theorem 1.10.5 For the sum over all characters we have _ f0 ifx #1, 2x) = _ “i feel (1.10.7) Proof. It is sufficient to observe that if x #4 1, then there is a character x such that x (x) 4 1. The theorem may now be proved as before. We now define the analog of the beta function. Definition 1.10.6 For two multiplicative characters x and y the Jacobi sum is defined by Jan = > xno). (1.10.8) x+y=1 The following theorem gives some elementary properties of the Jacobi sum. We denote the trivial or identity character by e. The reader should notice that the last result is the analog of the formula B(x, y} = TQ) Q)/ITG@ + y)]. Theorem 1.10.7 For nontrivial characters x and n, the following properties hold: J(e, x} = 0. (1.10.9) J(e,e) = p—2. (1.10.10) T(x. x7') = -x(- 1). (1.10.11) 8(x)8(n) Hxnx#e, then JX.nN= (1.10.12) g(xn) Remark 1.10.4 From the definition of characters it is clear that the product of two characters is itself a character and so the set of characters forms a group. The additive characters form a cyclic group of order p and the multiplicative characters a cyclic group of order p — 1. Also, xe) = x(x7!) = 1/x (x) and since |x (x}| = 1 it follows that x~!(x) = x(x). Proof. The first part of the theorem is a restatement of Theorem 1.10.3 and the second part is obvious. To prove (1.10.11), begin with the definition Foxx) =o x@x 71d -2) = SS x@d-x)7). x 10,1 Now note that as x runs through 2,..., p — 1, then x(1 — x) runs through 1,..., p—2. The value y = p—1 = —1(mod p) is not assumed because x = y(1+y)7!. Therefore, Fixx) = So x0) =-x(-D, ye—140 1 The Gamma and Beta Functions by Theorem 1.10.4. This proves the third part. The proof of the fourth part is very similar to Poisson’s or Jacobi’s proofs of the analogous formula for the beta function. Here one multiplies two Gauss sums and by a change of variables arrives at a product of a Jacobi sum and a Gauss sum. Thus, for x7 + e, g(XxXosa) = S- x (x jer™2/P > nl yertir/P x y = So x@ eter x.¥ = So x@)niyy+ SO x@)ne — 1ye/?. x+y=0 t=x+y40 The first sum is 5°, x(x)n(—x) = n(-L) 35, xn) = 0 since yn # id. The second sum with x = st is S> xs)x)n(nd = sel? = So xn(tye"!? S$” x (s\n — 5) t£0,5 t = 2(xnJ (xX. ). This proves the fourth part of the theorem. We were able to evaluate '(s) in anice form for positive integer values and half- integer values of s. Evaluations of special cases of Gauss sums are also possible and important, but in any case the magnitude of the Gauss sum can always be found. Theorem 1.10.8 For nontrivial multiplicative and additive characters x and w, = /p. So xxv (x) Proof. By (1.10.5) it is enough to prove that |g1(x)|? = leGol? = p: Ie? = So x rye? S* eye OV? x y =o xory ermievr, xyeO Set x = ry. Then IsGO? = So xMemmernr rv£0 p-l =Sox+ SO xO DS erMOV?, I 140 or | yx401.10 Gauss and Jacobi Sums 4] The first sum is p — | and the inner sum in the second term is —1. Thus Is =p-1-Sox@=p—-l+l=p, tH] and the result is proved. Corollary 1.10.9 If x, and xn are nontrivial characters, then F(x. )| = JP. (1.10.13) Proof. This follows from Theorems 1.10.7 and 1.10.8. As an interesting consequence we have: Corollary 1.10.10 If p = 4n + 1 is a prime, then there exist integers a and b such that p = a* + b’. Proof. The group Z(p)* is of order p — 1 = 4n, which is also isomorphic to the group of multiplicative characters on Z(p)*. Since the latter group is cyclic there exists a character x of order 4 that takes the value +1, -t/. It follows that J(X, x) = a + bi for integers a and b. Since x? ¥ id, apply Corollary 1.10.9 to obtain the desired result. Corollary 1.10.10 is a theorem of Fermat, though Euler was the first to publish a proof. See Weil [1983, pp. 66-69]. Later we shall prove a more refined result that gives the number of representations of a positive integer as a sum of two squares. This will come from a formula that involves yet another analog of the beta integral. We have seen that characters can be defined for cyclic groups. Since any abelian group is a direct product of cyclic groups, it is not difficult to find all the characters of an abelian group and their structure. The following observation may be sufficient here: If x1 is a character of a abelian group G1, and x2 of G2, then we can define a character x : G, x Gz > C* by x(x, y) = x1(*) x29). We thus obtain n additive characters of Z(n) and ¢(n) multiplicative characters of Z(n)*. The Gauss and Jacobi sums for these more general characters can be defined in the same way as before. Gauss [1808] found one derivation of the law of quadratic reciprocity by evaluating the Gauss sum arising from the quadratic character. (A character x + id is a quadratic character when x” = id.) Details of this connection are in Exercise 37 at the end of the chapter. One problem that arises here, and which Gauss dealt with, is evaluating the sum G = ye e2tie iN Ag in Theorem 1.10.8 one can show that G? = +N depending on whether N = 1 (4) or 3 (4). The problem is to determine the appropriate square root for obtaining G. According to Gauss, it took him four years to settle this question. Dirichlet’s evaluation of )."=) e?7/*"/" by means of Fourier series is given in Exercise 32. 742 1 The Gamma and Beta Functions Jacobi and Eisenstein also considered the more general Jacobi sum Js X00 XO = S- X1(t1) X2(f2) «>> Xe(te). (1.10.14) ttethsl This is the analog of the general beta integral in Theorem 1.8.6. Eisenstein’s result, corresponding to the formula in Theorem 1.8.6, follows. Theorem 1.10.11 If x1, X2,..-, Xe are nontrivial characters and xX, Xo --+ X¢ is nontrivial, then _ (X18 (X2) ++ BO) g(XIX2° Xe) The proof of this is similar to that of Theorem 1.10.7, and the reader should fill in the details. In Section 1.8 the volume of n-dimensional objects of the form a,xj' + agxy + -+ + ayx,' < b was determined by means of the gamma function. In the same way, for finite fields, the number of points satisfying a,x}! +a2x37 +-+-+a,x;' = bcan be found in terms of Gauss sums. Gauss himself first found the number of points on such (but simpler) hypersurfaces and used this to evaluate some specific Gauss sums. Weil [1949] observed that it is easier to reverse the process and obtain the number of points in terms of Gauss sums. For an account of this the reader should see Weil [1974]. It may be mentioned that Weil’s famous conjectures concerning the zeta function of algebraic varieties over finite fields are contained in his 1949 paper. It also contains the references to Gauss’s works. One may also consult Ireland and Rosen {1991] for more on Jacobi and Gauss sums and for references to the papers of Jacobi and Eisenstein. The form of the Gauss sums also suggests that they are connected with Fourier transforms. Let F denote the vector space of all complex valued functions on Z(N), the integers modulo N. Let F be the Fourier transform on F defined by F(X, X20 +005 Xo) (1.10.15) N-I 1 . (Ff)(n) = Tw S f (xjerrinesN (1.10.16) It can be shown that the trace of this Fourier transform with respect to the basis {59, 5),-.., dn}, where Oo, x , soya {o 22? is the quadratic Gauss sum $3") e2"*"/", Schur [1921] gave another evaluation of this sum from this fact. The details are given in Exercise 47. One first proves that the fourth power of F is the identity so that the eigenvalues are +1, +i and the essential problem is to find the multiplicity of these eigenvalues.1.11 A Probabilistic Evaluation of the Beta Function 43 Discrete or finite Fourier analysis was not applied extensively before 1965 be- cause of the difficulty of numerical computation. This changed when Cooley and Tukey [1965] introduced an algorithm they called the Fast Fourier Transform (FFT) to reduce the computation by several orders of magnitude. The reader may wish to consult the paper of Auslander and Tolimieri [1979] for an introduction to FFT, which emphasizes the connection with group theory. Some of the earlier instances of an FFT algorithm are mentioned here. Computational aspects are also interesting. See de Boor [1980] and Van Loan [1992, §1.3]. 1.11 A Probabilistic Evaluation of the Beta Function When a and are positive integers, [so —xPldy = COVEN It seems that it should be possible to arrive at this result by a combinatorial argu- ment. But working with only a finite number of objects could not give an integral. Here is a combinatorial-cum-probabilistic argument that evaluates the integral. Choose points at random from the unit interval [0, 1]. Assume that the probability that a point lies in a subinterval (a, b) is b — a. Fix an integer n and let P(x; < t) denote the probability that, of n points chosen at random, exactly k of them have values less than t. The probability density function for P(x; < ft) is _ Ply 0 At Now P(x, , qa; € {0,1,...p— 1}, and the partial sums are integers that converge to the p-adic number. So, if there is a function f defined on the positive integers and the values of f at two integers that are p-adically close are close to each other, then f has a unique continuous extension to Zp. Define a function f on the positive integers n by the formula fa) =(-1)" Tk. pik It is not difficult to show that f(n + p”£) = f(n)mod p”, where n,m, and é are positive integers. Now n + p”é and n are p-adically close to each other and the values of f at these points are also p-adically close. Consequently, f has an extension to Z,. This extension gives the p-adic gamma function due to Morita [1975]. The p-adic gamma function is defined by p(x) = —f(x — 1). This function also has a functional relation and other useful properties. There is a formula of Gross and Koblitz [1979] that gives the Gauss sum as a product of values of the p-adic gamma function.46 1 The Gamma and Beta Functions A good treatment of the p-adic numbers and functions is given in Koblitz [1977]. An account of the p-adic gamma function and the Gross—Koblitz formula is available in Lang [1980], including a reference to a paper by “Boyarsky” [1980]. In fact, p-adic extensions of the beta function, and more generally, the Mellin transform, are also available. Exercises 1. Use the change of variables s = ur to show that oo oO Pari) = [ | syle“ tds 0 Jo isT(x+ y) B(x, .y): (Poisson) 2, Let f= f° e*’dx. Observe that 1? = f>° f° e+" dxdy. Evaluate this double integral by converting to polar coordinates and show that J = ./7 /2. 3. A proof of Wallis’s formula is sketched below: (a) Show that [ dx == 0 x?+a 7 2a (b) Take the derivative of both sides n times with respect to the parameter a to conclude that [ dx _—1-3-5++:2n-1n 1 0 (x2 +a)rti ~~ 2.4.6---2n 2 qntc/2- (c) Setx = y/./n,a = 1, letn — ov, and use Exercise 2 to obtain Wallis’s formula. 4. Evaluate fia — t?)*~'dt in two different ways to prove the duplication formula given in Theorem 1.5.1. To get another proof evaluate n/2 / sin! 26 d6 0 in two ways. 5. Suppose that f is twice differentiable. Show that f” > 0 is equivalent to f(ax + By) < af(x) + Bf(y) for a and f nonnegative anda + Bp = 1. 6. Convexity can be used to prove some important inequalities, for example, Holder’s nea 1/p <{f ifirax } {[ ttae| . where f and g are integrable functions and 4 pt ; = |. We sketch a proof here. l Fear10. 11. 12. Exercises 47 (a) Note that e* is a convex function; use this and the result of Exercise 5 to show that if u and v are nonegative real numbers then Equality holds if and only if u? = v7. (b) Deduce Hélder’s inequality from (a). It might be appropriate to call this the Rogers—Hdlder inequality since Rogers {1888] had the result before Hélder [1889]. Other important results of L, J. Rogers are discussed later in the book. . Here is another proof of the functional relation xt B(x, y) = Bee yt 1). Bix, y+ 1) -/ pet(—t) dt 0 and perform an integration by parts to show that Write Bix, y+ D= Ty BO. . Show that B(x, y) -[° ldt fy (c+ 9" Take the derivative with respect to c and derive the functional equation x+ Box. y) = = Bay +). Give a similar argument using ¢ | tle —1)dt. 0 Tis Pa + &)/n) (22 )e-D/2_0/2)-2 ° . Write Gauss’s formula as TQ) = Show that the right side satisfies all the conditions of the Bohr—Mollerup theorem. This proves the formula. Give a proof of Gauss’s formula by using the definition of T(x). Prove Gauss’s formula by the method given in the remark after Theorem 1.5.2. It is clear from (x + 1) = xI(x) that pr log C(t)dt = xlogx ~x+C. Show that C = + log 2x. Stirling’s formula will work, but there is a more elegant argument using Gauss’s multiplication formula first.48 1 The Gamma and Beta Functions 13. There is another beta integral due to Cauchy defined by °° dt m2*-* YT (x+y—1) C(x, y= —eeo SS EOD -oo (1+ity (it) Px)r(y) (a) To prove this, show: (i) Integration by parts gives C(x, y+ 1) — 2C(x + 1, y). (ii) Write © {-1~it)4+2 C , = er (x,y) [. (l+tiry(-ityt! = 2C(x, y+ 1)—-C(x-1, y+ 1). , Re(xt+y)> 1. This together with (i) gives C(x, y) 2) cay ed) ,y) = — —C(x, ¥ ; w= oI ; (iii) Iteration gives 27" (n(n C(x, y) = ———— C(x +2. y +n). = Cty Da yn) oe dt C , = ooo? (x +n y +n) i. (1 +irytr(l _ itjyt" =[ dt oo + e2R + ity — in Set t > ¢/./n in the second integral and let n — >. (b) The substitution ¢ = tan @ leads to an important integral. Find it. 14. Use the method for obtaining Stirling’s formula to show that 1 1 1 | ttt sea Wit Ct et o(=a): c=-a4+va(I-5+B-Gt). where Sum " “1 Cn = (Ce —Ce-1) with ¢, = —= - S- k k-1 i DR k=1 and use some algebra to change c, — Cn—1 to an expression that goes to zero like n~*/ to show that (=F =(V2+)5— n=! » ala + Ti See Ramanujan [1927, papers 9 and 13] for further results of this type.Exercises 49 15. Here is an outline of a real variable proof of (1.2.1). Let No 1 . tan 2x N00 Ww n+x (a) g’(x) = —17/ sin? mx + Ss; I/(n+x)*. (b) g’(x) is continuous for 0 < x < Lif g’(0) = g/(1) =0. (c) g'(x/2) + g'((x + 1)/2) = 4g"(x). (d) Let M = maxoex<1 [g’(x)|. Then M < M/2so M =0. (e) a(x /2) — g( + 1)/2) = 2m /(sinazx) — 23> (—1)"/(n +x). —&% (f) g(x + 1) = g{x). (g) g(x) = constant. (h) | P+ odte= 0 (-)"/nt+xyt 0D" /(n 41-2) 0 n=) n=0 oO =S°(-b"/ +4), —oO so (1.2.1) holds. This proof, due to Herglotz, was published by Carathéodory f1954, pp. 269-270]. Bochner’s [1979] review of the collected works of Herglotz also includes this proof. 16. The following is Dedekind’s [1853] proof of [(a) (1 — x) = 2/sinax. Set co p71 ow = [ tar oO pro! ——dt = s * | st+1 O(x) oo 4x—l 1 dt=od(x)s*. [aaa o00 (a) Show that and (b) Deduce that x-l ox oo po! t—1 go SO. f 8) s—l 9 (st+1)(t+s) (c) Use the second formula in (a) to get [ p= fo ( ~ Eat \as: eu Jo stl [ t+s )as:50 17. 18. 19. 20. 1 The Gamma and Beta Functions then change the order of integration to obtain oo zx-1 oor = [Ear 0 t—1 y OO py—l _ poe | [o(x)Pdx = / ee at. l-y 0 t-—1 (e) Integrate (b) with respect to s over (0, 00) and use (d) to derive (d) Deduce x 3 _ © xl loge _ , a(x) J 160) ar=2 f dt = 29". (f) Show that @(x) = @(1 — x) implies ¢'(4) = 0 and / [o()Pdt = 2 / [e(t) at. l-x 1/2 (g) Deduce that o(x) / [o(t)Pdt = $'(x). 1/2 (h) Show that ¢ satisfies the differential equation @¢” — (’)? = 7. (1) Solve the differential equation with initial condition (5) = am and o'(5) = 0 to get P(x) =a cscrx. Show that 1 p-ld —p lar rar) = R O.R ; 0, . | la’ +b —pnP? abla +y)’ ex > ey>O0,a>0 Show that — Pa tix /or(t — ix/t) ie lim. ——————_____————- - e “ 1400 Paor(t) Prove that for a > 0, © si l b/2 / SMe dx = FT gh Seto)? 0 < Reb < 2, 0 xP 2 l(b) and oo 1 b/2 [ COS AN oy a bag? SEP) gC Reb 1. 0 x? 2 (db) For 4 > 0,x > O and —2/2 1 an integer, and retums to positive infinity. Hint: First prove that OO 45 1 et oC, o=a5 | foe" and then apply the ideas of the previous exercise. Note also that ¢(x, 5) is now52 25. 26. 27. 29. 1 The Gamma and Beta Functions defined as a meromorphic function by the contour integral with a simple pole ats = 1. . Prove the functional equation 2rad— 2 2 t(x,s)= Sou finns) 2 es 0. Show that the functional equation for ¢(x, s) easily implies (a) the functional equation for (s), (b) Kummer’s Fourier expansion for log T(x) /V22. The next five problems are taken from Artin [1964]. For 0 < x < 00, let @(x) be positive and continuously twice differentiable satisfying (a) @(x + 1) = (x), (b) (4)@(2*) = d(x), where d is a constant. Prove that ° is a constant. Hint: Let g(x) = £ log @(x). Observe that g(x +1) = g(x) and + q(a(5)+ (5) = g(x). Show that @(x) = ['(x) (1 —x) sin 2x satisfies the conditions of the previous problem. Deduce Euler’s reflection formula. - Prove that a twice continuously differentiable function f that is positive in 0 ov.30. 31. 32. 33. 35. 36. 37. Exercises 53 Prove that the example g(x) = >>°°, 3 sin(2"7rx) shows that just continuity is insufficient in the previous problem. Suppose f and g are differentiable functions such that f(x+y) = f(x) fQ)— g(x)e(y)and g(x+y) = f(x)g(y)+2(x) f(y). Prove that f(x) = e** cos bx and g(x) = e® sinbx, unless f(x) = g(x) = 0. Prove that S>N7} e27i°/N — Li" JN, where i = /—1. (a) Set f(t) = SONG} ett" Q < ¢ < 1, Note that f(0) = f(1) and extend f(t) as a periodic function to the whole real line. (b) Note that f(t) = 37°. ane?" , where ay = h rower Conclude that f(0) = 7) e 2mix? /N =" 4 (c) Show that a, = e~27#Nn’/4 Pennp! aniy" Nay. (d) Show that oo N- “me N-nN/2 amiy2/N ya=(Y/ “Sf jeray 00 Nn/2 Na/2 Reven Rodd a 2 = ai’) f ev IN gy, —& (e) Use Exercise 19 to evaulate the integral. Another way is to take N = | in (d). (Dirichlet) If p is an odd prime, then there is exactly one character x, that maps Z(p)* onto {+1}. Recall that Z(p)* is the integers modulo p without 0. Prove that xX,(a) = 1 if and only if x? = a mod p is solvable, that is, a is a square in Z(p)*. Usually one writes x, (a) = (5): which is called the Legendre symbol. . Prove that if a is a positive integer prime to p, then a?—!/? = (5) (mod p). Here p is an odd prime. (Use the fact that Z(p)* is a cyclic group.) For p an odd prime, use the previous problem to prove that (>)= = (-1)-)? and (2) = = (— 1)?’ ~1)/8 . (Use 9p/2 — (et/4 4 e TAP = _ (errs + e Pri/4y (mod p). Consider the two cases p = +1 (mod 8) and p = +3 (mod 8) separately.) Prove the law of quadratic reciprocity: For odd primes p and gq, (7)(4) = (- yee, (a) For S = > oP_ i a yermx/, show that $* = (—+ >). (The proof is similar to that of Theorem 1.10.8.) (b) Use (a) and Exercise 34 to prove that S77! = prs (*) (mod q). (c) Show that S4 = Poy (Eee ian/P = (£)S (modq). (d) Deduce the reciprocity theorem from (b) and (c). (Gauss) For integers a and N with N > 0, define Ga, N)= ee Be 2miax?/N (a) For p prime, show that G(1, p) = S7?_ i Ly ertix/P (b) For p prime show that G(a, p) = Cad, * p).54 38. 39. 40. 41. 1 The Gamma and Beta Functions (c) Prove that G(q, p)G(p, g) = G(1, pq) when p and q are odd primes. (d) Now use the result of Exercise 32 to deduce the reciprocity law. (Gauss) For a discussion of Exercises 32-37 and for references, see Scharlau and Opolka [1985, Chapters 6 and 8]. Prove Theorems 1.8.5 and 1.8.6. Prove Theorem 1.9.4. Weierstrass’s approximation theorem: Suppose f is a continuous function on a closed and bounded interval, which we can choose to be [0, 1] without any loss of generality. The following exercise shows that f can be uniformly approximated by polynomials on [0, 1]. (a) Show that it is enough to prove the result for f(0) = f(1) = 0. Now extend f continuously to the whole real line by taking f = Oonx <0 and x > 1, (b) Observe that (2n + 1)! (Gar Jent+1 (n! )2 1 / On(t)dt = 1. -1 Show that P, (x) = f ff + 1)Q,{(t)dt is a polynomial in x for x € [O, 1]. (c) Use Stirling’s formula to show that for 5 > Oand6é < |t| < 1,0,(1) > 0 uniformly as n — oo. (d) Note that for 0 f(x) uniformly on [0, 1], break up the integral into three parts, {- + P, + fe and use (c). Prove Plana’s formula (see Whittaker and Watson [1940, p. 145] for references to Plana): For positive integers m and n So) = Se | ods —i [ k=m m QOn{t) = y" is a polynomial such that 2 p(n + ty) —o(m + iy) — oH — iy) + o(m — iy) np ? where @(x + iy) is a bounded analytic function in m You might also like
The Subtle Art of Not Giving a F*ck: A Counterintuitive Approach to Living a Good LifeFrom EverandThe Subtle Art of Not Giving a F*ck: A Counterintuitive Approach to Living a Good LifeRating: 4 out of 5 stars4/5 (5823) The Gifts of Imperfection: Let Go of Who You Think You're Supposed to Be and Embrace Who You AreFrom EverandThe Gifts of Imperfection: Let Go of Who You Think You're Supposed to Be and Embrace Who You AreRating: 4 out of 5 stars4/5 (1093) Never Split the Difference: Negotiating As If Your Life Depended On ItFrom EverandNever Split the Difference: Negotiating As If Your Life Depended On ItRating: 4.5 out of 5 stars4.5/5 (852) Grit: The Power of Passion and PerseveranceFrom EverandGrit: The Power of Passion and PerseveranceRating: 4 out of 5 stars4/5 (590) Hidden Figures: The American Dream and the Untold Story of the Black Women Mathematicians Who Helped Win the Space RaceFrom EverandHidden Figures: The American Dream and the Untold Story of the Black Women Mathematicians Who Helped Win the Space RaceRating: 4 out of 5 stars4/5 (898) Shoe Dog: A Memoir by the Creator of NikeFrom EverandShoe Dog: A Memoir by the Creator of NikeRating: 4.5 out of 5 stars4.5/5 (541) The Hard Thing About Hard Things: Building a Business When There Are No Easy AnswersFrom EverandThe Hard Thing About Hard Things: Building a Business When There Are No Easy AnswersRating: 4.5 out of 5 stars4.5/5 (349) Elon Musk: Tesla, SpaceX, and the Quest for a Fantastic FutureFrom EverandElon Musk: Tesla, SpaceX, and the Quest for a Fantastic FutureRating: 4.5 out of 5 stars4.5/5 (474) Her Body and Other Parties: StoriesFrom EverandHer Body and Other Parties: StoriesRating: 4 out of 5 stars4/5 (823) The Sympathizer: A Novel (Pulitzer Prize for Fiction)From EverandThe Sympathizer: A Novel (Pulitzer Prize for Fiction)Rating: 4.5 out of 5 stars4.5/5 (122) The Emperor of All Maladies: A Biography of CancerFrom EverandThe Emperor of All Maladies: A Biography of CancerRating: 4.5 out of 5 stars4.5/5 (271) The Little Book of Hygge: Danish Secrets to Happy LivingFrom EverandThe Little Book of Hygge: Danish Secrets to Happy LivingRating: 3.5 out of 5 stars3.5/5 (403) The World Is Flat 3.0: A Brief History of the Twenty-first CenturyFrom EverandThe World Is Flat 3.0: A Brief History of the Twenty-first CenturyRating: 3.5 out of 5 stars3.5/5 (2259) The Yellow House: A Memoir (2019 National Book Award Winner)From EverandThe Yellow House: A Memoir (2019 National Book Award Winner)Rating: 4 out of 5 stars4/5 (98) Devil in the Grove: Thurgood Marshall, the Groveland Boys, and the Dawn of a New AmericaFrom EverandDevil in the Grove: Thurgood Marshall, the Groveland Boys, and the Dawn of a New AmericaRating: 4.5 out of 5 stars4.5/5 (266) A Heartbreaking Work Of Staggering Genius: A Memoir Based on a True StoryFrom EverandA Heartbreaking Work Of Staggering Genius: A Memoir Based on a True StoryRating: 3.5 out of 5 stars3.5/5 (231) Team of Rivals: The Political Genius of Abraham LincolnFrom EverandTeam of Rivals: The Political Genius of Abraham LincolnRating: 4.5 out of 5 stars4.5/5 (234) On Fire: The (Burning) Case for a Green New DealFrom EverandOn Fire: The (Burning) Case for a Green New DealRating: 4 out of 5 stars4/5 (74) The Unwinding: An Inner History of the New AmericaFrom EverandThe Unwinding: An Inner History of the New AmericaRating: 4 out of 5 stars4/5 (45)