DEFINITE INTEGRALS by MURRAY MANNOS Submitted in Partial Fulfillment of the Requirements for the Degree of Bachelor of Science in Mathematics Massachusetts Institute of Technology June 1937 Professor in Charge: Professor R. D. Douglass INTRODUCTION The first two chapters consist of the evaluation and extension of a number of definite integrals; whereas, in the third chapter, an attempt is made to show a possible method for obtaining the sum of a finite series by means of definite integrals somewhat of the form used in Chapters I and II. It is perhaps advisable to outline the useful notions and formulae which are used repeatedly in the evaluation of the definite integrals. This will enable the reader to follow the steps more readily without an undue repetition of the same theorems, etc., between equations. The notions and formulae, the proofs of which may be found in practically any text on Functions of a Complex Variable, will merely be stated as it is not deemed necessary to repeat such proofs since this would merely take up space. Methods of finding residues: (1) Simple pole as of function Then Lim (z - a) z-a if the Limit $ (z) = 1 f(z) z - a = (z) f (a) is the residue at is a definite number. a tHospital t s rule that, if (2) It follows from L' S(z) repeated factor of and if z - a and is a nonat the residue of V/(z), _4 ial_ is a are holomorphic at W(z) a - (a) ( f h (a) (3) If we have poles of higher multiplicity, then we use Taylorts expansion where f (z) a (z -a) is real or complex. $' (a) + (z - (a) + (z - a) (z - a) If n = 1 If n = 2 If n = 3 $ then + .... e... the residue at is (a) a) 2 •_a) 21 "1'(a) n nna _2! gn-1l If n = n (n-1) . Two useful theorems for showing that (where R--> oo and lim Z If lim (z - a) f (z) around r-- r z f (z) = K CO where =k f (z) dz o or around large circle k is a constant, z-- a lim R are the following: e91) K m SR f () dz = i (92 R--oo and 0 = r--o) I = (@2 - 1) k then ] The following useful formulae are given: z eiz + e-iz Cos z = - e,-.--- eiz sin z e + e e+e 2 cosh z - e-iz sin i z = i sinh z cos i z = cosh z sin 29 S 7T e z - e -Z sinh z 2i 2 = 1 = where 0 - z z2 g - 1 7r 2 TABLE OF CONTENTS Chapter I Chapter II Evaluation of Definite Integrals . . . . - Extension of the Evaluation of One Defirnite Integral to That of Many Others Chapter III - .. ..... . 20 . . . . . 23 . .............. Summation of a Finite Series by Means of Definite Integrals Chapter I EVALUATION OF DEFINITE INTEGRALS cos mx 0 x4 + a4 z.z za, a[Cos + i sin jvr---4+ 2k7Q] =- cos (7r+-2k7TI + 1~~sin (T--r where k = 0, 1, 2, 3 Consider imz z 4 + a4 dz 77 zi 1 = ae z2 = z3 = e = a e a 4 4 i 577 z4 4 I 777 % 4 Integral around r 0 9R- R4 - <R 4 --a 4 S0 2 e for im(R cos 9 + 1 sin 9) ------------------ - mR -iR SRe - sin 9 Q 0 R = 0 a4 2mR r d 2 R a4 R4 _ - 2R < ------- 4= ReigdQ -T o 2mR (R - a ) -mR sin 9 a0 as R--> im a Residue at 0( e Residue at ae 4 ima(cos T+i sT) sin 4 4 e 2t 4a S7r Residue at ae e 4 3 ma 4a3 (- -+i - e ima(cos 37T 3- + i sin 37T) 11) 4 4 4a 3 e - ma -F2 " e e 4 a0 (As 4 ma ,2- + i1e) Adding residues, we have g ma e 42 4 as ma ma i e 1 e-/2e 1 1 1 -2 41 S-1-1) i-J2 ma 7 -1 4 a3 (+ 1 + 2 ma 3 -()e ma a i i - 2 ema J2-2 e---F (e -7 4 a iea-3 2 sin (f4 ma +r2) 4 (-+i + 1 i) 2 ,-iJ (-J) e4 mx ei 0 - 00 x 4 2 e - 3 -----------a = + a4 dx ma ý sin( (a +7C) ma 2 Oo cos mx dx -0 x 4 + a 4 sin (E + -- ) e .............. = --- 4 3 a ma cos mx dx = 4 co o x4 + a e qT sin (Sý + - 2a 0o sin mx 4dx -So x+ 0 0 = x4 + a4 -co S00 ) . . . . . s. m x sin mx dx x4 + a4 and a positive Consider Seimz z4 +-----dz z4 + a 4 R = 0 Integral around ei Re mR for (cos 9 + i sin g) iRei 9 dg R4 e 419 + a4 0 r T2 -mR sin 9 2R 2 e-mRsiln 9 -- s2 e ------------- 4d9 e4 4 4 4 0 R -a R -a 0 2 d9 7T T7 S------S2 R4 - a4 2mR9 e 0 = - a4 )2 (R427T 229mR e dQ---2-R 2mR (R4 _ a4) -7TR- 4 [1 - e - R ] 4 _ a R -a as 0 R--7 7 0 e S Residue at OC i ma (- 1 + i iv 2 4a 47 e ma 4 a2 1 + ma 1) + ma 4 a i ma . 7E 2 4ma (e 4- Residue at ae 1 ) e------ 4 Residue at ae im Q -4 ma a2 1 e Adding residues ma ma e 2 = e--- sin ma 2 a2 si a2 ma '3 7 ie Sx ei mx x + adx a sin 2 JE ma oO x sin mxdx o oooxx4-------cos 4 dx + amx x S x +a o0 - 2 17fe sin- a + a = xx- o+ a dx = X 4cos mx 4 dx 0 ma o 0 x- sin mx dx x4 + a4 = 2a 2 sin ma -- 2E ma JE aŽ o00x 3 sin mx 4 x 44 + a4 0 S --------- dx z 4 eim dz z4 + a--------dz z + a4 0, R S -----------------------------R e i + a4 R 4 -mR sin 9 < 7 R--e 4 0 R 0 -R 7r R3 e3iG eimR (cos 9 + i sin 9) 0 m 4 2 -- S R4 - a 4 0 d9 Z - a4 4 ---- 2 2 -R R4 7r ---[ 4) R m 2 a (R4 . i e d 2mRg r7 R - e - mR] A dg e 0 R - -, o as Sia eim( Residue at o< 4 o( 71 Residue at ae e ma (+ e 3ma e i ma 3 i) e--------r-----4-2-- 4 4r Residue at ae I ma (- -L+ 1 e------ 4---------A) - ma ee4 41 4 Adding residues ma e 42 2 mama F e __2 + e 2 Sx3 ei MX _- x4 - 4 dx x + a ma 2e cos 2 = Tie ma J cos ma JE ma -- ma e , 10 3 co X x s cos mx ---------cX S x4 + a44 °on +i ma x s3 sin mx -- -dx+i -- = x@ x4 ++aa4 o x 3 cos mx dx dx + a0 -x eJ 1dx e7e4 = x +a ma sina4mxdx 0o0 x3x-4----dx =+ 2 cos dIT X2 dx o00 x 4 + 5x2 + 4 0 Consider z2 dz z4 + 5a2 z z + 2 = - 1 = + 1 - 4 or or G:4z 3 +2 10 residue at pole 4 ----------5+•125- 16 2 zA IoCI + 21 -------------40<, 3 + 106K 4 c\2 + 10 =-- residue at S ma -- " + 1 = ----2 2i 4 (-4) + 10 = 6 =- i 3 Since the exponent of the numerator is 2 less than that of the biggest exponent in denominator 2 the ;• = 0as (i.e., (i.e., z0 as R-'>co) R-> i \ o ma=i7 COS x 2 dx --- 03ox 4 + 5x2+ 4 - 2 = i i i since we have an even function - x 2 dx x 4 + 5x 2 + 4 2 So 0 00 x4 0 If M > 0 i 27Fi 7/- -i 6 x2 dx 67 5x 2 6 + + 4 evaluate soo cos mx dx 0 1 + x2 + x4 consider eimz dz 2 1 + + z 1 + z2 + z 6 = z z 4 2 1 + z + z wherever z = cos (0 + 2k-T) 6 + i *"* z->9 e eim z z + 4z 3a 0 , 4, 5) i 2( u , e 3 above i 2TT i e + 00 == (k = 0, 1, 2, 6 has poles at the x axis ) the residue at Lim z >e-3 = sin (0+-2k6f) eimZ + - z4 61 - 1 .. . - 12 - 1 the residue at + eimz Lim 2i 7E 2 3 2z + 4z e 3 12 im ( + i 2 e -- 22 ---- - - t)-----2 3 (1 + e 2e ) ) ------e 7_E 2----- 2------41 + 2i 2 e + i -- im (- e (1 + 2 e Li -M45 2 -i 3) -m 2 + ------- e--- 2[ e ----2(1 + 7 1)[1+2(+ 1 +47 i))] 2(1 + 2 1)2[1+g(12 2 2 2 2 2 e------------ e 2 ----------- -m 27 -m 2 1 -----4 ] 1))4] 2 ei 2 ------- -- -- + -- - ------------- (1 + JE i) zJ1 i e - i4 -m42/ e e 2 2e T (1 - i JE) + e e (i + 1 -i L (1 + 1 2 (1 - i JZ) (1 + 1 J4) S_-2 - 1 --3 i4X (- 1 + JE 1)(- JZ) -m + 2e Se 4t (i --- 2 cos ( JE, i 21 on the circle S coo = R ei dz = R i eig dG imx Se-m R sin 2+---- 4 dx + 0 2 9 z e im R cos 9 e +dQ ---------- Ri e 2 2 1 + R e g + R e4 1 i2 e i JE 2 cos (7L - ) 3 2 13 0--Owhen R -- co the second -m -co 1 Cos C + x2 x dx + i s x4 + sin mx -00oo Scos mx 0+x-+"0 i + x + S and e sin mx dx 1 + x 2+ xi 0 = 0 0 z-a 4 1 a-i x-- dx i - x PS cos (I -- ) 3 2 2 -m cos- ( - ) 2 3 2 jj dx x4 dx = 27Te +x2+x4 a-1 3 ALz - 1 dz z = r ei dz = r i ei Lim r --> 0 C0 9 ra- dQ 1 ei(a-1 ) 9 r i ei r 2IT ei 9 Lim Lim R->oo R0- oo S 27 0 2 dQ - 1 ra - 9 ir r+ ra 1 - r + i 0 S d9 -- >' 0 dG -- 0 Ra-i ei(a-!)9 R ei i ei9 d@ ----------9 R Re 27r 3' ES - 0 -1 Ra R+I d- Ra R+I1 27r 14 i-£ a-1 S x - 1 1 i- _ e 2 ari dx S----- 1 x- 0 a-1 dx+ a-1 & x- dx+ 1 xa-1 -dx 1 + f., xe1 5' 0 ,r - 1) P S 0 Xa1-X -1 dx 0 1-x (e 2 aITi pS0 0 0 = L e27(a-1) 1 + 7T i (e 2 1-:x- e ra eo (a-1) i + (eaTri + e-aTri) l1 ----------------I-- xa-1 e2a7ri = 1 -1 -a7r i e a 771i - e 2i = s nos a7 sin a7 0 00 0 sinh ax -sinh sinli 7w Consider since d 71ctn a7 x eaz sinh dz 7T Z 7Tz e-- V 7Tz sinh z = e 7-r 2 we have poles when e e2 when z = i 7w Z z Z -7p =e- 1 i.e., when z = 0 and z = ki need only concern us.using above contour. 15 z = = i dy dz 1 e iay - e-R e-iy - 0 eFR eiy R+iy 1 /2e aR 0 - eR 1 2 e-R/ =- - a--- --e S dy tends to 0 as R--> 0o 2 z = -R+iy dz = i dy tends to 0 as R --> 00 --e 0 e ax a dx o0osinh 7 x sinh Tr (x +i) 2z eaz Lim 0 zr Z Lim z-> dx dx ax -- dx -rx sinh dx + S-oo 2az eaz + e e7 z - e- ea(x+i) snh co sliiTy + 0 ea(x+i) -00 S0 z -> 0 + 2 eaz 7e z + T e--rz 1 (V ± _2(_z_- ) eaz ez residue about little semicircle - -i ei a Z e = -1-7T = i 7T Combining i )a (+ eia)S oo 0 eax e sih 7 x ia) dx + (1 + eia 0 eax ----- dx + i-ieia= 0 sinh 7T x -7X 16 (1+ e ia) sinh ax 00 7rx) dx = - -2 [1 - e i a ] (sinh e-a 2eaX so d e -i - 2 [i + eia ] sinh-rx 0 ia _ia [1 - eia e a 2 2 e 2 +e 2 sin a 2 _ 1 2 dbi 9 - a cot -------2 S0 tan a 2 a 2cos 2 COS 9 - --a -bi 2Scos-= d9 9 - a2 - bi z = eiQ dz = ie dz = iz 2T- i ±9-a-bl dG dG -j. e + e d dG S ------------------ = lu 2Tr ei - ia + b + 1 dQ = i e---------------1 dQ ei-ia+b - 1 e ia-b dz z +z e b-b i a z e z- = -2 = - 27ri + 4 7 = 2 7Ti i If b / 0 eia-b is outside unit circle i ia-b eia-b is inside unit circle If b ; 0 17 Lim z e bz(z - eia b) z z--O S2T cot - a- bi dG = S- cot- 9---------rw 2ýi 2-i when b ; 0 - 27ri = when b -0 0 < a <2 Soo 0 xa-1 ------- dx 1 + x + x Consider a- 1 1+ a + 1 + z + z' dz z2 + z + 1 Z (1 + z+ Z2 z (1 + z + z9) dz 1-+J32 2- Z z z = = - 2 + -3 2 inside figure i) ] za [ (_ 21 J+ 2 (-12 + 2 )][z - 2 Lim 2 z [z 2 e 2Z I' Y4t z za ------------ z0 -0 - 1 +x+x Z (1 + z + 0 xa-0i - . = Ti lirm •. )1) 3 v- Integrated around r 2 2 dx + S 0 a-i -------1 = 0 dx dx = z2 ) + x + X2 27iResidue 27Ti.Residue 18 co S xa- 1 xa-1 ri(a-1) -dx + S 1 - x + x' 0 ---------- 2 dx 1 + x + x 0 2V ST Cos 2_Uq-a + i sin _2-ffa cos 2 + i sin 2- 3 +i 83 3 (cos (cos 82L + sing ( 3 sin 82ELE)(cos j '3 (cos 2 2-- + i sin [cos (27Fa oe 1x 0 iS 0 0a-I xa-1 ----------1 - x +a sin- a-1 S 2 - x + x2 0 1---------- 1 log (x + 1) + x2 Consider S around R = 0, sin -~ for dz z f (z) = 0 lim Z -, co j- 27)] 3 O 3 7 a------------- sin -r a 2 + 2l•(• z2 + 1 277- ?Th) 1 sin 2;_) 2-7- sin 27a 3 log 2 - x 2 dx = sin j i sin (27 ra - 27) 3 3 a dx x= - 3- cosT a dx + S-x----- 2 dx 0 1+x+x 2 0 --- a-)(cos ?27 )+ i sin (27 a 3 a- 1 X+ Xx 0 2- 3 JEa 7 2,2[) 3 19 z_•ogl +__ _ = + log (z + i) -Z z+i z+ ------ 2 z 1 + -_z +- - ~z+i z = -- - - - - -2- - - -- Residue at = i - z--i_og_ +i) z.-> 1 ((z - i) (z + 1) sgx + - ---x2 +1 J -0o r?) lim -0 di = rr log 1_o_ 2= 2i 2i -74-- A-o2--.- / / 2- / --- v - 21 20 Chapter II EXTENSION OF THE EVALUATION OF ONE DEFINITE INTEGRAL TO THAT OF MANY OTHERS The following example will show from the evaluation of one definite integral how we can obtain a great number of other integrals: a ' 0 oo a cos x + x sin -o00 x dx + ad Consider S -- iz - dz z - ai 77r eiR(cos 9 + i sin 9) Riei9 do S0 < R ei r 7r 0 9 - ai -ResinG ~ Re --------s 2R ---R + a -71-2 R+a eR R+a R+ a0 9 0 - e-R] (R+ a) 2R[1 S+ _-_22R Limr Residue at z -->ia 0 as z -ai eiz z - ai R--> oo = e -a ix oo adx -c0 xx :-ai Soo Scos ---S00 S~ca •o - sin s2 -2 Re w- = x + iiaxsin x i- at 27ri e- a dx = 27Ti e - a (cos x + i sin x) x + ail dx = 2Y-i e -a +---------------------co (x - ai)(x + at) 21 o0 x cos x-a sin x 0o () + i ------- --- = 2i 27.i x---- a---dx T e = - - - 2- - -2 - - - d - oo x +a I (2) dx - a a cos x + x sin x ------ '. a cos x+x sin 00 i-d + = 0 because odd function S00 x cos x dx - a dx i x +a So (3) 00 sin-x -a o x 2 +a co0 x+ sin x .co x = + a a > +oo+ - - 00 -a cos x + x sinx dx -+-------- 0 = 0 0 0 = a------x" + a dx a' eiz Consider 0 -0o z + ai ix e0 x+ia = dx -00 (4) (x + ia)(x- (5) x cos x + a ---------+ a 5 0 cosx + i sin x (x 0c0 -co S oo dz co sin x dx+ ---- i - ia) - c0 a - o 00 -00 x 2 + a2 + a = d x cos x2 + a 2 sin x dx x 0 - a cos x + x sin x -------------dx = 0 x2 + a -_a-cos x + x sin x dx ----------------- = = -a e-a 22 Adding (1)and (4) yo 0 (6) x sin 2 X -0 2 a + 00 x-sin- (7) x2 0 27 e- a x= dx + dx rs-- a = a2 2 Adding (3) and (6) (8) oo x +2 1 x - 0o sin + a2 dx e- a = Adding (6) and (8) oo (x-dx+-1 sin x d-a = So x -o + a2 2 eare Generalizing, (9) oo nx -co X 12 2 + dx = nTTe - a s a 2 where n is an integer 23 Chapter III SUMMATION OF A FINITE SERIES BY MEANS OF DEFINITE INTEGRALS The scheme consists of evaluating 0 --x - dx x2n + 1 by a standard method; then, knowing the answer, we compare this with the finite summations in the real and imaginary parts of an equation gotten by taking a number (n) of poles on the unit circle, where we may let these poles approach arbitrarily close by increasing n 1 p 0 p - 1a 00 x-A-IF d C 0 sufficiently. 1 +x S P-1dz z 1+ Lim z f (z) = 0 z f = 0 1? RL Lim z-Ž>O z--> (z) 0 oo S1 xP-1 +x 0 (x e 2 _ dx + So z -Lim- since OO IO 0 p-1 -- 1 + x dx + 1 + xe + z 1--(z-+ p-1 -K- z) p - 0 e2pVri 00 = = (e 2iri e ( p - 1)Ti 1 p-1 ----------1 + x dx ) p-1 dx = 27i e (p-1) ri 24 (1- e 2 p •0 0 i - - dx + x xp-i l+x and n T7r sin py e-Prri + ep•) then thn m - 27Ti epi + 2v~i (- If = 1 0 S • x --d 1 7rsin piT 1 7--+X dx= are positive integers, and m 4 n S00 x2m 2+dx 0 =y x 2n + 1 m 2nx 2 n-1dx 1-2n 2n 1AS o00 n --- ..- 2n -- S 2n 0 Since 0 dy - ) 2n -----1 +y 2m + dy 2n dy 1-2n 2n dx = Y... 2n X = y~n dy 1 m 1 2n dy dx 1+y (2m+1 = = yn x2m (2m+i1 1 2n-- ---------- dy 771 = 1+y ) 2n sin -__T ---------------2n sin 2+ 2n -1 I J S 0 m+ 2n x2mn dx x2n + I We now attack the problem in a different manner which will permit us to get the sum of a finite series by means of definite integrals. We have the advantage now of knowing the dx value of the definite integral S2m ------ dx+ 25 m and n positive integers with m 4 n oo S-- x2m ---dx 0 x2n + 1 S z2m dz -n z2 n /~I~; -A + I z2n z = 1 (Cos I+ 2kTT 2n 2L 2n + i sin .'. we have poles at angles = (k = 0, 1, 2 ... -- .---- 2t --- .. 2n' 2n 2n-1) 2n on unit circle The poles from ..... 2n 2n - o unit circle are on n 2n poles in upper half of the plane and as long as n is finite there are no poles on the x-axis. Lim S R -- o0 AB ( f (z) dz = If f (z) - 91) a b __ ( V(z) where Y (z) is a polynomial of degree n and (z) is a polynomial of degree less than n; where a and b are the coefficients of zn - 1 and zn in (z) and /-(z) respectively. Since the degree of .'. the integral of $ (z) is Z-n - 2, a = 0 and f (z) round semicircle of radius R is 0. R 2m Sdx Taking pole at 27T i 'Y Residues = -R x 2 n + 1 e 2n 2)m ---- ---- P21 2n (e p 2n = p=1n 2n- 1 r 2n e p ) ..... 2 n e p -rr 2n e 2n - 1. Substitute KLm+1112p-1i7T p=l 2n -- - - - , 3, e K2m+1j --- Lp.Z P 7 2n = 2p - 1 2+1 (2p-1)7 n 2n 2n p e 2n p=1 n (cos 2m + 2n 2n p=1 d (A) S0 x2 x2 +1I (2p - 1)M7 + i sin 2m+t 1 (2p - 1)7F) 2n 2 = 2 771 2n 2m + 1 (2p - 1)7 2n p=1 p=1 S2772n-- p=7 p=IA sin 2m + 1 2n (2p - 1)7F since we have an even function oo n 2m 2m + S dx 2n p=1 (2p - 1) 7 27 n x2m 0 x 2n + 2n sin m+ 1 (2p - 1) r 2n p=1 2n sin 2m + 1 2n 7 from comparison with the answer obtained in the first method. cos 2m +1 Since (2p - 1)7 = 2n p=l for the imaginary must vanish in equation (A) as we know from the answer obtained in the first way. Checking the results of this specific finite series: n cos (ap + b) ei(ap + b) p=1 (only real part of it) p=1 eib eiap p=1 Using geom. series n eia eia(n+1) iap p=1 e = eia e - (2m+1) e ia -1 ii -1 a S2e 2 _a i.*-12 a e e 2_- e _- 2 ia e 2 sin a 2 28 '7 cos sin ia = -e--e---- (b + a) ib (ap + b) sin -g 2 p1j 2----- sin a 2 2 sin 0 s------ -- 0 2 n 1 (2p - sin 2m-+l 2n sin 2m --+ 1 7n p=1l for ei(ap+b) sin (ap + b) p=1 (only imag. coeff.) p=l n = = eiap e b b eib ie i sin a 2 p=1 cos (b + a) 2 sin a 2 cos O - sin (r_+ 2n .17T 1 sin 2n 7-T We note the peculiarity of the finite series n cos m (2 - )7 = that is, no matter how many terms we have in the series given by n the sum is always zero (i.e., if n is a finite 29 integer no matter how large n is (with m < n) then wherever the series is broken off (at n = 5, 1000, 1010, or any such integral number) the sum is zero. Checking the result in special cases Suppose for s~implicity = M 1 For n = 2 2 + cos 977 cos 1)7F = (2p- cos p=1 - 2 1- 4 = 0 2 For n = 3 3 Z cos p=1 6 (2p- 1)I = cos- (2p- 1)7/ = cos 2 + cos 2 + cos5 = 2 0 For n = 4 4 cos 8 8 + cos -• 8 + cos 15 8 p=1 + 8 =I10 Since from the diagram and knowledge of sign of cosine being positive in the first and third quadrants and negative in the A second and fourth quadrants. 30 cos 37r --- + cos 217C -- T 8 8 cos --r + cos---15= 8 8 etc., for n > 4 - = 0 0 (continuing the same process).