s6_95-100aSOLUTION

# s6_95-100aSOLUTION - ACM95a/100a Last updated Problem Set...

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ACM95a/100a Last updated November 18, 2009 Problem Set VI Solutions Prepared by Faisal Amlani 1. (20 pts.) Use contour integration to evaluate the integrals 2π (a) (10 pts.) 0 dθ , 2 + sin(θ) π (b) (10 pts.) −π cos(nθ) dθ (|a| < 1). 1 − 2a cos(θ) + a2 For the second integral, assume n is a non-negative integer. SOLUTION : (a) Let C be the positively oriented unit circle about the origin. Deﬁning a parametrization z = eiθ , dz = i eiθ dθ, θ ∈ (0, 2π ), we can write sin θ and the diﬀerential dθ in terms of z to get 2π 0 1 dθ = 2 + sin θ C 1 dz . 2 + (z − 1/z )/(2i) iz Then by the residue theorem, 2π 0 1 dθ = 2 + sin θ = C C = C 1 dz 2 + (z − 1/z )/(2i) iz 2 dz z 2 + 4iz − 1 2 √ √ dz z+ 2+ 3 i z+ 2− 3 i z+ 2+ √ 2 3i √ z+ 2− 3 i z+ 2− = 2πi Res 2 = 2πi √ 2 3i 2π =√. 3 (b) If a = 0, we have √ 3 i , z = −2 + √ 3i π cos(nθ ) dθ = −π 0 2π for n = 1, 2, ... for n = 0. If |a| < 1, a = 0 then sin(nθ ) 1 − 2a cos θ + a2 is an even function and hence its integral vanishes over the domain. So we can add it to the integral to get π −π cos(nθ ) dθ = 1 − 2a cos θ + a2 = π −π π −π cos(nθ ) dθ + 1 − 2a cos θ + a2 einθ dθ 1 − 2a cos θ + a2 π −π sin(nθ ) dθ 1 − 2a cos θ + a2 Let C be the positively oriented unit circle about the origin. Deﬁning a parametrization z = eiθ , dz = i eiθ dθ, θ ∈ (−π . . . π ), zn dz . 1 − a(z + 1/z ) + a2 iz we can write the integrand and the diﬀerential dθ in terms of z to get π −π einθ dθ = 1 − 2a cos θ + a2 C Then by the residue theorem, π −π einθ dθ = 1 − 2a cos θ + a2 zn dz 1 − a(z + 1/z ) + a2 iz C zn dz = −i 2 2 C −az + (1 + a )z − a i zn = dz a C z 2 − (a + 1/a)z + 1 zn i dz = a C (z − a)(z − 1/a) zn i = 2πi Res ,z = a a (z − a)(z − 1/a) 2π an =− a a − 1/a 2πan = 1 − a2 Summarizing the two cases we get π −π cos(nθ ) dθ = 1 − 2a cos θ + a2 2π 2πan 1−a2 for a = 0, n = 0 otherwise 2. (30 pts.) Use contour integration to evaluate the integrals ∞ (a) (b) (10 pts.) −∞ ∞ dx , 1 + x4 x2 dx , (1 + x2 )2 (10 pts.) −∞ ∞ (c) (10 pts.) −∞ cos(x) dx. (1 + x2 ) SOLUTION : To aid with the rest of this assignment, we state the following result: Result 1 Let f (z ) be analytic except for isolated singularities, with only ﬁrst order poles on the real axis. Let CR be the semi-circle from R to −R in the upper half plane. If R→∞ lim R max |f (z )| z ∈CR =0 n then − ∞ m f (x) dx = 2πi k =1 Res (f (z ), zk ) + πi k =1 Res (f (z ), xk ) −∞ where z1 , . . . zm are the singularities of f (z ) in the upper half plane and x1 , . . . , xn are the ﬁrst order poles on the real axis. Now let CR be the semi-circle from R to −R in the lower half plane. If R→∞ lim R max |f (z )| z ∈CR =0 n then − ∞ −∞ m f (x) dx = −2πi k =1 Res (f (z ), zk ) − πi Res (f (z ), xk ) k =1 where z1 , . . . zm are the singularities of f (z ) in the lower half plane and x1 , . . . , xn are the ﬁrst order poles on the real axis. (a) Consider ∞ −∞ x4 1 dx. +1 The z = {e , e plane. Since integrand z 41 is analytic on +1 iπ/4 i3π/4 i5π/4 i7π/4 ,e ,e lim the real axis and has isolated singularities at the points }. Let CR be the semi-circle of radius R in the upper half z4 1 +1 = lim R→∞ R→∞ R max z ∈CR R R4 we can apply Result 1 to get ∞ −∞ 1 −1 = 0, x4 1 dx = 2πi +1 Res z4 1 , eiπ/4 +1 + Res z4 1 , ei3π/4 +1 . The residues are Res 1 , eiπ/4 4+1 z z − eiπ/4 = lim 4 z → eiπ/4 z + 1 1 = lim 3 z → eiπ/4 4z 1 = e−i3π/4 4 −1 − i =√ 42 and Res 1 , ei3π/4 4+1 z = 1 4( ei3π/4 )3 1 = e−iπ/4 4 1−i = √, 42 Thus the integral from the residue theorem is ∞ −∞ x4 1 −1 − i 1 − i √ +√ dx = 2πi +1 42 42 π = √. 2 x2 dx. (x2 + 1)2 (b) Consider ∞ −∞ The integrand is analytic on the real axis and has second order poles at z = ±i. Since the integrand decays suﬃciently fast at inﬁnity, lim R max z ∈CR R→∞ z2 (z 2 + 1)2 = lim R→∞ R R2 (R2 − 1)2 = 0, we can apply Result 1 to get ∞ −∞ x2 dx = 2πi Res (x2 + 1)2 z2 ,z = i . (z 2 + 1)2 The residue is d z2 (z − i)2 2 z →i dz (z + 1)2 2 z d = lim z →i dz (z + i)2 (z + i)2 2z − z 2 2(z + i) = lim z →i (z + i)4 i =− . 4 Thus the integral from the residue theorem is simply Res = lim ∞ −∞ z2 ,z = i (z 2 + 1)2 x2 π dx = . 2 + 1)2 (x 2 ∞ −∞ (c) Consider cos(x) dx. (1 + x2 ) Since sin(x) 1 + x2 is an odd function, it vanishes on the domain. We can then add a multiple of it to our integral to get ∞ −∞ cos(x) dx = 1 + x2 = ∞ −∞ ∞ −∞ cos(x) dx + i 1 + x2 eix dx. 1 + x2 ∞ −∞ sin(x) dx 1 + x2 Since e /(1 + z ) is analytic except for simple poles at z = ±i, and the integrand decays suﬃciently fast in the upper half plane, R→∞ iz 2 lim R max z ∈CR eiz 1 + z2 = lim R→∞ R we can apply Result 1 to get ∞ −∞ 1 R2 − 1 = 0, ∞ −∞ eix dx = 2πi Res 1 + x2 e−1 = 2πi 2i π = e =⇒ π cos(x) dx = . 2 1+x e eiz ,z = i (z − i)(z + i) 3. (20 pts.) For 0 < a < 1 show that ∞ (a) (b) (10 pts.) −∞ ∞ eax π dx = , ex + 1 sin(πa) cosh(ax) πa dx = π sec . cosh(x) 2 (10 pts.) −∞ [Hint: For the ﬁrst integral integrate exp(az )/(exp(z ) + 1) around a rectangle with vertices at −R, R, R + 2πi and −R + 2πi. The second integral can be obtained similarly, or deduced from the ﬁrst.] SOLUTION : (a) Consider eax dx. x −∞ e + 1 Let C be the positively-oriented rectangular contour with corners at ±R and ±R + 2πi. From the maximum modulus integral bound we see that the integrals on the vertical sides of the contour vanish as R → ∞: ∞ R+i2π R −R −R+i2π eaz eaR dz ≤ 2π R → 0 as R → ∞ ez + 1 e −1 e−aR eaz dz ≤ 2π → 0 as R → ∞. ez + 1 1 − e−R Thus in the limit R → ∞, the integral on the rectangular contour is the sum of the integrals along the top and bottom sides: eaz dz = ez + 1 ∞ −∞ C eax dx + ex + 1 ∞ −∞ −∞ ∞ ax ea(x+2πi) dx ex+2πi + 1 dx. = (1 − e−2aπi ) e ex +1 Since the only singularity of the integrand inside the contour is a ﬁrst order pole at z = iπ , we can use the residue theorem to get eaz dz = 2πi Res ez + 1 eaz , iπ ez + 1 (z − πi) eaz = 2πi lim z →iπ ez + 1 a(z − πi) eaz + eaz = 2πi lim z →iπ ez aπi = −2πi e . C Equating the two results gives ∞ (1 − e−2aπi ) −∞ ∞ −∞ eax dx ex + 1 eax dx ex + 1 = =⇒ = = −2πi eaπi eaπi 2πi − e−aπi π sin(πa) (b) Consider ∞ −∞ cosh(ax) dx. cosh x From a change of variables x → 2x in the result of part (a), ∞ −∞ ∞ −∞ e2ax 2 dx e2x + 1 2e dx ex + e−x e dx cosh x (2a−1)x (2a−1)x = =⇒ = =⇒ = π sin(πa) π sin(πa) π . sin(πa) ∞ −∞ So if we let b = 2a − 1, we have that ∞ −∞ π π ebx dx = = cosh x sin(π (b + 1)/2) cos(πb/2) for − 1 < b < 1. Since cosine is even, we also have that ∞ −∞ e−bx π dx = cosh x cos(πb/2) for − 1 < b < 1. Thus ∞ −∞ cosh(ax) dx = cosh x = = = = eax + e−ax dx 2 cosh x −∞ 1 ∞ e−ax 1 ∞ eax dx + dx 2 −∞ cosh x 2 −∞ cosh x 1 π 1 π + 2 cos(πa/2) 2 cos(πa/2) π cos(πa/2) πa π sec . 2 ∞ 4. (30 pts.) Obtain the formulae ∞ (a) (b) (15 pts.) −∞ ∞ 1 − cos(x) dx = π , x2 sin(πx) dx = π . x(1 − x2 ) (15 pts.) 0 SOLUTION : (a) This is integrable since (1 − cos x)/x2 has a removable singularity at x = 0 and decays 1 like x2 at inﬁnity. Since (sin x)/x2 is a odd function with a simple pole at x = 0, the principal value of its integral vanishes so we can subtract a multiple of it to get ∞ −∞ 1 − cos x dx = x2 ∞ 1 − cos x sin x dx − i − dx 2 2 x −∞ −∞ x ∞ 1 − cos x − i sin x dx =− x2 −∞ ∞ 1 − eix =− dx. x2 −∞ ∞ Let CR be the semi-circle of radius R in the upper half plane. Since R→∞ lim R max z ∈CR 1 − eiz z2 = lim R R→∞ 2 =0 R2 the integral along CR vanishes as R → ∞, CR 1 − eiz dz → 0 as R → ∞. z2 1 − eiz ,z = 0 z2 1 − eiz πi lim z →0 z −i eiz πi lim z →0 1 −i2 π π πi Res π. Then applying Result 1 gives ∞ − −∞ 1 − eix dx x2 = = = = = =⇒ = ∞ −∞ 1 − cos x dx x2 (b) Consider ∞ 0 Again, this is integrable since the integrand has removable singularities at the points x = 0, ±1. Then, since the integrand is also clearly even, we have that ∞ 0 sin(πx) dx. x(1 − x2 ) 1 sin(πx) dx = x(1 − x2 ) 2 ∞ −∞ sin(πx) dx. x(1 − x2 ) Since cos(πx) is an odd function with ﬁrst order (removable) poles at x = 0, ±1, it’s x(1 − x2 ) integral vanishes over (−∞, ∞) so we can add it as follows: ∞ 0 1 ∞ sin(πx) sin(πx) dx = dx x(1 − x2 ) 2 −∞ x(1 − x2 ) ∞ i sin(πx) = dx 2i −∞ x(1 − x2 ) ∞ 1 ∞ cos(πx) sin(πx) i dx + − dx = 2i −∞ x(1 − x2 ) 2i −∞ x(1 − x2 ) 1 ∞ cos(πx) + i sin(πx) = − dx 2i −∞ x(1 − x2 ) eiπx 1∞ − dx = 2i −∞ x(1 − x2 ) i∞ eiπx =−− dx. 2 −∞ x(1 − x2 ) eiπz z (1 − z 2 ) 1 R (R 2 − 1) Let CR be the semi-circle of radius R in the upper half plane. Since lim R max z ∈CR R→∞ = lim R R→∞ =0 the integral along CR vanishes as R → ∞, i.e. CR eiπz dz → 0 as R → ∞, z (1 − z 2 ) so we can apply Result 1 to get eiπx i∞ −i −− dx = πi 2 −∞ x(1 − x2 ) 2 Res + Res eiz , z = 0 + Res z (1 − z 2 ) eiz ,z = 1 z (1 − z 2 ) π 2 π = 2 =π = ∞ 0 eiz , z = −1 z (1 − z 2 ) eiπz eiπz eiπz + lim − lim lim z →0 z (1 − z ) z →0 z (1 + z ) z →0 1 − z 2 −1 −1 + 1− 2 −2 sin(πx) dx = π. x(1 − x2 ) =⇒ ...
View Full Document

## This note was uploaded on 01/14/2010 for the course ACM 1 taught by Professor Prof during the Fall '09 term at Caltech.

Ask a homework question - tutors are online