Real & Complex Analysis Qualifying Exam Solution, Fall 2010 Shiu-Tang Li January 1, 2013 A-1 Note that {x : f (x) > a} = ∩b∈Q,b<a {x : f (x) > b} is measurable, {x : f (x) < a} = ∩b∈Q,b>a {x : f (x) < b} is measurable, and {x : f (x) ∈ O} is measurable for every open set O in R since O may be decomposed into a countable union of open intervals. Also, F := {A : f −1 (A) is measurable } is a sigma-algebra containing all open sets in R. It means that B(R) ⊆ F . Thus, for every B ∈ B(R), f −1 (B) is measurable and we’re done. A-2 (1) kfn gn − f gkL1 ≤ kfn gn − fn gkL1 + kfn g − f gkL1 ≤ kfn kLp kgn − gkLq + kgkLq kfn − f kLp ≤ M1 kgn − gkLq + M2 kfn − f kLp which goes to 0 as n → ∞. (2) Since gn → g in L∞ , gn is bounded by M in L∞ -norm, and when arbitrary > 0 is chosen there is some N > 0 so that kgn − gkL∞ < , 1 kfn gn − f gkL1 ≤ kfn gn − fn gkL1 + kfn g − f gkL1 ≤ kfn (gn − g)kL1 + k(fn − f )gkL1 ≤ kfn kL1 + kM (fn − f )kL1 ≤ M 0 + M k(fn − f )kL1 and the result follows from arbitrariness of . A-3 Let x0n : y 7→ hxn , yi , H → C be a sequence of elements in H ∗ . Since supn |x0n (y)| = supn | hxn , yi | ≤ My < ∞ for each y ∈ H, by uniform boundedness principle we have Dkx0n k is Ebounded by M in n, Therefore, M ≥ sup{y∈H,kyk=1} | hxn , yi | ≥ | xn , kxxnn k | = kxn k for every n. A-4 If x ∈ LP ∩ Lr , then xχx>1 ∈ Lp ⊆ Lq and xχx≤1 ∈ Lr ⊆ Lq . By Minkowski’s inequlity, kxkLq ≤ kxχx>1 kLq + kxχx≤1 kLq , giving us the result. A-5 Define d := inf{|x − z|, z ∈ M }. Pick a sequence {yn } ⊆ M so that |x − yn | < d + n1 . Apply parallelogram’s law to 12 (x − yn ) and 12 (x − ym ), we m 2 k + 14 kyn − ym k2 . have 12 (kx − yn k2 + kx − ym k2 ) = kx − yn +y 2 Thus, 41 kyn −ym k2 ≤ 12 (2d2 + 2d + 2d + n12 + m12 )−d2 , meaning {yn } is Cauchy n m and hence yn → y ∈ M . Therefore, d ≤ kx − yk ≤ kx − yn k + ky − yn k → d as n → ∞, which shows that y is closest to x than any other element in M is. If both y, y 0 minimize the distance to x, then by parallelogram’s law applied to x − y, x − y 0 , we have 4d2 = k2x − y − y 02 k2 + ky − y 0 k2 = 02 4kx − y+y k2 + ky − y 0 k2 ≥ 4d2 + ky − y 0 k2 , forcing y = y 0 . 2 2 B-6 To solve 1 1 − cos(a + ib) = 1 − (ei(a+ib) + ei(a−ib) ) 2 1 = 1 − (e−b (cos(a) + i sin(a)) + eb (cos(a) + i sin(a))) = 0, 2 first we observe that the imaginary part of both sides are zero, so sin(a) = 0. If cos(a) = −1, then we have 1 + 12 (e−b + eb ) = 0, which is impossible. If cos(a) = 1, then we have 2 = eb + e−b , so b = 0. Therefore the zeros for 1 − cos(z) are {2kπ : k ∈ Z}. B-7 (i) The only isolated singularity is −1. Let z = −1 + bi, we have | sin( z −1 + bi i )| = | sin( )| = | sin 1 + | z+1 bi b i i 1 = | (ei(1+ b ) − e−i(1+ b ) )| 2i −1 1 1 = | (ei+ b − e−i+ b )| → ∞ 2i as b → 0 + . However, if we let z = −1 + a, | sin( z )| = | sin(1 − 1/a)| ≤ 1 for every a 6= 1. z+1 Thus 0 is neither a pole or a removable singularity ⇒ it must be an essential singularity. (ii)sin(z) = z ∈ C. Let z = P∞ n=0 w w+1 = f (n) (1) (z − 1)n = n! 1 1 − 1+w , we have ∞ (n) n f (1) (1 n=0 (−1) n! P∞ − z)n for every −∞ X X w f (n) (1) f (−n) (1) sin( )= (−1)n (w + 1)−n = (−1)n (w + 1)n . w+1 n! (−n)! n=0 n=0 (iii) The coefficient of 1 1+w is −f 0 (1) = − cos( 21 ) · 3 z+1−z | (1+z)2 z=1 = − 14 cos( 21 ). B-8 See B-10 of Spring 2011. B-9 See B-6 of Fall 2007. B-10 Rouche’s theorem states that if f and g are holomorphic on a closed disc D with circle C as its boundary, and |f (z)| > |g(z)| on C, then f and f + g has the same number of zeros inside C. Take f (z) = 8, g(z) = 2z 5 − z 3 + 3z 2 − z, we find that |f | > |g| on {z : |z| = 1}, thus f + g = 2z 5 − z 3 + 3z 2 − z + 8 has the same zeros as f on {z : |z| < 1}, namely no zeros. Since |2z 5 − z 3 + 3z 2 − z| = 6 8, f (z) + g(z) 6= 0 when |z| = 1. By the fundamental theorem of algebra, f (z) + g(z) has 5 zeros on C, thus it also has 5 zeros on {z : |z| > 1}. 4