LECTURE 12: BESSEL’S INEQUALITY AND PARSEVAL’S THEOREM MINGFENG ZHAO July 24, 2015 The L2 space and Bessel’s inequality Definition 1. The space L2 ([−L, L]) is the set of all functions f (x) on [α, β] such that Z L |f (x)|2 dx < ∞. −L For any f ∈ L2 ([−L, L]), define E(f ) = 1 L Z L |f (x)|2 dx. −L Remark 1. Let f (x) be a bounded function on [−L, L], then f (x) ∈ L2 ([−L, L]). For f ∈ L2 ([−L, L]), we can define the Fourier series of f by: ∞ nπ i nπ a0 X h + x + bn sin x , an cos 2 L L n=1 where (1) an = (2) bn = 1 L Z L nπ x dx, L −L Z nπ 1 L x dx, f (x) sin L −L L f (x) cos ∀n ≥ 0, ∀n ≥ 1. For any N ≥ 1, let N SN (x) = nπ nπ i a0 X h an cos + x + bn sin x . 2 L L n=1 Theorem 1 (Bessel’s inequality). Let f ∈ L2 ([−L, L]), then N a0 X 2 + [an + b2n ] ≤ E(f ), 2 n=1 ∀n ≥ 1. In particular, we have ∞ a20 X 2 + [an + b2n ] ≤ E(f ) < ∞. 2 n=1 1 2 MINGFENG ZHAO Proof. In fact, for any N ≥ 1, we have E(f − SN ) = = L 1 L Z 1 L "Z |f (x) − SN (x)|2 dx −L L −L |f (x)|2 dx − 2 Z L Z f (x)SN (x) dx + −L # L |SN (x)|2 dx −L ≥ 0. Then " Z # Z L L 1 2 2 f (x)SN (x) dx − |SN (x)| dx ≤ E(f ). L −L −L Notice that Z L f (x)SN (x) = −L = = = Z # N nπ nπ i a0 X h + an cos x + bn sin x dx f (x) 2 L L −L n=1 Z L |SN (x)|2 dx = −L = = " L Z L Z Z L N N nπ nπ X X a0 L bn an x dx + f (x) sin x dx f (x) dx + f (x) cos 2 −L L L −L −L n=1 n=1 # " N N a20 X 2 X 2 + an + bn By (1) and (2) L 2 n=1 n=1 " # N a20 X 2 2 L + [an + bn ] 2 n=1 Z L N h nπ i2 nπ a0 X x + bn sin x dx an cos + 2 L L −L n=1 # " # Z L" N N N N nπ X nπ mπ X mπ X a0 a0 X + an cos x + bn sin x · + am cos x + bm sin x dx 2 L L 2 L L −L n=1 n=1 m=1 m=1 Z L" 2 N N mπ a X mπ a0 a0 X 0 + am cos x + bm sin x 4 2 m=1 L 2 m=1 L −L + N N X N nπ mπ nπ X a0 X an cos x + an am cos x cos x 2 n=1 L L L n=1 m=1 N N X N N X N nπ X nπ nπ X nπ mπ a0 X + bn sin x + bn am sin x cos x + bn bm sin x sin x 2 n=1 L L L L L n=1 m=1 n=1 m=1 = a20 4 Z L −L Z L Z L N N mπ mπ a0 X a0 X dx + am cos x dx + bm sin x dx 2 m=1 L 2 m=1 L −L −L Z L Z L N N X N nπ nπ mπ X a0 X cos x dx + an am cos x cos x dx an 2 n=1 L L L −L −L n=1 m=1 # dx LECTURE 12: BESSEL’S INEQUALITY AND PARSEVAL’S THEOREM 3 Z L Z L N N X N nπ nπ nπ X a0 X + bn sin x dx + x cos x dx bn am sin 2 n=1 L L L −L −L n=1 m=1 + N X N X n=1 m=1 Z L bn bm sin −L mπ nπ x sin x dx L L N N X X a20 ·L+0+0+0+ a2n · L + 0 + 0 + b2n · L 2 n=1 n=1 " # N a2 X 2 = L· 0 + [an + b2n ] . 2 n=1 = So we get N a20 X 2 + [an + b2n ] ≤ E(f ). 2 n=1 Parseval’s Theorem Theorem 2 (Parseval’s identity). Let f ∈ L2 ([−L, L]), then ∞ a20 X 2 + [an + b2n ] = E(f ). 2 n=1 Moreover, we have Z L |f (x) − SN (x)|2 dx = 0. lim N →∞ −L Proof. It’s a little technical to prove the Parseval’s identity, so here we assume it. By the computations in the proof of Theorem 1, then 1 L Z L 2 |f (x) − SN (x)| dx = −L = = # N a20 X 2 2 |f (x)| dx − + [an + bn ] 2 −L n=1 " # ∞ N a20 X 2 a20 X 2 2 2 + + [an + bn ] − [an + bn ] 2 2 n=1 n=1 1 L Z " L 2 ∞ X [a2n + b2n ] n=N +1 → 0, as N → ∞, By Theorem 1. 4 MINGFENG ZHAO Theorem 3 (Parseval’s identity for odd functions). Let f ∈ L2 ([0, L]) and f (x) = ∞ X bn sin n=1 2 L ∞ X L Z |f (x)|2 dx = 0 nπ x in 0 < x < L, then L b2n . n=1 f (x), if 0 < x < L, Proof. Consider the odd extension of f : fo (x) = , since f ∈ L2 ([0, L]), then f0 ∈ −f (−x), if −L < x < 0. L2 ([−L, L]), and 1 L Z L |fo (x)|2 dx 2 L = −L fo (x) = L Z |f (x)|2 dx 0 ∞ X bn (x) sin n=1 nπ x , L x ∈ (−L, 0) [ (0, L). By Theorem 2, then ∞ X L 1 L Z 2 L Z |f0 (x)|2 dx = −L b2n . n=1 Therefore, we have L |f (x)|2 dx = 0 ∞ X b2n . n=1 Example 1. For f (x) = x in (0, 2), then the Fourier sine series of f is: ∞ nπ 4 X (−1)n+1 x= sin x , π n=1 n 2 ∀x ∈ (0, 2), that is, bn = 4 (−1)n+1 · , π n ∀n ≥ 1. Notice that 2 2 Z 2 2 Z |f (x)| dx = 0 2 x2 dx = 0 8 . 3 By Parsavel’s identity, then ∞ X n=1 b2n = Z ∞ 16 X 1 8 2 2 · = = |f (x)|2 dx. π 2 n=1 n2 3 2 0 So we get ∞ X π2 1 = . n2 6 n=1 LECTURE 12: BESSEL’S INEQUALITY AND PARSEVAL’S THEOREM 5 Example 2. For f (x) = x2 in (−π, π), then the Fourier series of f is: x= ∞ X π2 (−1)n cos(nx), +4 3 n2 n=1 ∀x ∈ (−π, π), that is, 2 π , if n = 0, 3 an = n 4(−1) , if n ≥ 1. n2 Notice that 1 π Z π 1 |f (x)| dx = π −π 2 Z π 2 x dx = π −π 4 Z π 0 x4 dx = 2 π5 2π 4 · = . π 5 5 By Parsavel’s identity, then ∞ ∞ X a20 X 2 π4 1 2π 4 + an = + 16 = . 4 2 9 n 5 n=1 n=1 So we get ∞ X 1 π4 = . 4 n 90 n=1 Department of Mathematics, The University of British Columbia, Room 121, 1984 Mathematics Road, Vancouver, B.C. Canada V6T 1Z2 E-mail address: mingfeng@math.ubc.ca