Parseval`s Theorem (complex version): Let ∈ 2[βˆ’ , ]. Then 2 =

Parseval’s Theorem (complex version): Let 𝑓 ∈ 𝐿2 [βˆ’πœ‹, πœ‹]. Then 𝑓 𝑑
πœ‹
βˆ’πœ‹
𝑓(𝑑) 2 𝑑𝑑 = 2πœ‹
∞
𝑛=βˆ’βˆž
π‘Žπ‘› 2 , where π‘Žπ‘› =
Proof: We will instead prove that 𝑓, 𝑔 = 2πœ‹
πœ‹
βˆ’πœ‹
2πœ‹
1
∞
𝑛=βˆ’βˆž π‘Žπ‘› π‘π‘š
2
=
𝑓(𝑑)𝑒 βˆ’π‘–π‘›π‘‘ 𝑑𝑑.
, where 𝑓 =
∞
𝑖𝑛𝑑
,
𝑛=βˆ’βˆž π‘Žπ‘› 𝑒
π‘–π‘šπ‘‘
𝑖𝑛𝑑
π‘–π‘šπ‘‘
𝑔= ∞
, 𝑓𝑁 = 𝑁
, and 𝑔𝑁 = 𝑁
. We will examine the case where
π‘š =βˆ’βˆž π‘π‘š 𝑒
𝑛=βˆ’π‘ π‘Žπ‘› 𝑒
π‘š =βˆ’π‘ π‘Žπ‘š 𝑒
𝑁
𝑓 = 𝑔. First, we need to show that 𝑓𝑁 , 𝑔𝑁 = 2πœ‹ 𝑛=βˆ’π‘ π‘Žπ‘› π‘π‘š . Now, 𝑓𝑁 , 𝑔𝑁
𝑖𝑛𝑑
π‘–π‘šπ‘‘
𝑖𝑛𝑑
π‘–π‘šπ‘‘
= 𝑁
, 𝑁
= 𝑁
, 𝑁
by linearity on the first term. By
𝑛=βˆ’π‘ π‘Žπ‘› 𝑒
π‘š =βˆ’π‘ π‘π‘š 𝑒
𝑛=βˆ’π‘ π‘Žπ‘› 𝑒
π‘š =βˆ’π‘ π‘π‘š 𝑒
𝑁
𝑖𝑛𝑑
additivity on the second term and conjugate symmetry, we obtain 𝑛=βˆ’π‘ π‘Žπ‘› 𝑁
, 𝑒 π‘–π‘šπ‘‘ .
π‘š =βˆ’π‘ π‘π‘š 𝑒
0, 𝑛 β‰  π‘š
𝑒 𝑖𝑛𝑑 𝑒 π‘–π‘šπ‘‘
Now, since 2πœ‹ , 2πœ‹ is an orthonormal set, then 𝑒 𝑖𝑛𝑑 , 𝑒 π‘–π‘šπ‘‘ =
. Since π‘š = 𝑛, then we
2πœ‹, 𝑛 = π‘š
𝑁
𝑁
𝑁
obtain 𝑁
𝑛=βˆ’π‘ π‘Žπ‘› π‘š =βˆ’π‘ π‘π‘š 2πœ‹ = 𝑛=βˆ’π‘ π‘Žπ‘› π‘π‘š 2πœ‹. Hence, 𝑓𝑁 , 𝑔𝑁 = 2πœ‹ 𝑛=βˆ’π‘ π‘Žπ‘› π‘π‘š .
Now we must show that 𝑓𝑁 , 𝑔𝑁 β†’ 𝑓, 𝑔 as 𝑁 β†’ ∞. We will do so by showing that 𝑓𝑁 , 𝑔𝑁 βˆ’
𝑓, 𝑔 β†’ 0 as 𝑁 β†’ ∞. Now, 𝑓𝑁 , 𝑔𝑁 βˆ’ 𝑓, 𝑔 = 𝑓𝑁 , 𝑔𝑁 βˆ’ 𝑓, 𝑔𝑁 + 𝑓, 𝑔𝑁 βˆ’ 𝑓, 𝑔 =
𝑓𝑁 βˆ’ 𝑓, 𝑔𝑁 + 𝑓, 𝑔𝑁 βˆ’ 𝑔 by additivity of the first and second terms. By the Schwarz inequality, we
obtain 𝑓𝑁 βˆ’ 𝑓, 𝑔𝑁 + 𝑓, 𝑔𝑁 βˆ’ 𝑔 ≀ 𝑓𝑁 βˆ’ 𝑓 𝑔𝑁 + 𝑓 𝑔𝑁 βˆ’ 𝑔 . By convergence in the norm, we
obtain 𝑓𝑁 βˆ’ 𝑓, 𝑔𝑁 + 𝑓, 𝑔𝑁 βˆ’ 𝑔 ≀ 0 𝑔𝑁 + 𝑓 0 = 0. Hence, 𝑓𝑁 , 𝑔𝑁 βˆ’ 𝑓, 𝑔 β†’ 0 as 𝑁 β†’ ∞.
This completes the proof. ∎
Define the convolution of 𝑓 and 𝑔, denoted 𝑓 βˆ— 𝑔 𝑑 , as
𝑓(π‘₯) =
∞
𝑛=βˆ’βˆž
π‘Žπ‘› 𝑒 𝑖𝑛π‘₯ and 𝑔(π‘₯) =
∞
𝑛=βˆ’βˆž
1 Ο€
2πœ‹ βˆ’Ο€
𝑓 𝑑 𝑔(π‘₯ βˆ’ 𝑑) 𝑑𝑑. If
𝑏𝑛 𝑒 𝑖𝑛π‘₯ , then 𝑓 βˆ— 𝑔 𝑑 =
Proof: From the definition of complex Fourier series, we know that 𝑓 βˆ— 𝑔 𝑑 =
need to show that 𝑐𝑛 = π‘Žπ‘› 𝑏𝑛 . Now, 𝑐𝑛 =
1 Ο€
2Ο€ βˆ’Ο€
∞
𝑖𝑛π‘₯
.
𝑛=βˆ’βˆž π‘Žπ‘› 𝑏𝑛 𝑒
∞
𝑖𝑛π‘₯
,
𝑛=βˆ’βˆž 𝑐𝑛 𝑒
so we
𝑓 βˆ— 𝑔 𝑑 𝑒 βˆ’π‘–π‘›π‘₯ 𝑑π‘₯ =
1 Ο€
1 Ο€
1 Ο€ 1 Ο€
𝑓 𝑑 𝑔(π‘₯ βˆ’ 𝑑) 𝑑𝑑 𝑒 βˆ’π‘–π‘›π‘₯ 𝑑π‘₯ = 2Ο€ βˆ’Ο€ 2πœ‹ βˆ’Ο€ 𝑓 𝑑 𝑔(π‘₯ βˆ’ 𝑑) 𝑒 βˆ’π‘–π‘›π‘₯ 𝑑𝑑 𝑑π‘₯ =
2Ο€ βˆ’Ο€ 2πœ‹ βˆ’Ο€
Ο€
1 Ο€ 1
𝑓 𝑑 βˆ’Ο€ 𝑔(π‘₯ βˆ’ 𝑑) 𝑒 βˆ’π‘–π‘›π‘₯ 𝑑π‘₯ 𝑑𝑑 by Fubini’s Theorem. Now, let 𝑒 = π‘₯ βˆ’ 𝑑. Then 𝑑𝑒
2Ο€ βˆ’Ο€ 2πœ‹
= 𝑑𝑑,
π‘₯ = 𝑒 + 𝑑, and the limits of integration do not change. Hence, we obtain
Ο€
1 Ο€ 1
𝑓 𝑑 βˆ’Ο€ 𝑔(𝑒) 𝑒 βˆ’π‘–π‘› (𝑑+𝑒) 𝑑𝑒 𝑑𝑑
2Ο€ βˆ’Ο€ 2πœ‹
1 Ο€
1 Ο€
𝑓 𝑑 𝑒 βˆ’π‘–π‘›π‘‘ 2πœ‹ βˆ’Ο€ 𝑔(𝑒) 𝑒 βˆ’π‘–π‘›π‘’ 𝑑𝑒
2Ο€ βˆ’Ο€
This completes the proof. ∎
1
= 2Ο€
𝑑𝑑 =
Ο€ 1
Ο€
𝑓 𝑑 βˆ’Ο€ 𝑔(𝑒) 𝑒 βˆ’π‘–π‘›π‘‘ 𝑒 βˆ’π‘–π‘›π‘’ 𝑑𝑒 𝑑𝑑 =
βˆ’Ο€ 2πœ‹
1 Ο€
1 Ο€
𝑓 𝑑 𝑒 βˆ’π‘–π‘›π‘‘ 𝑏𝑛 𝑑𝑑 = 2Ο€ βˆ’Ο€ 𝑓 𝑑 𝑒 βˆ’π‘–π‘›π‘‘
2Ο€ βˆ’Ο€
𝑑𝑑 𝑏𝑛 = π‘Žπ‘› 𝑏𝑛 .
Riemann-Lebesque Theorem: Let 𝑓(π‘₯) be a piecewise-continuous function on
𝑏
π‘Ž
[π‘Ž, 𝑏]. Then limπ‘˜β†’βˆž
𝑓 𝑑 cos π‘˜π‘‘ 𝑑𝑑 = limπ‘˜β†’βˆž
𝑏
π‘Ž
𝑓(𝑑) sin π‘˜π‘‘ 𝑑𝑑 = 0.
𝑡
𝒏=𝟏
Lemma 1: Let 𝒇(𝒙) be a πŸπ…-periodic function and 𝑺𝑡 𝒙 = π’‚πŸŽ +
𝝅
𝟏
Then 𝑺𝑡 𝒙 = 𝝅
𝒇(𝒙 + 𝒖)
βˆ’π…
𝟏
+
𝟐
𝑡
π’Œ=𝟏 𝐜𝐨𝐬 π’Œπ’–
𝒅𝒖.
𝑁
π‘˜=1
Proof: By the definitions of Fourier coefficients, we have 𝑆𝑁 π‘₯ = π‘Ž0 +
1 πœ‹
2πœ‹ βˆ’πœ‹
𝑁
1 πœ‹
βˆ’πœ‹
πœ‹
π‘˜=1
𝑓(𝑑) 𝑑𝑑 +
πœ‹
1
𝑓(𝑑) 𝑑𝑑 + πœ‹
βˆ’πœ‹
𝑁
𝑓
π‘˜=1
Now, let 𝑒 = 𝑑 βˆ’ π‘₯. Then we obtain
we obtain
πœ‹
βˆ’πœ‹
𝑓(𝑑) sin π‘˜π‘‘ 𝑑𝑑 sin π‘˜π‘₯ = 2πœ‹
1
πœ‹
βˆ’πœ‹
𝑓(𝑑) 𝑑𝑑 +
𝑓 𝑑 cos π‘˜π‘‘ cos π‘˜π‘₯ + sin π‘˜π‘‘ sin π‘˜π‘₯ 𝑑𝑑 = 2πœ‹
πœ‹
βˆ’πœ‹
1
π‘₯) 𝑑𝑑 = 2πœ‹
πœ‹
1
πœ‹
1
2
𝑓 π‘₯+𝑒
βˆ’πœ‹
+
1
𝑓 π‘₯+𝑒
βˆ’πœ‹βˆ’π‘₯
𝑁
π‘˜=1 cos π‘˜π‘’
1
2
1
2
+
cos π‘˜π‘’ sin
𝑒
2
𝑁
then
sin π‘˜
π‘˜=1
1
βˆ’2 𝑒
1
cos π‘˜π‘’ sin
1
sin
2
1
1
1
𝑒
2
𝑁 + 2 𝑒 β‡’ sin
1
2
= sin 𝑁 +
𝑒
2
1
2
+
𝟏 𝟏
+
𝟐
Lemma 3: Let 𝑷𝑡 𝒖 = 𝝅
1
2
βˆ’πœ‹
𝑁
π‘˜=1 cos 0
1
=2+
𝑒
𝑁
π‘˜=1 1
βˆ’ π‘₯) 𝑑𝑑.
𝑑𝑒. Since 𝑓(π‘₯) is 2πœ‹-periodic,
1
1
2
1
𝑒
2
𝑒 βˆ’ sin
1
2
1
𝑒
2
1
= 2 sin 𝑁 + 2 𝑒 β‡’ 2 +
𝑡
π’Œ=𝟏 𝐜𝐨𝐬 π’Œπ’–
. Then
𝝅
𝑷
βˆ’π… 𝑡
1
=2+
𝐬𝐒𝐧 𝑡+𝟏 𝟐 𝒖
.
,𝒖 β‰  𝟎
𝟐 𝐬𝐒𝐧 𝒖 𝟐
1
𝑁 = 𝑁 + 2. Now, if 𝑒
1
π‘˜=1 2
=
1
, we have sin
1
𝑡 +𝟏 𝟐,𝒖 = 𝟎
=
𝑁
𝑒
= 2 sin 𝑁 + 2 𝑒 βˆ’ 2 sin
𝑁
π‘˜=1 cos π‘˜π‘’
𝑓 𝑑 cos π‘˜(𝑑 βˆ’
𝑁
cos π‘˜(𝑑
π‘˜=1
+
𝑡
π’Œ=𝟏 𝐜𝐨𝐬 π’Œπ’–
sin π‘˜π‘’ + 2 βˆ’ sin π‘˜π‘’ βˆ’ 2
𝑒
2
𝑓(𝑑) 𝑑𝑑 +
𝑑𝑒 as desired. ∎
=2+
= 2 sin 𝑁 + 2 𝑒 βˆ’ sin
𝑁
π‘˜=1
𝑁
π‘˜=1 cos π‘˜π‘’
𝑁
1
=
π‘˜=1 2
1
2
𝑓 𝑑
𝑁
π‘˜=1 cos π‘˜π‘’
+
Lemma 2: For all π’Œ ∈ ℝ+ and for all 𝒖 ∈ [βˆ’π…, 𝝅], 𝟏 𝟐 +
Proof: If 𝑒 = 0, then
1 πœ‹
βˆ’πœ‹
πœ‹
π‘˜=1
πœ‹
1
πœ‹βˆ’π‘₯
πœ‹
βˆ’πœ‹
𝑁
𝑑 cos π‘˜(𝑑 βˆ’ π‘₯) 𝑑𝑑 = πœ‹
1
πœ‹
π‘Žπ‘˜ cos π‘˜π‘₯ + π‘π‘˜ sin π‘˜π‘₯ =
1
𝑓(𝑑) cos π‘˜π‘‘ 𝑑𝑑 cos π‘˜π‘₯ + πœ‹
𝑁
1 πœ‹
βˆ’πœ‹
πœ‹
π‘˜=1
𝒂𝒏 𝐜𝐨𝐬 𝒏𝝅𝒙 + 𝒃𝒏 𝐬𝐒𝐧 𝒏𝝅𝒙 .
𝑒
2
β‰  0,
1
sin π‘˜ + 2 𝑒 βˆ’
. Since
𝑁
+
cos π‘˜π‘’ sin
π‘˜=1
𝑁
π‘˜=1 cos π‘˜π‘’
=
𝑒
2
=
sin 𝑁+1 2 𝑒
.
2 sin 𝑒 2
∎
𝒖 𝒅𝒖 = 𝟏 .
Pointwise Convergence Theorem: If 𝑓(π‘₯) is continuous and 2πœ‹-periodic, then for
every point π‘₯ where 𝑓’(π‘₯) exists, the trigonometric Fourier series, 𝐹(π‘₯), converges to
𝑓(π‘₯).
Proof: We need to show that lim 𝑆𝑁 π‘₯ = 𝑓(π‘₯). We will instead prove that lim 𝑆𝑁 π‘₯ βˆ’ 𝑓 π‘₯ = 0. By
π‘β†’βˆž
π‘β†’βˆž
πœ‹
Lemma 1, lim 𝑆𝑁 π‘₯ βˆ’ 𝑓 π‘₯ = lim βˆ’πœ‹ 𝑃𝑁 𝑒 𝑓(π‘₯ + 𝑒)𝑑𝑒 βˆ’ 𝑓 π‘₯ β‹… 1. By Lemma 3, we obtain
π‘β†’βˆž
π‘β†’βˆž
πœ‹
πœ‹
πœ‹
lim βˆ’πœ‹ 𝑃𝑁 𝑒 𝑓(π‘₯ + 𝑒)𝑑𝑒 βˆ’ 𝑓 π‘₯ βˆ’πœ‹ 𝑃𝑁 𝑒 𝑑𝑒 = lim βˆ’πœ‹ 𝑃𝑁 𝑒 𝑓(π‘₯ + 𝑒) βˆ’ 𝑓 π‘₯ 𝑑𝑒. By Lemma 2,
π‘β†’βˆž
π‘β†’βˆž
πœ‹
πœ‹
sin 𝑁+1 2 𝑒
sin 𝑁𝑒 cos 𝑒 2 +sin 𝑒 2 cos 𝑁𝑒
we obtain lim
𝑓(π‘₯
+
𝑒)
βˆ’
𝑓
π‘₯
𝑑𝑒
=
lim
𝑓(π‘₯ +
2πœ‹
sin
𝑒
2
2πœ‹ sin 𝑒 2
π‘β†’βˆž βˆ’πœ‹
π‘β†’βˆž βˆ’πœ‹
πœ‹
sin 𝑁𝑒 cos 𝑒 2
cos 𝑁𝑒
𝑒) βˆ’ 𝑓 π‘₯ 𝑑𝑒 = lim
+ 2πœ‹
𝑓(π‘₯ + 𝑒) βˆ’ 𝑓 π‘₯ 𝑑𝑒 =
2πœ‹ sin 𝑒 2
π‘β†’βˆž βˆ’πœ‹
πœ‹
lim
π‘β†’βˆž βˆ’πœ‹
𝑓(π‘₯+𝑒)βˆ’π‘“ π‘₯ sin 𝑁𝑒 cos 𝑒 2
2πœ‹ sin 𝑒 2
and 𝑕(𝑒) =
πœ‹
lim
π‘β†’βˆž βˆ’πœ‹
𝑓(π‘₯+𝑒)βˆ’π‘“ π‘₯
2πœ‹
+
𝑓(π‘₯+𝑒)βˆ’π‘“ π‘₯ cos 𝑁𝑒
2πœ‹
𝑑𝑒. Now, let 𝑔(𝑒) =
𝑓(π‘₯+𝑒)βˆ’π‘“ π‘₯ cos 𝑒 2
2πœ‹ sin 𝑒 2
. By substitution and the Riemann-Lebesque Theorem, we obtain
πœ‹
π‘β†’βˆž βˆ’πœ‹
𝑔(𝑒) sin 𝑁𝑒 + 𝑕(𝑒) cos 𝑁𝑒 𝑑𝑒 = lim
πœ‹
π‘β†’βˆž βˆ’πœ‹
𝑔(𝑒) sin 𝑁𝑒 𝑑𝑒 + lim
𝑕(𝑒) cos 𝑁𝑒 𝑑𝑒 = 0. ∎