Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier series Fourier series of a function f (t) with period T and corresponding angular frequency ω = 2π/T : ∞ a0 X f (t) = + (an cos(nωt) + bn sin(nωt)) , 2 n=1 Fourier series is a linear sum of cosine and sine functions with discrete frequencies that are integer multiples of the frequency of f (t). This gives rise to a discrete frequency spectrum given by the Fourier coefficients (=frequency amplitudes). Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Complex Fourier series Complex representation of Fourier series of a function f (t) with period T and corresponding angular frequency ω = 2π/T : f (t) = ∞ X cn e inωt , where n=−∞ (an − ibn )/2, n > 0, a0 /2 n = 0, cn = (a|n| + ib|n| )/2 n < 0 Note that the summation goes from −∞ to ∞. Now have a negative frequencies as well. Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Positive and negative frequencies Positive frequencies (n > 0): e inωt = cos(nωt) + i sin(nωt) Negative frequencies (n < 0): e −i|n|ωt = cos(|n|ωt) − i sin(|n|ωt) Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier series demonstration Applet for Fourier Series: http://www.westga.edu/~jhasbun/osp/Fourier.htm Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function The Fourier transform For non-periodic (or periodic) functions f (t), we can define the Fourier transform Z ∞ 1 Fourier transform F[f (t)] = g (ω) = √ f (t)e −iωt dt 2π −∞ Z ∞ 1 −1 inverse transform F [g (ω)] = f (t) = √ g (ω)e iωt dω, 2π −∞ where g (ω) corresponds to a continuous frequency spectrum. Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Properties of the Fourier transform Linearity Shifting Scaling Power spectrum Parseval’s theorem δ-function F[f (t) + g (t)] = F[f (t)] + F[g (t)] F[f (t − t0 )] = e −iωt0 g (ω) F−1 [g (ω − ω0 )] = e iω0 t f (t) 1 F[f (αt)] = |α| g (ω/α) 1 −1 F [g (βt)] = |β| f (t/β) F[f (t)] (F[f (t)])∗ = |F[f (t − t0 )]|2 = |g (ω)|2 cannot reconstruct f(t) from power spectrum since phase information is lost. (t −Rt0 )]|2 = |F[f (t)]|2 R ∞Note: |F[f ∞ 2 (t)| dt = −∞ |g (ω)|2 dω −∞ |f | {z } | {z } Power Power R ∞ ixt spectrum 1 δ(x) = 2π −∞ e dt δ(−x) = δ(x) Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier Transform - Symmetry properties The Fourier transform obeys certain symmetries. Consider the Fourier transform and its complex conjugate: Z ∞ 1 g (ω) = √ f (t)e −iωt dt 2π −∞ Z ∞ 1 ∗ g (ω) = √ f ∗ (t)e iωt dt 2π −∞ If f (t) is real (f ∗ (t) = f (t)), then Z ∞ 1 ∗ g (ω) = √ f (t)e −i(−ω)t dt = g (−ω) 2π −∞ Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier Transform - Symmetry properties g ∗ (ω) = g (−ω) means the real part of the transform Re(g (ω)) is even, while Im(g (ω)) is odd: g (ω) = Re(g (ω)) + iIm(g (ω)) g (−ω) = Re(g (−ω)) + iIm(g (−ω)) g ∗ (ω) = Re(g (ω)) − iIm(g (ω)) Conversely, if f (t) is purely imaginary, then Z ∞ 1 ∗ √ g (ω) = − f (t)e −i(−ω)t dt = −g (−ω) 2π −∞ Hence, g (−ω) = −g ∗ (ω), which means that Re(g (ω)) is odd, while Im(g (ω)) is even. Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier Transform - Symmetry properties Let f (t) be even: f (−t) = f (t), then can use scaling property to show: F[f (−t)] = F[f ((−1)t)] = 1 g (ω/(−1)) | − 1| = g (−ω) But, F[f (−t)] = F[f (t)] = g (ω) Therefore, g (ω) = g (−ω). The transform is even too. Similarly, can show that if f (t) is odd, then g (ω) is odd as well. Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier Transform - Symmetry properties The symmetry properties of the Fourier transform can be summarized as follows: f (t) f (t) f (t) f (t) f (t) f (t) f (t) f (t) real imaginary even odd real and even real and odd imaginary and even imaginary and odd Re(g (ω)) even and Im(g (ω)) odd Re(g (ω)) odd and Im(g (ω)) even g (ω) even g (ω) odd g (ω) real and even g (ω) imaginary and odd g (ω) imaginary and even g (ω) real and odd Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier Series vs. Fourier transform Let’s compare Fourier series, complex representation of Fourier series and Fourier transform: Consider f (t) = sin(ωt), where ω = 2π/T . What are its real Fourier coefficients? ∞ f (t) = a0 X + (an cos(nωt) + bn sin(nωt)) , 2 n=1 By inspection, an = 0, b1 = 1. bn = 0 for n 6= 1. (1) Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function More formally, substitute f (x) into the equations below (lecture 3, eqns. (15,16)) an = bn = 2 T 2 T Z T /2 f (t) cos(nωt)dt (2) f (t) sin(nωt)dt (3) −T /2 Z T /2 −T /2 Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Therefore, an = 2 T Z T /2 sin(ωt) cos(nωt) dt {z } | {z } −T /2 | odd even = 0 (5) Z bn = = (4) T /2 2 sin(ωt) sin(nωt)dt T −T /2 Z 2 1 π sin(x) sin(nx)dx , T ω −π | {z } (6) (7) πδ1n = 2 T πδ1n T 2π where x = ωt. Therefore, bn = δ1n . (8) Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Now let’s find the Fourier coeffcients of f (t) = sin(ωt) for the complex representation of the Fourier series: ∞ X f (t) = cn e inωt , (9) n=−∞ where, 1 cn = T Z T /2 f (t)e −inωt dt Substituting f (t), we obtain Z 1 T /2 cn = sin(ωt)(cos(nωt) − i sin(nωt))dt T −T /2 = 1 T (10) −T /2 Z T /2 Z T /2 sin(ωt) cos(nωt)dt −i sin(ωt) sin(nωt))dt −T /2 −T /2 | {z } =0 Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function cn i = − 2π Z −π | = π − 2i + 2i sin(x) sin(nx)) dx {z } πδ1n n=1 n = −1 Hence, c1 = −i/2 and c−1 = i/2, all other cn are zero. Note: A single harmonic (sin or cos) is represented by two Fourier coefficients in the complex Fourier series. Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Now consider the Fourier transform of f (t) = sin(Ωt). Rewrite as iΩt −iΩt f (t) = e −e : 2i Z ∞ 1 g (ω) = √ f (t)e −iωt dt 2π −∞ Z ∞ iΩt 1 e − e −iΩt −iωt = √ e dt 2i 2π −∞ = 2π √ 2i 2π Z ∞ 1 Z ∞ 1 i(Ω−ω)t −i(Ω+ω)t e dt − e dt 2π −∞ 2π −∞ | {z } | {z } √ = δ(Ω−ω) i 2π (δ(Ω + ω) − δ(Ω − ω)) 2 Note that δ(x) = δ(−x) . δ(−(Ω+ω))=δ(Ω+ω) Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Therefore, √ i 2π (δ(Ω + ω) − δ(Ω − ω)) F[sin(Ωt)] = 2 yields two δ-function peaks at ω = ±Ω with imaginary amplitudes. This is analogous to the two complex Fourier coefficients we obtained earlier. Similarly, one can obtain √ F[cos(Ωt)] = 2π (δ(Ω + ω) + δ(Ω − ω)) 2 Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function What is the Fourier transform of a constant function f (t) = C ? Z ∞ 1 g (ω) = √ f (t)e −iωt dt 2π −∞ Z ∞ 1 Ce −iωt dt = √ 2π −∞ Z ∞ 2πC 1 = √ e −iωt dt 2π 2π −∞ | {z } √ =δ(−ω)=δ(ω) = C 2πδ(ω) δ-function peak at ω = 0. Zero frequency mode. Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier transform of periodic functions Fourier transform can operate on non-periodic, but also periodic functions, which we can express in terms of Fourier series. Let f (t) be a periodic function with angular frequency Ω: ∞ f (t) = a0 X + (an cos(nΩt) + bn sin(nΩt)) , 2 n=1 Therefore, # ∞ a0 X + (an cos(nΩt) + bn sin(nΩt)) , F[f (t)] = F 2 " n=1 Now use linearity of transform F[f (t)] = F[a0 /2] + ∞ X n=1 (an F[cos(nΩt)] + bn F[sin(nΩt)]) Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Fourier transform of periodic functions F[f (t)] √ a0 2π = δ(ω) 2 √ ∞ 2π X an (δ(nΩ + ω) + δ(nΩ − ω)) + 2 n=1 √ ∞ i 2π X + bn (δ(nΩ + ω) − δ(nΩ − ω)) 2 n=1 Therefore, Fourier transforms of periodic functions yield a sum of δ-functions located at integer multiples of the frequency Ω. Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Gaussian peak: f (t) = e −αt 2 F[f (t)] = = = = Z ∞ 1 2 √ e −αt e −iωt dt 2π −∞ Z ∞ iω 2 iω 2 1 t+( 2α −α t 2 + iω ) −( 2α ) α √ dt e 2π −∞ Z ∞ iω 2 iω 2 1 √ e −α(t+ 2α ) e α( 2α ) dt 2π −∞ Z ∞ ω2 1 2 √ e − 4α e −αx dx 2π | −∞ {z } √ = = ω2 1 √ e − 4α 2α π/α Fourier Series and Transform - summary Fourier Transform - Symmetry properties Fourier Series and Transform - Comparison Fourier Transform example - non-periodic function Table of common Fourier transforms f (t) δ(t) constant C sin(Ωt) cos(Ωt) 2 Gaussian (α > 0): e −αt Exponential (α > 0): e −α|x| g (ω) √1 √ 2π C 2πδω √ i 2π δ(ω+Ω)−δ(ω−Ω) 2 √ δ(ω+Ω)+δ(ω−Ω) 2π 2 2 √1 e −ω /(4α) 2α q α 2 π α2 +ω 2