Scribd Logo
Upload

Welcome to Scribd!

  • Fourier Transform

    Uploaded by

    Ryandi
    4 views40 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    4 views40 pages

    Fourier Transform

    Uploaded by

    Ryandi

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 40
    Frames Introduction eo to the Fourier transform Learning outcomes When you have completed this Programme you will be able to: Convert a trigonometric Fourier series into a doubly infinite sum of complex exponentials Derive the complex Fourier series of a function that satisfies Dirichlet’s conditions Recognise the function sinc (¢) ‘© Separate a discrete complex spectrum into an amplitude spectrum and a phase spectrum ‘ State Fourier’s integral theorem in terms of complex exponentials Define and derive the Fourier transform of a function satisfying Dirichlet’s conditions Separate a continuous complex spectrum into an amplitude spectrum and a phase spectram Recognise the functions I(t) and As(t) and derive their Fourier transforms along with those of the Dirac delta and the Heaviside unit step © Recognise alternative forms of the function-transform pair © Reproduce a collection of properties of the Fourier transform Evaluate the convolution of two functions and describe its Fourier transform Derive the Fourier sine and cosine transformations. 231 232 Programme 7 Complex Fourier series introduction In the previous Programme we saw how a periodic function can be represented by an infinite sum of periodic, trigonometric harmonics. Each harmonic has a definite frequency which is an integer multiple of the fundamental frequency. A non-periodic function can be similarly represented, not as a sum but as an integral over a continuous range of frequencies. Before we do this, however, we shall convert the infinite Fourier series in terms of sines and cosines into a doubly infinite series involving complex exponentials. Complex exponentials Recall the exponential form of a complex number and its relationship to the polar form, namely z=1(cos6 + jsin8) = re” From this equation we can see that cost + jsind =e and so cos(-8) +jsin(—0) =e = cos0—jsind Using these two equations we can find the complex exponential form of the trigonometric functions as and sind Because cos0+jsind =eland cosé— jsind = so adding these two equations gives Peer 2cos0— el" +e that is coso— 2 + a and subtracting the two equations gives ; 2jsno =e? —e# ats sno EE” @) ‘These two equations permit us to develop an alternative representa- tion of a Fourier series. Introduction to the Fourier transform 233 In the previous Programme we found that the Fourier series of the piecewise continuous function f(t) with piecewise continuous derivative and where f(t +) = f(t) is given as P10) = 2+ (an cosmant +by sin man!) @ wie 9% so we 2" ¥ f(b cosngtat and by, 27 pysnmatar Now, if we substitute the right-hand sides of equations (1) and (2) into equation (3) we obtain fi chat perhat ginat —erhat 5 ECE) In the next frame we shall make some notational changes to simplify this expression 234 Programme 7 ® If we now define ¢, so that the complex conjugate of ¢y is. 2 n+ jn 5? we can write this sum as Fit)=c+ D(e™ +e") Note that we have taken mot bo = 0. There is no problem about this. There is no term a a sin Quot in the Fourier series = cor doce + Scre™" and so bo = 0 at ont _ S For notational convenience cot Sogo + Se e7mat_ we denote c bY én. This = = means that @_» =a, and bn=—bn |As n ranges from 1 to 00 so =n ranges from —1 to ~co = 3 eee p04 Soelmt ‘Notice the reversed order of 2 Summation in the fst sum Feet Combining atthe tems za Into the dbly nite sum ; mn soa -#i pal ()(Cosmoot — jsinnaot) at, That is 1 poeta In the next frame we shall look at some examples BS Example 1 ‘To find the complex Fourier series for the function 0 -T/2 0. This means that as the period increases the space between lines in the spectrum decreases so the spectrum lines ‘come closer together and in the limit merge into a continuous spectrum. Thats, for large T ‘uy = nd and as T + 00 $0 bw w where w is the continuous frequency variable. To see the effect of this (on the general form of the Fourier series we start with FOO = SO ene where wy = 2 and where c= 4" fierm*at rp! Substituting the integral form of cy into the sum gives arya ro= > Er poem roe where w is a dummy variable in place of the variable t. Now, w» = 2% and so T r= > fl 2 fade aa ae! If Tis large then uy = éw and r= SLL remade Introduction to the Fourier transform 241 In the limit as T > oo $0 nd + w, the sum becomes an integral and 6 becomes the differential du giving vo LEL Flue au eM dus = a [C ake - _fine tau eM ay 1 eft ruetan women =ef” rae ‘These two integrals form the conclusion of Fourier’s integral theorem. Next frame Fourier’s integral theorem Given function f(t) with derivative /'(t) where (@) f(t) and f'(t) are piecewise continuous in every finite interval ©) ft) i ast itogble in (2, that is [yo is finite ~ then 10 = Fe Fledeh da where Fs) = p= The discrete harmonic values muy of the periodic function are now replaced by the continuous harmonic variable w and the discrete spectra cy =|crleM™ are replaced by the continuous spectra F(.) = [F(u)leM,F(u) I referred to as the Fourier transform of f(t) and can also be written as #(/(t)). Deriving the Fourler transform of a function is then a matter of applying the second of these two integrals. The expressions f(t) and F(.) form a Fourier transform pair where f(t) can be referred to as the inverse Fourier transform of F(.). That is, f= FF). Fite at Next frame 242 Programme 7 E73) Example 3 Find the Fourier transform of 0 t<-a/2 re)= {i a/2 0 can be also expressed in the form f(t) = e~*u(t) and has ‘the Fourier transform Fw) 250 Programme 7 Bs 1 FO tari) Because Fw) -zi. e*u(te™ ar LP? terion aL ewmrar Vir to 1 V2n(a + jw) The triangle function (at+o/@ -aa AO fy “2 ot ‘The Fourier transform of A(t) is F(e) Because Feo = Fel aemar Ef sorts Lela ort =~ Gf - near Ze fa -nerar changing the variable of integration in the frst integral fom tt0 -t “+z 2 (1—A)cosutdt and integration by parts yields _ 2. (1sin2(w/2) © Von Ne al oy , ?) sine (u/2) aa Introduction to the Fourier transform 251 Alterna’ Tt should be noted that there ae a numberof alternative forms for the (28 J Fourier transform ~ each dealing with a different location for the constant 2x. Other forms are f(t) JLo diy where F(e) = f fide at or fo ip Jot day wher ut Ef reoetraswnee r= [roe ot by boring the 2x the exponent by defining =v 10) =F Foye where re) =f" feo ar We shall remain with our original form because it has the simplest exponential factor and we do not need to remember which integral has the constant in front of it and which does not. Next frame Properties of the Fourier transform ‘We now list a number of properties of the Fourier transform that are useful in their manipulation. Linearity If the Fourier transforms #(fj(t)) = F(z) and # (f(t) = Fo(w) then F (oapilt) + aafe(t)) = oF (fi(t) + a2F (f(O) = anFi(e) + anFa(w) where a; and a2 are constants. Example ‘The Fourier transform of f(t) = 2ilx(t) ~ 6A2(¢) is Flu) 252 Programme 7 Jine (w)(1— sinc (w)) Because If f(® = ta(t) then F( sinc (w) and if f(t) = Ax(t) then pgsine a) Since f(t) = 201(t) ~ 6A2(¢) then Flu) = - sine) 6 Jeane) eee eo Time shifting If #(f(t)) = F(w) then ¥(f(t — to) = eF(u) Example 1 i The Fourier transform of Ih(t) is 7-sine (x) s0, by the time shifting property, the Fourier transform of I(t ~ 5) is and of z(t +3) is Seine) and Frequency shifting If F (f(t) — Fl) then # (f(te!*) = Flow») Example If the Fourier transform of f(t) is F(w) then the transform of Fit) cost is Introduction to the Fourier transform 253 1 &) FFw+4)+Fle—4) Because Pecosat = pe ss jae) ae =F rine +5 rte = Hiroe + fide) andthe ute tan slo 4) Fo) byte nay and the frequency shifting properties Time scaling W¥(/(t) = Fle) then wen Ff (kt)) = w@) $0, for example, given f(t) = H(t) with Foutler transform F(), If Ft) is shrunk to half its width then P(w) is stretched to twice its width but shrunk to half its height. Example If F(w) is the Fourier transform of f(t) then the Fourier transform of f(-t)is FTF ——TTT TTT (3) Few) (33) Because Kr 'R(/k) = ee fikteM ae and when k= 1 then {= 11 Fe/[-1) = zal. P(-te™ at =F(—w) symmetry KF) F(w) then ¥(F(t)) = f(—w) Example The Fourier transform of f(t) = Iz(t) is Fs) Fourier transform of F(t) = esine(t) is esses Vor 254 Programme 7 & F(-w) = ~The) Because ‘The Fourier transform of F(t) is f(-w) = Ue(-u) = —Ih(u) Try one yourself. sine (t) Example “The Fourier transform of the unit constant function f() = 1 is Fi) = & ViR6(u) Because F(6(t)| =e so ¥[_] = 514), therefore #[1] = V2n5(u) iz Vin Vin] Differentiation If F(t) +0 as t+ 400 and if #(f(t)) = Flu) then Fr )= & jak (w) Because FfOl= defo = palo Fel = 0+ juFlw) In general, if f(t) 0 as t+ 00 and if #(f(t)) = F(a) then 1 F(f(0) =H(o) then # (F()) = Go)"F) ‘where the superscript (n) indicates the mth derivative. fear Example ‘The differential equation for unforced and undamped harmonic motion is of the form mf"(t) +kf(t) =O. If we take the Fourier transform of this equation we immediately find that the permitted frequencies of oscillation are Introduction to the Fourier transform 255 [+5] Because If F(w) is the Fourier transform of f(f) then taking the Fourier transform of both sides of the equation mf”(t) + f(t) = 0 gives by the differentiation property nia)? Fw) + KEW) = (ma? + KF) = 0 so if F(u) #0 then m.? =k and so the permitted frequencies are vai Next frame The Heaviside unit step function ‘The Heaviside unit step function is defined as u(t) where (38) wo={t =o If we follow the definition of the Fourier transform we find that (je dt Fu) FU) = pees — {1~ Lim fem} Fao Hi Because Lp Fw) == the dt (= Ff me = Lf? ne “Vazh © 1 = fh - [1 -Lim fe" Fags Bale} Because eM = coswt — jsinut we cannot say what happens to the exponential at t + oo. So how do we resolve the problem? Next fame 256 Programme 7 Let Fu(t Now, u(t) +u( from Frame 35, Flo) + F(-w) = V2ra(w) We now assume that F(.) consists of a combination of the Dirac delta and an arbitrary function G(e) () and so, by the scaling property, Fu( , therefore ¥[u(t)] + ¥ [a]. That is, F(w) = af(u) + G(w) so that F(u) + F(-w) = abu) + Glu) + a6(-w) + G(—) = 2ad(u) + Glu) +G(-w) since 6(—w) = 6 = VPx6(u) 0. That is, G(w) = -G(-u), Consequently #[u(0)] = Fl) (2) + Gu). Now, # [u(t] = VJew+ ce} and since u(t) =(t) vent sof Fee + cu} “i 1 Since f(a) =0, then juGla) = Ae and 30 Glu) = Tp thereby siving UU] = Foe {80s +f } ‘The next property deals with the Fourier transform of a product of functions but before we go any further we need to discuss what is meant by the convolution of two functions. then ¥(w(t)] = ¥{6(t)] = Next frame Introduction to the Fourier transform 257 Convolution The convolution of two functions f(t) and g(t) is defined as @ f(t)es(t) = £ F(x)g(t — x) dx = h(t) where the + denotes the operation of convolution. You will note that this isa function of variable t and here we have denoted it by h(t). To illustrate an interpretation of this operation, consider the two functions f(t) = wIl,(t) and g(t) = cost. Then £(3)8(—A) = alle) €08(—x) = ally (x) co8 is that part of the cosine function that lies between 4/2 and is zero elsewhere nll,o)c08x. the intl [" fot-niar= f° stya)onsds=2. Tis te area beneath the single loop of the cosine curve. We shall call this ‘value h(0) because the loop is centred on the origin. That is (0) Now, the graph of cos(n/3 ~ x) has the same shape as cos(-x) = cosx but it is shifted to the right by x/3 radians. Consequently, [(2)s(x/3 — x) = ally(x)cos(x/3 —x) is that part of the shifted cosine function that lies between 4/2 and zero elsewhere, so now FF roasenis -2) a8 = measures the amount of the shift of the cosine curve. That is (x/3) = 1. Proceeding in this manner to define values of h(t) we see that the function formed from these integrals is Fees) = J fener aps =m |. We shall call this value (7/3) because x/3 nl, cos(n/3-») ensy=1 258 Programme 7 Example To find the convolution f(t) + g(t) where sect t| 7/4 = [tance a)" = {-tan(t—x/4) + tant tant—1 “T+tant L+tanét tant Next frame The convolution theorem (a) If F(u) and G(x) are the Fourier transforms of f(t) and g(t) respectively then (a) The Fourier transform of the convolution of f(t) and g(t) is equal to the product of the individual Fourier transforms. That is Fist) s(t) = VIF(s) Gu) and 0 1 = Linn. FPG) = Feo +00) (©) The Fourier transform of the product f(tg(®) 4s equal to the convolution ofthe individual Fourier transforms. That is Fif(Os(0l = Fe FF) «Glo = VORP) These provide useful methods of finding inverse transforms F() + G(w) and so Introduction to the Fourier transform 259 Example To find the inverse transform of Fw) -—1_,-_1 _ ,__1 __ wnerea>o Bela tjay Vana + fo)” V2x(a+ jo) ‘we note that if Fi(u) then from Frame 26 — Vora hs) f(t) = Fw] = fAilt) = e*tu(t) 4 Now, because Fu) = ie)Fi) then F(t) = FO F(w)] = FFi(w)Faw)] = Feel fleniee-nvae =P enue outs “Sf. e*u(a)eu(t—x) de = EI meoue—ayax Now, u(x)u(t—x) =0 when x <0 or when t—x <0, that is when xo 1 Jaq f+ HO) 1 if00 0 ifta 2a sinc?(.oa) Because A) = 2 reosinacar = \2[fsnetar “fe 5 2 - Ve) - Fes “2) - are sine?(ua) The Fourier cosine and sine transforms are useful when f(t) is only defined fort > 0 and where an extension can be added to (0) for t < 0 that makes the extended f(t) into an even or odd function respectively. Introduction to the Fourier transform 263 Table of transforms r= { MegHBSt0 L 1 He wae re Feel a+ph The triangle ction 0 |t}>1 1 |t}<1 Line? Jy eel?) w= {2 SQ) the Fourer transform AQ) = has the Fourier transform Introduction to the Fourier transform 265 6 Alternative forms ‘There are a number of alternative forms for the Fourier transform. each dealing with a different location for the constant 2x. Other forms are 10) =F eed du wnere Fa) = 2[" fpeMat or 10) = EP Floeh du where Ma) = J" fipeMat or frie reear. Pet) = J Fle du where Fa 7 Properties of the Fourier transform Time shifting If Fw) = z 0 = Bap pee at Li If Fy(v) and Fy(z) are the Fourler transforms of fi(t) and f(t) respectively then a1F\(w) +a9F2(w) is the Fourier transform of aufi(t) + o2f2(t) where a; and az are constants. Frequency shifting If F(a) is the Fourler transform of f() then the Fourier transform of F(terH* is Flo — un). Tine ling 1 Fe) =a feet then oe were) = ef ranebat, Symmetry If F(w) is the Fourier transform of f(t) then the Fourier transform of F(t) is F(-w). Differentiation ae] _ foe Mae ten (a)"F(w) = zl. A je at and F(0) = Fel Cie oe Mat f(He™ dt then 266 Programme 7 8 Convolution The convolution of two functions f(t) and g(t) is defined as FO xa =| forete—2yax= ne ‘The convolution theorem If F(w) and G(s) are the Fourier transforms of f(t) and s(t) respectively then (@) The Fourier transform of the convolution of f(¢) and g(t) is equal to the product of the individual Fourler transforms. That is FIO + 8(0)] = Fw)Gw). (b) The Fourier transform of the product f({)s(t) is equal to the convolution of the individual Fourier transforms. That is, FIFOs) = Flv) + Gu). 9 Fourier cosine and sine transforms Given that (t)= ee” FlojeMat where =e Vir where f(t) is even then f= af Fel) coset du where Fe(e) = ran f(0) costae and where F(z) is called the Fourier casine transformation. ‘This transformation is useful when f(¢) is defined only for t= 0 and where an extension can be added to f(t) for t <0 that makes the extended f(¢) into an even function. If f(t) is odd then f= ef F,(e) sind where F,(e) = er f(@)sinutde and where F,(w) is called the Fourier sine transformation. This ‘transformation is useful when f(¢) is defined only for t > 0 and where an extension can be added to f(t) for t < 0 that makes the extended (t) into an odd function. Fe) = Fite de Introduction to the Fourier transform 267 % Can You? Checklist 7 (51) ‘Check this list before and after you try the end of Programme test. On a scale of 1 to 5 how confident are you that Frames you can: © Convert a trigonometric Fourier series into a doubly infinite sum of complex exponentials? ys OF OF OF O QO No ‘* Derive the complex Fourier series of a function that satisfies Dirichlet’s conditions? ys O O O O O No ‘# Recognise the function sinc(t)? ys O O O O O wW ‘ Separate a discrete complex spectrum into an amplitude spectram and a phase spectrum? ys O OF Of O QO No ‘* State Fourier’s integral theorem in terms of complex exponentials? Ys O OF OF O DO No ‘* Define and derive the Fourier transform of a function satisfying Dirichlet’s conditions? ys O OF DO O O Ww ‘* Separate a continuous complex spectrum into an amplitude spectrum and a phase spectrum? ww Ys O OF OF O O Wo ‘* Recognise the functions Ilp(¢) and Aq(t) and derive their Fourier transforms along with those of the Dirac delta and the Heaviside unit step? Ys O OF OF O O WN ‘* Recognise alternative forms of the function-transform pair? = vs O O O O O WB ‘® Reproduce a collection of properties of the Fourier transform? wv O O O O O wo Evaluate the convolution of two functions and describe its Fourier transform? Se Ys O O O O G wo « Derive the Fourier sine and cosine transformations? (oH) ys O 0 O O DO WN 268 — Programme 7 (g Test exercise 7 1 Find the complex Fourier series of the sawtooth wave fi)=t, O 0 and k > 0. ( Further problems 7 1._By comparing the trigonometric Fourier series of a periodic function with its complex exponential counterpart show that lel = 2 2 Prove Parseval's identity for the periodic function with period T Ff {WOrae= 3 lal? = $4de +n) ei and show that 3° =, ah + BR and by = arctan{—Pe} where cy = lene. 3 Draw the graph and find the complex Fourier series of the rectified sine wave f(t) =sinxt, O0. 13. Given that { 1 for -1

    You might also like