# Fourier2

Description
fourier
Categories
Published

View again

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
Fourier integrals and thesampling theorem Anna-Karin TornbergMathematical Models, Analysis and SimulationFall semester, 2011 Fourier Integrals Read: Strang, Section 4.3.Fourier series are convenient to describe  periodic   functions(or functions with support on a ﬁnite interval).What to do if the function is not periodic? -  Consider  f    : R → C . The  Fourier transform  ˆ f    =  F  ( f    ) is given byˆ f    ( k  ) = ￿   ∞−∞ f    ( x  ) e  − ikx  dx  ,  k   ∈ R . Here,  f    should be in  L 1 ( R ). -  Inverse transformation: f    ( x  ) = 12 π ￿   ∞−∞ ˆ f    ( k  ) e  ikx  dk  Note: Often, you will se a more symmetric version by using adi ﬀ  erent scaling. -  Theorem of Plancherel:  f    ∈ L 2 ( R ) ⇔ ˆ f    ∈ L 2 ( R ) and ￿   ∞−∞ | f    ( x  ) | 2 dx   = 12 π ￿   ∞−∞ | ˆ f    ( k  ) | 2 dk  .  Fourier Integrals: The Key Rules    df    / dx   =  ik  ˆ f    ( k  )    ￿   x −∞ f    ( ξ  ) d  ξ   = ˆ f    ( k  ) / ( ik  )    f    ( x   − d  ) =  e  − ikd  ˆ f    ( k  )  󿿿 e  icx  f    = ˆ f    ( k  − c  ) Examples of Fourier transforms: f    ( x  ) =  sin ( ω x  ) ˆ f    ( k  ) = − i  π ( δ  ( k  − ω ) − δ  ( k   +  ω )) f    ( x  ) =  cos  ( ω x  ) ˆ f    ( k  ) =  π ( δ  ( k  − ω ) +  δ  ( k   +  ω )) f    ( x  ) =  δ  ( x  ) ˆ f    ( k  ) = 1 for all  k   ∈ R . f    ( x  ) = 􏿿 1 ,  if   − L ≤ x   ≤ L ,0 ,  if   | x  |  >  L .ˆ f    ( k  ) = 2sin kLk   = 2 L sinc( kL ) , where sinc( t  ) = sin( t  ) / t   is the  Sinus cardinalis   function. Sampling of functions - The Nyquist sampling rate How many samples do we need to identify sin ω t   or cos ω t  ?Assume equidistant sampling with  step size T  . Deﬁnition:  Nyquist sampling rate:  T   =  π / ω .The Nyquist sampling rate, is exactly 2 equidistant samples over a fullperiod of the function. (period is 2 π / ω ).For a given sampling rate  T  , frequencies higher than the  Nyquist frequency   ω N   =  π / T   cannot be detected. A higher frequency harmonic ismapped to a lower frequency one. This e ﬀ  ect is called  aliasing  .Figure from Strang.  Band-limited functions –  When the continuous signal is a combination of many frequencies  ω ,the largest frequency  ω max   sets the Nyquist sampling rate at T   =  π / ω max  . –  No aliasing at any faster rate. (Do not set sampling rate equal tothe Nyquist sampling rate. Consider sin ω T   to see why.) –  A function  f    ( x  ) is called band-limited if there is a ﬁnite  W   such thatits Fourier integral transform ˆ f    ( k  ) = 0 for all  | k  | ≥ W  .The Shannon-Nyquist sampling theorem states that such a function  f    ( x  )can be recovered from the discrete samples with sampling frequency T   =  π / W  .(Note that relating to above,  W   =  ω max   +  ε ,  ε  >  0. ) The Sampling Theorem Theorem:  (Shannon-Nyquist) Assume that  f    is band-limited by  W  , i.e.,ˆ f    ( k  ) = 0 for all  | k  | ≥ W  . Let  T   =  π / W   be the Nyquist rate. Then itholds f    ( x  ) = ∞ 󟿿 n = −∞ f    ( nT  )sinc( π ( x  / T   − n ))where sinc( t  ) = sin( t  ) / t   is the  Sinus cardinalis   function.Note: The sinc function is band-limited: 󿿿 sinc( k  ) = 􏿿 1 ,  if   − π  ≤ k   ≤ π ,0 ,  elsewhere.Also note: This interpolation procedure is not used in practice. Slowdecay, summing an inﬁnite number of terms. Approximations to the sincfunction are used, introducing interpolation errors.  The Sampling Theorem: Proof  Assume for simplicity  W   =  π .By the inverse Fourier transform, f    ( x  ) = 12 π ￿   ∞−∞ ˆ f    ( k  ) e  ikx  dk   = 12 π ￿   π − π ˆ f    ( k  ) e  ikx  dk  . Deﬁne˜ f    ( k  ) = 􏿿 ˆ f    ( k  ) ,  if   − π  <  k   <  π , periodic continuation, if   | k  | ≥ π .˜ f    can be represented as a Fourier series in  k  -space,˜ f    ( k  ) = ∞ 󟿿 n = −∞ ˆ˜ f   n e  ink  , whereˆ˜ f   n  = 12 π ￿   π − π ˜ f    ( k  ) e  − ink  dk   = 12 π ￿   π − π ˆ f    ( k  ) e  − ink  dk   =  f    ( − n ) . Hence, for  − π  <  k   <  π ,ˆ f    ( k  ) = ˜ f    ( k  ) = ∞ 󟿿 n = −∞ f    ( − n ) e  ink  . The Sampling Theorem: Proof, contd. Therefore, f    ( x  ) = 12 π ￿   π − π ˆ f    ( k  ) e  ikx  dk  = 12 π ￿   π − π   ∞ 󟿿 n = −∞ f    ( − n ) e  ink   e  ikx  dk  = ∞ 󟿿 n = −∞ f    ( − n ) 12 π ￿   π − π e  ik  ( x  + n ) dk  = ∞ 󟿿 n = −∞ f    ( − n )sin π ( x   +  n ) π ( x   +  n )= ∞ 󟿿 n = −∞ f    ( n )sin π ( x   − n ) π ( x   − n )If   W   di ﬀ  erent from  π , rescale  x   by  π / W   and  k   by  W  / π  to complete theproof.

Jul 23, 2017

#### IG6.1.2IS3.3.1ApplicationGuide.pdf

Jul 23, 2017
Search
Similar documents
Tags