Search Mathematics of the DFT
Would you like to be notified by email when Julius Orion Smith III publishes a new entry into his blog?
The length DFT is particularly simple, since the basis sinusoids are real:
The DFT sinusoid is a sampled constant signal, while is a sampled sinusoid at half the sampling rate.
Figure 6.4 illustrates the graphical relationships for the length DFT of the signal .
Analytically, we compute the DFT to be
and the corresponding projections onto the DFT sinusoids are
Note the lines of orthogonal projection illustrated in the figure. The ``time domain'' basis consists of the vectors , and the orthogonal projections onto them are simply the coordinate axis projections and . The ``frequency domain'' basis vectors are , and they provide an orthogonal basis set that is rotated degrees relative to the time-domain basis vectors. Projecting orthogonally onto them gives and , respectively. The original signal can be expressed either as the vector sum of its coordinate projections (0,...,x(i),...,0), (a time-domain representation), or as the vector sum of its projections onto the DFT sinusoids (a frequency-domain representation of the time-domain signal ). Computing the coefficients of projection is essentially ``taking the DFT,'' and constructing as the vector sum of its projections onto the DFT sinusoids amounts to ``taking the inverse DFT.''
In summary, the oblique coordinates in Fig.6.4 are interpreted as follows: