# Find an efficient procedure to compute the discrete cosine & sin transforms using fft

• Oct 5th 2011, 06:36 PM
drewprogden
Find an efficient procedure to compute the discrete cosine & sin transforms using fft

Consider the discrete cosine and sine transforms of a real sequence of numbers
${x_j, j=0,....,2N-1}$

$C_k = \sum_{j=0}^{2N-1} x_j cos(\frac{jk\pi}{N}),$
$S_k = \sum_{j=0}^{2N-1} x_j sin(\frac{jk\pi}{N})$

Find an efficient procedure to compute $C_k, S_k$ using a Fast Fourier Transform of $\{x_j, j=0,...2N-1\}$.

Can someone give me a hint/explain to me the way to go about doing this problem? I am completely lost. Thanks!
• Oct 6th 2011, 11:56 PM
CaptainBlack
Re: Find an efficient procedure to compute the discrete cosine & sin transforms using
Quote:

Originally Posted by drewprogden

Consider the discrete cosine and sine transforms of a real sequence of numbers
${x_j, j=0,....,2N-1}$

$C_k = \sum_{j=0}^{2N-1} x_j cos(\frac{jk\pi}{N}),$
$S_k = \sum_{j=0}^{2N-1} x_j sin(\frac{jk\pi}{N})$

Find an efficient procedure to compute $C_k, S_k$ using a Fast Fourier Transform of $\{x_j, j=0,...2N-1\}$.

Can someone give me a hint/explain to me the way to go about doing this problem? I am completely lost. Thanks!

What is the real part of the DFT of a real sequence? What is the imaginary part?

CB