Circular Convolution

Fourier Analysis · Axiom Academy

Understanding periodic convolution and its relationship to the DFT 1. Definition of Circular Convolution Circular convolution (denoted by ⊛) is defined for two N-point sequences x[n] and y[n]. The key difference from linear convolution is the modulo operation, which creates a periodic, "wrap-around" effect. The modulo operation (n - m) mod N ensures all indices stay within [0, N-1], creating a circular indexing pattern. 2. Why "Circular" - The Wraparound Effect The term "circular" comes from visualizing the sequences arranged on a circle. When indices exceed N-1 or go below 0, they wrap around to the other side of the sequence. For example, if N=8, then indices wrap as: ...,-1,0,1,2,3,4,5,6,7,0,1,2,... 3. The DFT Convolution Theorem The circular convolution theorem is one of the most powerful properties of the DFT. It states that circular convolution in the time domain corresponds to pointwise multiplication in the frequency domain. This means: to compute a circular convolution, we can take the DFT of both sequences, multiply them element-wise, and then take the inverse DFT. This is often much faster than direct computation! 4. Linear vs Circular Convolution - Avoiding Wraparound Linear convolution produces a result of length (N + M - 1) for sequences of length N and M. Circular convolution produces a result of length max(N, M) due to wraparound aliasing.

This is the written version of the interactive lesson above. See the full Fourier Analysis course.