By pre-multiplying the input arrays with a linear phase the fast multiplication via FFT and inverse FFT computes a negative wrapped convolution corresponding to a modulus of x^n+1.