Hi! I am new to convolutions. I am looking to implement a convolution with wrapped/toroidal boundaries (i.e. for CircularArrays).

```
# matrix = N x M matrix
# smaller_matrix = some sort of kernel
convolve2d(matrix, smaller_matrix, mode='same', boundary='wrap')
```

In other words, a smoothing filter that acts periodically. I know

```
DSP.conv(vector, vector, matrix)
```

achieves something similar, but I don’t know how to implement wrapped boundaries.

Thanks!