Is this implementation of calculating L2 distance matrix efficient?

1

I need to calculate L2 distance between all elements in a mini-batch. The following code is find on the web. The question is that is this implementation efficient? Is there any API to do such operation?

def distanceMatrix(mtx1, mtx2):
    """ 
    mtx1 is an autograd.Variable with shape of (n,d)
    mtx1 is an autograd.Variable with shape of (n,d)
    return a nxn distance matrix dist
    dist[i,j] represent the L2 distance between mtx1[i] and mtx2[j]
    """
    m = mtx1.size(0)
    p = mtx1.size(1)
    mmtx1 = torch.stack([mtx1]*m)
    mmtx2 = torch.stack([mtx2]*m).transpose(0, 1)
    dist = torch.sum((mmtx1 - mmtx2)**2, 2).squeeze()
    return dist

Thanks~

2

See here Efficient Distance Matrix Computation. Approach B is the most efficient if its numerical accuracy is enough for you.

3

Thanks very much.
Your suggestion exactly solved my problem. I will have a try.