Hi, I notice that in your paper these is a formular stated like this :$A_{ij}=exp(-\frac{2D_{ij}^2}{\Delta})$. But in your repo, the corresponding code is np.exp(-dis_vec[0, node_idx] ** 2 / (m_prob * avg_dis) ** 2) , seems like computing $A_{ij}=exp(-\frac{D_{ij}^4}{\Delta})$. Could please explain it's this or mistake or i misundersood your code ?
Hi, I notice that in your paper these is a formular stated like this :$A_{ij}=exp(-\frac{2D_{ij}^2}{\Delta})$. But in your repo, the corresponding code is$A_{ij}=exp(-\frac{D_{ij}^4}{\Delta})$ . Could please explain it's this or mistake or i misundersood your code ?
np.exp(-dis_vec[0, node_idx] ** 2 / (m_prob * avg_dis) ** 2), seems like computing