Where do the normalization constants w1 and w2 come from?

Hi,

I am having some trouble understanding why are we using the normalization constants w1 and w2 and why they are defined in this way.

```
   w1 = (train_adj.shape[0] * train_adj.shape[0] - train_adj.sum()) / train_adj.sum()
   w2 = train_adj.shape[0] * train_adj.shape[0] / (train_adj.shape[0] * train_adj.shape[0] - train_adj.sum()
```

Thanks a lot for your help beforehand!