I wanted to interpret array indexing in the following code snippet. What does State[t,Con] mean, where Con itself is an array?
for t in range(T): # 0 .. T-1
State[t+1] = Bool[:, sum(Pow * State[t,Con],1)].diagonal()
And Con is given as below (where N>K):
Con = apply_along_axis(random.permutation, 1, tile(range(N), (N,1) ))[:, 0:K]
Con is a (N,K) array of integers.
State presumably is (T,N) array.
State[t,Con] will be a (N,K) array of values selected from the t row of State. Since Con has repeats, some values of the State row will be repeated.
`Bool[:, sum(Pow * State[t,Con],1)].diagonal()`
It then does a element by element multiplication with Pow (also a (N,K) array, or something compatible). Then sum over the last axis (columns), giving a N,) array (N element vector). Then select those columns from array Bool (a (N,N) array?). Finally get the main diagonal - again N values.
The last step is to assign those values to the t+1 row of State.
