Y lack lags

For X = (X_t, …, X_t-n) and similarly Y, transfer entropy is H(Xt|X) - H(Xt|X, Y).

The make space code includes lags of X, but no lags of Y.

This means when doing the array manipulations with numpy and KdTree, each history windows of X is only pared against one value of Yt. The histories of Y is never taken into account of.