@@ -153,7 +153,7 @@ while heap_max(intervals)[0] > $\epsilon$:
In the above, `loss` only gets the data associated with a single interval;
in order to support loss functions that rely on data from neighboring intervals we would need to maintain a separate datastructure that encodes the neighborhood information.
For example, if `data` were a binary tree storing `(x, f(x))` then we could query neighboring points in $\mathcal{O}(log N)$ time.
For example, if `data` were a binary tree storing `(x, f(x))` then we could query neighboring points in $\mathcal{O}(\log N)$ time.
#### As an example, the interpoint distance is a good loss function in one dimension.
An example of such a loss function for a one-dimensional function is the interpoint distance.