Closed matthewcarbone closed 1 year ago
This is for the situations where each (usually high-D) input data point has an index associated with it. For example, in DKL, we index image patches and, at each step, choose the index of a patch to measure next, even though our DKL inputs are actual images. Hence, the distance penalty must be applied to the (i, j) space of indices instead of the D image space.
I see, interesting. That could use some documentation but in that case it's not an issue. Thanks!
https://github.com/ziatdinovmax/gpax/blob/a35febaa3f0a33d276457b2ced198349f73560e1/gpax/acquisition/batch_acquisition.py#L45
Above, it looks like
X_
is either set to indexes or data. Should this reador something like it? Just trying to understand a bit better what this is doing.