We are working on interactive visualization of multi-dimensional data
Usually O(2-7) dimensions (still dense) Currently we are using numpy implementation for multidimensional histogram filling , slicing, summing:

I am not familiar with this particular numpy function
I don’t think we have such function in pytorch at the moment, but I’m sure we will be happy to accept a PR to add this new feature.

Do you know how it is implemented? Is it only a wrapper around the 1D version, or does it need substantially different algorithms?

In case searchsorted will be difficult to implement (I did not find it in pytorch= only external https://github.com/aliutkus/torchsearchsorted), uniform binning will be sufficient.

Algorithm:

Calculate bin edges if not provided

Compute the bin number each sample falls into - assuming non uniform binning np.searchsorted

Compute the sample indices in the flattened histogram matrix np.ravel_multi_index

Compute the number of repetitions in xy and assign it to the flattened histmat. np.bincount