* New array mapping implementation using sparse matrices which is ~100x faster and allows for very large input values
0.5
* Default is now 64 bit integers for indexing; compatible with `bw_processing` 1.0
0.4.3
* Allow `MappedMatrixDict` to take extra `kwargs` for compatiblity with custom matrix classes
0.4.2
* Fix diagonal `MappedMatrixDict` - `col_mapper` is not required * Unroll tuples when multiplying `SparseMatrixDict` * Switch from `fs` to `fsspec` (following `bw_processing`)
0.4.1
* Allow passing `MappedMatrix` subclass to `MappedMatrixDict`
0.4
* Add `SparseMatrixDict` * Defined matrix multiplication for `MappedMatrixDict`