mapToParent
Maps child (view) coordinates to parent tensor coordinates.
This method performs the core coordinate transformation, converting indices from the view's coordinate system to the parent tensor's coordinate system. The transformation must handle:
Dimensional mapping (view dimensions to parent dimensions)
Offset calculations (handling slice starting positions)
Stride transformations (handling step sizes and memory layout)
Coordinate Validation
Implementations should validate that:
childIndices.size matches the expected view dimensionality
All child indices are within valid bounds for the view
The resulting parent indices are within parent tensor bounds
Performance Requirements
This method is called for every element access operation, so implementations must be highly optimized. Consider:
Pre-computing stride multipliers
Using lookup tables for common patterns
Minimizing array allocations
Return
the corresponding coordinates in the parent tensor's coordinate system
Parameters
the coordinates in the view's coordinate system
Throws
if childIndices are invalid or out of bounds
if childIndices has incorrect dimensionality