A little project for vectorized N-Dimensional Arrays in native mojo
- Vectorized, robustly statically typed, nd arrays with all of the standard matrix operations
- Eventually to be made compatible with mojo implementations or ABI of LAPACK and BLAS for linear algebra operations.
- Compatability with numpy arrays eventually for interactions with the rest of python.
- Once the standalone compiler is available and we can use more MLIR features MLIR tensor and Tensor Operations Standards will be implemented
- The current state of this project is for toy usage and nailing down desirable behavior for the complete future version.
- DTypePointer-based data storage 2d arrays
- Basic guard rails for getting and setting
- add, subtract, mult, truediv, floordiv, and pow for both arrays of the same size, slices of arrays, and single values(but not commutatively yet for all)
- Arrange, transpose, shape, to_numpy, and print basic implementations
- All single SIMD input operations from Standard Math: trig(except atan2) and hyperbolic trig, etc.
- Getting and setting from slices.
- git clone https://github.com/MadAlex1997/Mojo-Arrays
- run demo.mojo for examples