Very simple BLAS (Basic Linear Algebra Subroutines) implementation for embedded devices inspired by the numpy library.
This is a minimal library absolutely not meant to be perfect:
- It lacks many operations compared to a standard numerical computing library.
- The operations are all potentially computationally expensive since this library was not designed to be optimized.
- The methods used are numerically the simplest possible, but the library seems to be numerically stable if the sizes of the matrices used for calculations remain small.
- Its use is recommended with matrices of extremely small maximum size, otherwise the program may crash due to memory exhaustion.
- In order to avoid crashes due to memory exhaustion, manual deallocation of the arrays in use is required, or specifically through the destroy(float*) function, or completely through the destroyAll() function.
- Install cmake, make and the g++ compiler.
- ./start.sh
- If you want to try some specific function, you can use the main.cpp file under src.