NumPy C-API#
NumPy provides a C-API to enable users to extend the system and getaccess to the array object for use in other routines. The best way totruly understand the C-API is to read the source code. If you areunfamiliar with (C) source code, however, this can be a dauntingexperience at first. Be assured that the task becomes easier withpractice, and you may be surprised at how simple the C-code can be tounderstand. Even if you don’t think you can write C-code from scratch,it is much easier to understand and modify already-written source codethan create itde novo.
Python extensions are especially straightforward to understand becausethey all have a very similar structure. Admittedly, NumPy is not atrivial extension to Python, and may take a little more snooping tograsp. This is especially true because of the code-generationtechniques, which simplify maintenance of very similar code, but canmake the code a little less readable to beginners. Still, with alittle persistence, the code can be opened to your understanding. Itis my hope, that this guide to the C-API can assist in the process ofbecoming familiar with the compiled-level work that can be done withNumPy in order to squeeze that last bit of necessary speed out of yourcode.
- Python types and C-structures
- System configuration
- Data type API
- Array API
- Array structure and data access
- Creating arrays
- Dealing with types
- Array flags
- ArrayMethod API
- API for calling array methods
- Functions
- Auxiliary data with object semantics
- Array iterators
- Broadcasting (multi-iterators)
- Neighborhood iterator
- Array scalars
- Data-type descriptors
- Data Type Promotion and Inspection
- Custom Data Types
- Conversion utilities
- Including and importing the C API
- Array iterator API
- ufunc API
- Generalized universal function API
- NpyString API
- NumPy core math library
- Datetime API
- C API deprecations
- Memory management in NumPy