LLVM math library
The purpose of this project is to provide portable math functions, many of which are in C99 and not available on all platforms. It is based on NumPy’s umath and tries to support all floating point and complex types.
The library can be compiled with any C compiler, or to LLVM assembly using Clang, to be linked into modules containing functions for jitting.
$ git clone email@example.com:ContinuumIO/llvmmath.git $ cd llvmmath $ python setup.py install