How to return numpy ndarray objects from Rust to Python?

[ifsheldon]

Hi everyone! I'm new to pyo3 and I'd like to ask how I can return numpy ndarrays from Rust to Python.

Simply put, I want to write a function whose input and output are numpy ndarrays, write it in Rust with Rust-native data types and glue it to Python. I've been reading the documentations of pyo3 and rust-numpy, but I didn't find out related contents.

I know Vec<T> in Rust can be converted to list[T] in Python, but I don't know how much overhead it has to convert Vec<T> to list[T] and then to np.ndarray. If there's no direct mapping from np.ndarray to Rust's data representation, I think we will have to do something like np.ndarray -> list[T] -> Vec<T> -> Process in Rust -> Vec<T> -> list<T> -> np.ndarray.

If you have done that before or you know where the related documentation is, please give me a pointer. Thanks!

[davidhewitt]

You're probably looking for rust-numpy. See also #2029

[ifsheldon]

OK, I found the example in rust-numpy, thank you!