It is possible to use generic Python objects as the dtype for a NumPy array, but if you do this, you’ll get no better performance with NumPy than you would with Python generally.

You can file this post under “if Rhett doesn’t write it down, he will forget”. I guess you are thinking that I am a python expert. Perhaps you picture me like this: Oh, I pretend to be an ...

I think that it makes sense to support NumPy arrays especially given that this is the default datatype that pandas uses when reading arrays from parquet files, a fairly common use-case.

Here’s how to use Cython to accelerate array iterations in NumPy. NumPy gives Python users a wickedly fast library for working with data in matrixes.