pybind11 VS NumPy

I'm investigating using C++ to build a REST server, and would love to know of people's experiences with Crow-- or whether they would recommend something else as a "medium-level" abstraction C++ web server. As background, I started off experimenting with Python/FastAPI, which is great, but there is too much friction to translate from pybind11-exported C++ objects to the format that FastAPI expects, and, of course, there are inherent performance limitations using Python, which could impact scaling up if the project were to be successful.

I have a C++ function computing a large tensor which I would like to return to Python as a NumPy array via pybind11.

This is based on the Google Ink Stroke Modeler C++ library, and using pybind11 to make it available on python.

pybind11 makes it easy to call C++ from Python if you want to mix.

If you've never used Pybind before these pybind tests[1] and this repo[2] have good examples you can crib to get started (in addition to the docs). Once you handle passing/returning/creating the main data types (list, tuple, dict, set, numpy array) the first time, then it's mostly smooth sailing.

Pybind offers a lot of functionality, but core "good parts" I've found useful are (a) use a numpy array in Python and pass it to a C++ method to work on, (b) pass your python data structure to pybind and then do work on it in C++ (some copy overhead), and (c) Make a class/struct in C++ and expose it to Python (so no copying overhead and you can create nice cache-aware structs, etc.).

[1] https://github.com/pybind/pybind11/blob/master/tests/test_py...

I have a python module for which the logic is written in C++ and I use pybind11 to expose the objects and functions to Python.

Reading from Python requires a wrapper, using pybind11 this is fairly done.

Also I can highly recommend starting using C++ to augment your Python code, i.e. find the parts that are slow or undoable in Python and write those in C++ then expose them as Python functions. You can use https://github.com/pybind/pybind11 to call C++ code from Python.

In NumPy with @, dot() or matmul():

In NumPy with * or multiply(). ` or multiply()` can multiply 0D or more D arrays by element-wise multiplication.

Data science projects often use numpy. However, numpy objects are not JSON-serializable and therefore require conversion to standard python objects in order to be saved:

numpy: A library for scientific computing in Python

Python has become a preferred language for data analysis due to its simplicity and robust library ecosystem. Among these, NumPy stands out with its efficient handling of numerical data. Let’s say you’re working with numbers for large data sets—something Python’s native data structures may find challenging. That’s where NumPy arrays come into play, making numerical computations seamless and speedy.

I was looking at all the big projects like numpy, pytorch, flask, etc.

I wish the NumPy devs would more thoroughly consider adding full fluent API support, e.g. x.sqrt().ceil(). [Issue #24081]