Numba is a tool that compiles fast, specialized versions of Python functions at runtime. In this chapter, we learned how to compile, inspect, and analyze functions compiled by Numba. We also learned how to implement fast NumPy universal functions that are useful in a wide array of numerical applications. Finally, we implemented more complex data structures using the nb.jitclass decorator.

Tools such as PyPy allow us to run Python programs unchanged to obtain significant speed improvements. We demonstrated how to set up PyPy, and we assessed the performance improvements on our particle simulator application.

We also, briefly, described the current ecosystem of the Python compilers and compared them with each other. 

In the next chapter, we will learn about concurrency and asynchronous programming. Using these techniques, we will be able to improve the responsiveness and design of applications that spend a lot of time waiting for network and disk resources.