Contents

Numba

Aims & Outcomes:

  1. Look at the benefits of just in time compilation with numba

  2. Understand how the python ecosystem allows for performance tuning should you need it

Tip

Another library dask is useful for scaling up across cores and/or clusters

numba is the foundational package for accelerating numerical python code by translating a subset of python and numpy code into fast machine code using just in time compilation (JIT).

Personal Note:

This package saved me weeks of work during my PhD. I was able to speed up my python code to the point you could interact with the analysis in real time. My work went from overnight runs that took 8 hours to getting the results straight away.

It can be a hugely useful tool.

Here is the QuantEcon lecture on numba

We will work through the following numba demo notebook.

Local Option

You can download the notebook from here

Then browse to your download location and load jupyter:

cd ~/Downloads
jupyter notebook intro-to-numba.ipynb

Cloud Based Option

You can launch the notebook

statsmodels Other Packages