Machine Learning in Physics

Code Repository for a Special Topics Course in Machine Learning in Physics

Introduction

This repository contains machine learning tutorials and lectures associated with the Special Topics Course on Machine Learning in Physics at Florida State University, Department of Physics. The course is designed for graduate students and senior undergrads who have little to no experience with machine learning, but have some experience with Python.

Dependencies

The notebooks in this repository depend on one or more of several well-known well-engineered and free Python modules.

modules	description
pytorch	a powerful, flexible, research-level machine learning toolkit
scikit-learn	easy to use machine learning toolkit
numpy	array manipulation and numerical analysis
matplotlib	a widely used plotting module for producing high quality plots
scipy	scientific computing
sympy	an excellent symbolic mathematics module
pandas	data table manipulation, often with data loaded from csv files
imageio	photo-quality image display module
iminuit	a rewrite of the venerable CERN minimizer Minuit
emcee	one of many, many, Markov chain Monte Carlo modules
tqdm	progress bar
joblib	module to save and load Python objects
importlib	importing and re-importing modules

Installation

The simplest way to install these Python modules is first to install a software environment system. You could just bite the bullet and install Anaconda! However, it may be better to install miniconda3, which is a slim version of Anaconda, on your laptop. Do so by following the instructions at:

https://docs.anaconda.com/miniconda/

Software environment systems such as Anaconda (conda for short) make it possible to have several separate self-consistent named environments on a single machine, say your laptop. For example, you may need to use Python 3.11.8 and an associated set of compatible packages and at other times you may need to use Python 3.12.4 with packages that require that particular version of Python. If you install software without using environments there is the danger that the software on your laptop will eventually become inconsistent. Anaconda (and its lightweight companion miniconda) provide a way, for example, to create a software environment consistent with Python 3.11.8 and another that is consistent with Python 3.12.4 without conflict.

Of course, like anything humans make, miniconda3 is not perfect. There are times when the only solution is to delete an environment and rebuild by reinstalling the desired packages.

Miniconda3

After installing miniconda3, it is a good idea to update conda using the command

conda update conda

Step 1

Assuming conda is properly installed and initialized on your laptop, you can create an environment, here called mlphysics using the command

conda create --name mlphysics

and activate it by doing

conda activate mlphysics

You need create the environment only once, but you must activate the desired environment whenever you create a new terminal window.

Step 2

First install pytorch. (Tip: search the web for conda install and the module name to get the exact syntax just in case it has changed.)

	conda install pytorch –c pytorch

Step 3

Install jupyterlab, matplotlib, scikit-learn, etc.

	conda install jupyterlab notebook
	conda install matplotlib
	conda install scikit-learn
	conda install pandas
	conda install sympy
	conda install imageio
	conda install tqdm

Again be sure to check the exact syntax. This does change from time to time!

Step 4

Install git if it is not yet on your system, then download the mlinphysics package.

	conda install git
	mkdir tutorials
	cd tutorials
	git clone https://github.com/hbprosper/mlinphysics
	cd mlinphysics
	pip install -e .

In the above the GitHub package mlinphysics has been downloaded into a directory called tutorials and installed so that it is available from any other folder on your machine.

Step 5

Open a new terminal window, navigate to the folder (directory) that contains mlinphysics and run jupyter lab in that window (in blocking mode).

	jupyter lab

If all goes well, the jupyter lab environment will appear in your default browser. Navigate to the mlinphysics folder and under the Files menu item, click on the notebook test.ipynb and execute it. This notebook tries to import several Python modules. If it does so without error messages, you are ready to try out the other notebooks.

In general, it is better to work in another folder and to copy the pieces of mlinphysics you wish to work with. That way you avoid conflicts if you need to update mlinphysics by navigating to that folder and doing

	git pull

to download the latest version of this package.