FDQLearn

Feynman Diagrams Quantum Learner:

The goal of the project is to create a quantum neural network that can efficiently compute the squared matrix element of a tree-level Feynman Diagram, given its topology and the necessary dynamical dofs. This work is get inspired by the work: “Learning Feynman Diagrams using Graph Neural Networks”. In: arXiv preprint arXiv:2211.15348 (2022), Harrison Mitchell, Alexander Norcliffe, and Pietro Liò, where they define a graph neural network for high energy QED.

This project aims to be an original work that goes deeper in the theory of quantum graph neural network and equivariant quantum machine learning applied to QFT. Using a QML approach seems natural to study QFT, because they follow the same physical laws of quantum theory, and exploiting them can bring us to perform straightforward computation in quantum regime that are unfeasible in classical regime, in a sort of quantum advantage in terms of code complexity.

In this project I develop a quantum graph neural network because it is the best way to encode the pictorial Feynman Diagrams, which contain essential information on the kind of interaction and then on the output results.

Some of the future aspects of the project is to enlarge the kind of diagrams and interactions that can be studied with this approach and try to build a network that encodes the symmetries of the interaction (both gauge and accidental symmetries) and study more nice computations that can't be done (either at all or in a feasible way) by classical machine learning.

Installing the virtual environment (venv command)

Here are showed the steps to setup the virtual environment with the right versions of the packages I've used; the version of Python is 3.9.13.

Create the environment:

1. $ python -m venv ./.venv

Activate the environment:

for Linux and MacOs:

1. $ source .venv/bin/activate

for Windows (cmd.exe): 2) $ .venv/Scripts/activate.bat

for Windows PowerShell: 2) $ .venv/Scripts/Activate.ps1

Install 'requirements.txt' and my library: 3) $ pip install -r requirements.txt 4) $ pip install -e .

5. In repo's root run setup.py

Regular use of the virtual environment (venv command)

for Linux and MacOs:

1. $ source .venv/bin/activate

for Windows (cmd.exe): 2) $ .venv/Scripts/activate.bat

for Windows PowerShell: 2) $ .venv/Scripts/Activate.ps1

3. run whatever cmd, es: $ python scripts/FDQLearn_e_mu_massive.py

4. to deactivate just run: deactivate

Dataset generation (venv command)

In the repo there are some existent datasets, but there's the original notebook from the previous paper that create a dataset; the command to execute the jupyter notebook (after activating the virtual environment) is:

$ ipython kernel install --user --name=.venv

Installation of the virtual environment (conda)

Here are showed the steps to setup the virtual environment with the right versions of the packages I've used; the version of Python is 3.9.13.

Create the environment:

1. $ conda env create -p ./.venv --file env.yml

Activate the environment: 2) $ conda activate ./.venv

Install 'requirements.txt' and my library: 3) $ pip install -f requirements.txt 4) $ pip install -e .

5. In repo's root run setup.py

Regular use of the virtual environment (conda)

1. $ conda activate ./.venv

2. run whatever cmd, es: $ python scripts/FDQLearn_e_mu_massive.py

3. to stop the virtual environment: conda deactivate

Dataset generation (conda)

In the repo there are some existent datasets, but there's the original notebook from the previous paper that create a dataset; the command to execute the jupyter notebook (after activating the virtual environment) is:

$ conda run -p ./.venv jupyter nbconvert --to notebook --stdout --execute scripts/Dataset_builder.ipynb