Physics Data Explorer – CERN Open Data (CMS Dimuon Sample)

This repository contains a small analysis of real proton–proton collision data from the CMS experiment at CERN, using the public dimuon dataset
Dimuon_DoubleMu.csv from the CERN Open Data portal.

The goal of this project is to demonstrate basic high-energy physics data analysis using the standard Python scientific stack:

• NumPy for numerical work
• pandas for data handling
• Matplotlib and Seaborn for visualisation

---

Notebook

Main notebook:

• Physics_Data_Explorer_CERN_Open_Data.ipynb

The notebook is written as a step-by-step mini analysis and includes:

1. Data loading

   • Load the CSV file Dimuon_DoubleMu.csv from CERN Open Data into a pandas DataFrame.
   • Inspect the structure, data types, and first few events.

2. Data cleaning

   • Drop rows with missing values.
   • Apply simple quality cuts:
     • Require positive energy (E1, E2) and transverse momentum (pt1, pt2) for both muons.

3. Summary statistics

   • Compute mean, median, and standard deviation for:
     • Muon energies: E1, E2
     • Muon transverse momenta: pt1, pt2
     • Dimuon invariant mass: M
   • Cross-check results using both pandas and NumPy.

4. Visualisation
   Using Matplotlib and Seaborn:

   • Histograms

     • Compare E1 vs E2 energy distributions.
     • Compare pt1 vs pt2 transverse momentum distributions.
     • Plot the dimuon invariant mass M to look for structures (e.g. Z-boson region).

   • Line plot (NumPy running average)

     • Convert the dimuon mass column to a NumPy array.
     • Compute a running (moving) average using np.convolve.
     • Plot raw event-by-event masses together with the smoothed curve.

   • Scatter plots

     • E1 vs pt1 and E2 vs pt2 (full range and zoomed views).
     • Visualise how energy and transverse momentum are related for each muon.

   • Correlation heatmap

     • Compute the Pearson correlation matrix for E1, E2, pt1, pt2, and M.
     • Display it as a Seaborn heatmap to summarise relationships between key variables.

5. Simple physics interpretation

   • Discuss the typical energy and momentum scales of the muons.
   • Comment on the shape of the dimuon mass distribution and the region compatible with Z→μ⁺μ⁻ decays.
   • Compare muon kinematics in different dimuon mass regions (e.g. low-mass vs Z-peak) in a basic way.

6. Conclusion

   • Summarise what was learned from the dataset.
   • Outline possible next steps (e.g. fitting the Z peak, background studies, or extending to other CMS Open Data samples).

---

Dataset

• Source: CERN Open Data – CMS experiment
• Record: Dimuon_DoubleMu – Education and outreach dataset
• File used: Dimuon_DoubleMu.csv

Each row corresponds to one dimuon event, with columns including:

• E1, E2 – energies of muon 1 and muon 2 (GeV)
• px1, py1, pz1, pt1, eta1, phi1 – kinematics of muon 1
• px2, py2, pz2, pt2, eta2, phi2 – kinematics of muon 2
• M – dimuon invariant mass (GeV)
• Additional event and detector information

---

Technologies

• Python
• NumPy
• pandas
• Matplotlib
• Seaborn

---

How to run the notebook

1. Clone this repository

   git clone https://github.com/Mahsuba/cern-physics-data-explorer.git
   cd cern-physics-data-explorer

Install dependencies

It is recommended to use a virtual environment, but for a simple test:

bash

pip install numpy pandas matplotlib seaborn

Open the notebook

You can run the notebook in:

Jupyter Notebook / JupyterLab

VS Code with the Python & Jupyter extensions

Google Colab by uploading the .ipynb file

Then run the cells from top to bottom to reproduce all results and plots.

Notes

This project is intentionally kept at an introductory level. It is not a full physics analysis, but it demonstrates:

Working with real CERN Open Data

Using the Python scientific ecosystem for exploratory data analysis

Producing clear visualisations

Connecting plots to basic physical interpretation

It can be extended with more advanced techniques, such as fitting the dimuon mass peak to extract the Z boson mass and width, estimating backgrounds, or incorporating additional CMS Open Data samples.