CosmoPH

CosmoPH is a full-stack web application for applying Topological Data Analysis (TDA) to Cosmic Microwave Background (CMB) maps. It helps users upload or select CMB data, preprocess maps, compute persistent homology, visualize topological signatures, compare results with Gaussian null models, and classify inflation-model patterns using machine learning.

The project is designed for students, researchers, and educators who want an accessible browser-based workflow for exploring primordial non-Gaussianity without manually running a complex Python/TDA pipeline.

Key Features

• FITS file upload and dataset selection for CMB analysis.
• Interactive preprocessing for CMB patches.
• Robust FITS loading with support for both 2D image HDUs and HEALPix-style table HDUs.
• Persistent homology computation for H0 and H1 features.
• Persistence diagram, Betti curve, persistence image, ECC curve, and lifespan histogram visualizations.
• Gaussian null-hypothesis comparison using calibrated TDA thresholds.
• Defect and anomaly detection overlays.
• Hybrid ML classifier using raw CMB map information and extracted TDA features.
• Demo workflow for quick testing without custom data.
• Export-ready result views and report page.
• Full frontend included in the main repository.

Screenshots

Page / View	Screenshot Path
Home
Upload Dataset
Results Dashboard
Defect Detection
Anomaly Classification
Graphs & Analytics
Overall Gaussianity / Workflow

Tech Stack

Layer	Technologies
Frontend	Next.js, React, TypeScript, Tailwind CSS
Visualization	Plotly.js, react-plotly.js
UI Icons	lucide-react
Backend	FastAPI, Pydantic, Uvicorn
Astronomy / FITS	Astropy, Healpy
TDA	Ripser, Persim
Machine Learning	PyTorch, scikit-learn, joblib
Data Processing	NumPy, SciPy
Testing	Pytest, ESLint

Project Structure

Code1/
|-- backend/
|   |-- app/
|   |   |-- main.py
|   |   |-- routes/
|   |   |-- schemas/
|   |   |-- services/
|   |   |-- models/
|   |   `-- utils/
|   |-- tests/
|   |-- requirements.txt
|   `-- Dockerfile
|-- frontend/
|   |-- app/
|   |   |-- dashboard/
|   |   |-- demo/
|   |   |-- docs/
|   |   |-- report/
|   |   |-- results/
|   |   `-- upload/
|   |-- components/
|   |-- lib/
|   |-- public/
|   |-- types/
|   |-- package.json
|   `-- Dockerfile
|-- dataset/
|-- scripts/
|-- uploads/
|-- docker-compose.yml
|-- PRD.txt
|-- UPDATES.md
`-- README.md

How To Run

Prerequisites

• Python 3.10 or newer
• Node.js 18 or newer
• npm
• Git

1. Clone The Repository

git clone https://github.com/Yog964/CosmoPH.git
cd CosmoPH

2. Run The Backend

cd backend
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000

Backend URLs:

• API: http://localhost:8000
• Swagger docs: http://localhost:8000/docs

3. Run The Frontend

Open a new terminal:

cd frontend
npm install
npm run dev

Frontend URL:

• App: http://localhost:3000

4. Run With Docker

From the project root:

docker-compose up --build

Main Workflow

1. Open the frontend in the browser.
2. Upload a FITS file or select a sample dataset.(https://irsa.ipac.caltech.edu/data/Planck/release_3/)
3. Configure preprocessing options.
4. Generate a clean CMB patch.
5. Run TDA computation.
6. View persistence diagrams, Betti curves, persistence images, and related plots.
7. Compare the sample against Gaussian null models.
8. Inspect detected defects or anomalies.
9. Run the ML classifier.
10. Export or review the generated report.

API Endpoints

Method	Endpoint	Purpose
GET	/health	Check backend health
GET	/api/datasets	List available datasets
POST	/api/upload	Upload a FITS file
POST	/api/preprocess	Run preprocessing
POST	/api/compute-tda	Start TDA computation
GET	/api/compute-tda/{job_id}	Check TDA job status
GET	/api/results/{job_id}	Fetch result data
GET	/api/export/{job_id}	Export result files
POST	/api/demo	Run demo analysis

Useful Scripts

Generate or prepare sample datasets:

python scripts/download_datasets.py

Generate Gaussian and non-Gaussian FITS samples:

python scripts/generate_fits.py

Train or refresh the classifier:

python scripts/train_classifier.py

Testing

Backend tests:

cd backend
pytest tests/ -v

Frontend lint:

cd frontend
npm run lint

Notes

• Large generated files such as .zip, .fits, .npy, and .npz are ignored.
• Uploaded files and generated outputs are stored locally during development.
• The frontend is tracked as normal source code inside this repository.

License

MIT License.