powerxrd

Minimal, extensible Python tools for powder XRD analysis with a lightweight Rietveld refinement engine.

---

✨ Overview

powerxrd is a modular framework for:

• Parsing powder XRD data
• Background subtraction and smoothing
• Peak analysis
• Lightweight Rietveld refinement
• Extensible lattice definitions

It is not a replacement for FullProf or GSAS.

It is a compact, inspectable refinement engine designed for:

• Teaching
• Prototyping
• Research exploration
• Algorithm experimentation
• Rapid lattice model development

---

🧱 Architecture

The package is organized into clear layers:

Lattice  → geometry only (d-spacing, HKL generation)
Model    → profile + background + peak physics
Refine   → least-squares optimization
Workflow → staged refinement orchestration
Chart/Data → preprocessing utilities

This separation allows you to extend:

• New lattice systems (tetragonal, hexagonal, etc.)
• Alternative peak shapes
• Custom refinement strategies
• Structure factor implementations

The lattice layer is fully abstracted through BaseLattice, enabling clean geometric extensions.

---

📦 Installation

From PyPI (for users)

pip install powerxrd

---

🛠 Development Setup (recommended: uv)

PowerXRD uses uv for fast, reproducible dependency management.

Install uv:

pip install uv

Then inside the repository:

uv sync --group dev

This will:

• Create a virtual environment automatically
• Install runtime dependencies
• Install development tools (pytest, ruff, mypy)

No manual venv activation required.

To run anything inside the project environment:

uv run python script.py

---

🚀 Quick Example

from powerxrd.lattice import CubicLattice
from powerxrd.model import PhaseModel
import powerxrd.refine as rr

# Define lattice
model = PhaseModel(lattice=CubicLattice(a=4.0))

# Load experimental data (example)
x_exp, y_exp = rr.load_data()

# Staged refinement
rr.refine(model, x_exp, y_exp, ["scale"])
rr.refine(model, x_exp, y_exp, ["bkg_intercept", "bkg_slope"])
rr.refine(model, x_exp, y_exp, ["a"])

# Plot final fit
rr.plot_fit(model, x_exp, y_exp, model.pattern(x_exp))

Full examples are available in:

examples/

Including extended hello_rietveld workflows.

Run them via:

uv run python examples/hello_rietveld_long.py

---

🧪 Development

Run tests:

make test

Lint:

make lint

Format:

make format

Type check:

make type

Build package:

make build

---

🧠 Philosophy

This project prioritizes:

• Clarity over feature bloat
• Extensibility over monolithic design
• Numerical transparency
• Minimal abstraction overhead
• Inspectable scientific computation

It is intentionally small.

The goal is not to compete with large crystallographic suites — but to provide a clean, programmable refinement core.

---

📚 References

See RIETVELD_CLASSIC.md for foundational Rietveld literature and background resources.

---

License

MIT License © Andrew Garcia