alibz

LIBS (Laser-Induced Breakdown Spectroscopy) data analysis toolkit for spectral peak finding, Voigt profile fitting, elemental indexing, and PCA-based peak-shape decomposition.

Installation

pip install -e .

This installs alibz as an editable package with all dependencies (numpy, scipy, scikit-learn, scikit-optimize, matplotlib, pulp).

For GPU acceleration (optional):

pip install -e ".[gpu]"

Modules

Module	Description
alibz.peaky_finder	Peak detection, FFT background removal, multi-Voigt fitting
alibz.peaky_indexer_v3	Whole-pattern spectral indexer (Bayesian optimisation + NNLS); the only supported indexer
alibz.peaky_maker	Forward spectral synthesis via Saha-Boltzmann
alibz.peaky_corpus	Batch loading, standardisation, parallel fitting, width statistics
alibz.peaky_pca	PCA peak-shape decomposition and broadening classification
alibz.background_pca	Corpus-level PCA for detector artifact and baseline removal
alibz.detector	Three-segment detector model: junction detection, artifact removal, segmented background subtraction
alibz.utils	Physical constants, Voigt utilities, NIST database interface, Saha-Boltzmann solver

alibz.peaky_indexer_v3 is the only indexer module. PeakyFitter has been removed.

Quick Start

from alibz import PeakyFinder, PeakyIndexer

# Load data and fit peaks
finder = PeakyFinder("path/to/raw/data")
fit_dict = finder.fit_spectrum_data(0, n_sigma=1, plot=False)

# Whole-pattern indexing
indexer = PeakyIndexer(fit_dict["sorted_parameter_array"])
result = indexer.run(shift_tolerance=0.1, max_ion_stage=2, n_calls=20, verbose=False)

print(result.temperature, result.ne)       # plasma T and log10(ne)
print(result.concentrations)               # per-species concentrations

The indexer returns a FitResult dataclass with temperature, electron density, per-species concentrations, residuals, R-squared, and peak assignments.

CLI

Peak-shape PCA pipeline

run-corpus-pca /path/to/libs/data \
    --pattern '**/*AverageSpectrum.csv' \
    --n-components 5 \
    --workers 4 \
    --timeout 120 \
    --fit-checkpoint data/fits.pkl \
    --out data/results.pkl

Background / artifact PCA

background-pca /path/to/libs/data \
    --n-components 20 \
    --batch-size 200 \
    --out data/bg_pca.pkl

Both commands accept --gpu / --no-gpu flags and support multiple input directories.

Notebooks

• notebooks/peaky_demo_v1.ipynb -- Peak finding, fitting, and elemental indexing workflow
• notebooks/peaky_data.ipynb -- Corpus-level PCA, peak width statistics, peak-shape decomposition, and PC-based peak characterisation
• notebooks/peaky_syndexer.ipynb -- End-to-end synthetic spectrum generation, fitting, and indexing stress tests
• notebooks/peaky_analyzer.ipynb -- Interactive peak-shape and PCA analysis scratchpad
• notebooks/peaky_tester.ipynb -- Interactive exploratory testing notebook

Tests

pip install -e ".[dev]"
pytest tests/

License

Unlicense (public domain)