TLDecPy: An open Python framework for robust thermoluminescence glow-curve deconvolution and metrological uncertainty quantification

Thermoluminescence (TL) glow-curve deconvolution is routinely used to separate overlapping peaks and to estimate kinetic parameters (e.g., activation energy E and frequency factor s). However, many available tools remain closed-source, tied to licensed environments, or provide limited transparency regarding numerical stability, constraint enforcement, robustness to artefacts, and uncertainty reporting. This work presents TLDecPy, an open Python library for TL glow-curve modelling, simulation, and multi-peak deconvolution, coupled to a web-based graphical user interface TLDec Web built with Dash/Plotly.TLDecPy implements multiple kinetic families (first-, second-, and general-order; mixed-order; OTOR with Lambert-W / Wright-Omega; and continuous trap distributions), hybrid global--local optimisation, robust loss functions, and a structured uncertainty-budget module (covariance propagation, Type A/B contributions, and Monte Carlo / bootstrap validation). The TLDec Web service provides a guided workflow from data import and optional pre-processing to fitting diagnostics and fully reproducible exports (figures, tables, JSON results, and complete analysis sessions).Validation is demonstrated through a multi-phase protocol: (i) machine-precision evaluation of the first-order temperature-integral backend using an adaptive barycentric rational approximation (AAA), (ii) optimiser benchmarking under multi-minima landscapes, (iii) robustness to extreme outliers, (iv) intercomparison on reference glow-curve datasets (Refglow/GLOCANIN), (v) advanced OTOR and CTD fits, and (vi) metrological uncertainty validation using large-scale Monte Carlo re-fitting.