pyIRM-unmix: An Efficient Python Tool for IRM Unmixing

Identification and semi-quantification of magnetic mineral assemblages in natural samples are essential for understanding geological processes and paleoenvironmental evolution. Among various rock magnetic approaches, the unmixing of isothermal remanent magnetization (IRM) curves has been widely used to semi-quantify magnetic components and characterize their coercivity distributions. Although various mathematical models and unmixing tools have been developed, existing approaches usually rely on manual parameter setting. This procedure may introduce a degree of subjectivity, limits processing efficiency for high-resolution, continuous environmental magnetic datasets, and compromises the consistency of unmixing criteria among samples. To address these limitations, we developed pyIRM-unmix, an Python-based package for IRM unmixing. Compared to conventional workflows, pyIRM-unmix reduces the reliance on manual parameter tuning by enabling robust automatic parameter initialization and full-parameter fitting within a Bayesian modeling framework. Comparative tests demonstrate that the program has high statistical reliability. Furthermore, it substantially reduces processing time while maintaining a fitting accuracy comparable to, or even better than, that of established tools. With its computational efficiency, explicit statistical criteria, and applicability to large datasets, pyIRM-unmix effectively mitigates the subjectivity caused by manual intervention. It provides a practical new tool for high-throughput data analysis in environmental and rock magnetism.