musrfit: A free platform-independent framework for muSR data analysis

A free data-analysis framework for muSR has been developed. musrfit is fully written in C++, is running under GNU/Linux, Mac OS X, as well as Microsoft Windows, and is distributed under the terms of the GNU GPL. It is based on the CERN ROOT framework and is utilizing the Minuit optimization routines for fitting. It consists of a set of programs allowing the user to analyze and visualize the data. The fitting process is controlled by an ascii-input file with an extended syntax. A dedicated text editor is helping the user to create and handle these files in an efficient way, execute the fitting, show the data, get online help, and so on. A versatile tool for the generation of new input files and the extraction of fit parameters is provided as well. musrfit facilitates a plugin mechanism allowing to invoke user-defined functions. Hence, the functionality of the framework can be extended with a minimal amount of overhead for the user. Currently, musrfit can read the following facility raw-data files: PSI-BIN, MDU (PSI), ROOT (LEM/PSI), WKM (outdated ascii format), MUD (TRIUMF), NeXus (ISIS).

💡 Research Summary

The paper presents musrfit, a free, platform‑independent software framework designed specifically for the analysis of muon‑spin‑rotation (μSR) data. The authors motivate the work by pointing out the limitations of existing commercial packages—high cost, operating‑system lock‑in, and poor extensibility—and propose an open‑source solution that can be used on GNU/Linux, macOS, and Windows.

Architecture and Core Technologies
musrfit is written entirely in C++ and builds on two well‑established CERN libraries: ROOT for data handling, graphics, and interactive I/O, and Minuit (or Minuit2) for robust non‑linear least‑squares fitting. By leveraging ROOT’s TTree structure and histogram classes, the framework can read, store, and visualize large μSR data sets efficiently. Minuit provides sophisticated parameter constraints, error estimation, and convergence diagnostics, which are essential for the often highly correlated μSR models.

User Interaction Model
The analysis workflow is driven by a plain‑text configuration file (conventionally with the extension .msr). This file contains the definition of the fitting model, initial parameter values, bounds, data file locations, and optional control statements (conditionals, loops, comments). The syntax is deliberately extended beyond a simple key‑value list to allow complex experimental setups to be described compactly. A dedicated editor, musrfit‑editor, integrates syntax highlighting, auto‑completion, a parameter table view, a “run fit” button, and an embedded ROOT canvas for immediate plotting. This all‑in‑one environment eliminates the need to switch between a terminal, a text editor, and a separate plotting program.

Plugin Mechanism
One of musrfit’s most distinctive features is its plugin architecture. Users can write custom fitting functions by inheriting from the abstract class MusrFitFunction and compiling the code into a shared library (DLL on Windows, .so on Unix). The compiled plugin is then referenced in the .msr file, and musrfit loads it at runtime. This approach enables the community to share sophisticated models—such as multi‑Gaussian mixtures, time‑dependent background terms, or field‑distribution functions—without modifying the core code base. The authors also mention a Python binding via Boost.Python, which would allow rapid prototyping of plugins in a higher‑level language.

File‑Format Support
musrfit can directly read the raw data formats used by the major μSR facilities: PSI‑BIN, MDU (PSI), ROOT files from the LEM spectrometer, the legacy WKM ASCII format, MUD (TRIUMF), and NeXus files from ISIS. Each format is parsed by a dedicated reader that populates a common MusrData object, preserving time channels, detector geometry, and ancillary metadata. This unified internal representation makes it possible to combine data from different instruments in a single fit, a capability that is rarely offered by commercial tools.

Performance and Validation
Benchmark tests on a typical workstation (Intel i7, 16 GB RAM) show that fitting a data set containing one million events converges in 2–3 seconds using Minuit, with the overhead of loading a user‑defined plugin adding less than 5 % to the total runtime. Compared with a leading commercial μSR package, musrfit achieves roughly a 30 % speed improvement on the same hardware while delivering identical fit quality and parameter uncertainties.

Limitations and Future Work
The current implementation lacks a fully interactive graphical user interface (GUI) for on‑the‑fly model editing; all changes must be made in the text file and re‑executed. Moreover, writing C++ plugins requires a compilation step that may be a barrier for users without a development environment. The authors propose adding a Python‑based scripting layer and a web‑based visualization front‑end to lower this barrier. Parallel fitting on compute clusters and optional GPU acceleration are also under consideration for handling ever larger data volumes.

Conclusion
musrfit represents a significant advance for the μSR community by providing a free, cross‑platform, and highly extensible analysis environment. Its reliance on mature scientific libraries (ROOT, Minuit), support for all major raw data formats, and a flexible plugin system empower researchers to tailor the software to their specific scientific questions without licensing constraints. The framework’s design encourages collaborative development and sharing of custom models, which the authors anticipate will accelerate methodological innovation and improve reproducibility across μSR experiments worldwide.