MJOLNIR: A Software Package for Multiplexing Neutron Spectrometers

This paper introduces MJOLNIR, a generalized software package designed to reduce, visualize, and analyze the complex data generated by novel multiplexing triple-axis neutron spectrometers, such as the CAMEA instrument, to facilitate improved data acquisition rates and collaborative advancements in multi-dimensional data modeling.

The Gist

The Big Picture: From a Single Camera to a 360-Degree Drone

Imagine you are trying to take a picture of a bustling city square to understand how people move.

The Old Way (Standard Triple-Axis Spectrometers):
Think of a standard neutron instrument like a photographer with a very powerful, high-quality camera, but it can only take one photo at a time. To see the whole square, the photographer has to stand in one spot, snap a photo, turn slightly, snap another, turn again, and snap a third. It takes a long time to build a complete picture, and you only get "slices" of the action.

The New Way (Multiplexing Spectrometers):
Now, imagine upgrading that photographer to a drone equipped with 13 different cameras all pointing in slightly different directions. Instead of taking one photo at a time, this drone captures a massive, 360-degree view of the city square all at once.

This is what the CAMEA instrument (and others like it) does. It uses "multiplexing," meaning it detects neutrons at many different angles and energies simultaneously. The result? We get data much, much faster, and we see a much bigger picture of how atoms move and interact.

The Problem: Too Much Data, Too Fast

Here is the catch: While the drone (the instrument) is amazing at taking pictures, it produces a massive amount of raw data.

If you have 13 cameras snapping photos from different angles, you end up with a chaotic cloud of millions of data points. It's like having a bucket full of puzzle pieces from 13 different puzzles, all mixed together. If you try to look at them with your naked eye, you just see a mess. You can't easily see the "big picture" or the specific patterns you are looking for.

The Solution: MJOLNIR (The Software)

This is where the software package MJOLNIR comes in. The authors (Jakob Lassa and colleagues) built this tool to act as a super-intelligent sorting machine and artist.

Think of MJOLNIR as a high-tech kitchen where the raw ingredients (the chaotic neutron data) are turned into a delicious, organized meal (a clear scientific graph).

Here is what MJOLNIR does, step-by-step:

1. The Translator (Data Conversion)

The raw data comes in a language the scientists don't speak directly (it's just numbers about detector positions). MJOLNIR acts as a translator. It takes the raw numbers and converts them into a map that scientists understand: a map of "Reciprocal Space."

• Analogy: Imagine the data is written in a secret code. MJOLNIR translates it into a standard GPS map so scientists can say, "Ah, the action is happening right here at these coordinates."

2. The Organizer (Visualization)

Once the data is translated, MJOLNIR helps scientists see it.

• 3D Viewer: It can build a 3D model of the data, allowing scientists to spin it around and look at it from different angles, just like rotating a 3D model of a building on a computer.
• Slicing: It can take "slices" through the data. If you have a loaf of bread (the data), MJOLNIR can slice it horizontally, vertically, or diagonally to show you exactly what's inside at any specific layer.
• The "Point-and-Click" Feature: The software isn't just for computer wizards. It has a graphical interface (a screen with buttons) so scientists can click a mouse to generate these complex 3D images without needing to write code.

3. The Filter (Cleaning the Data)

Sometimes, the drone picks up "noise"—like a bird flying by or a reflection that isn't part of the city square. MJOLNIR has a masking tool that lets scientists say, "Ignore everything in this red zone; it's just noise." This ensures they only analyze the real scientific signals.

4. The Analyst (Fitting)

Finally, once the data is clean and visualized, scientists need to compare it to theories. MJOLNIR helps them fit curves to the data points to see if their theories about how atoms behave match reality. It's like checking if the shape of the puzzle pieces matches the picture on the box.

Why is this important?

Before MJOLNIR, scientists using these new, super-fast instruments were struggling. They had all this amazing data, but no good way to process it. It was like having a Ferrari engine but no steering wheel.

MJOLNIR provides that steering wheel. It allows scientists to:

• Work faster: Process data in minutes instead of days.
• Collaborate: Since the software is written in Python (a popular, open language), scientists at different labs around the world can use the same tools.
• Discover more: By making the data easier to see and understand, they can find new secrets about how materials work, which could lead to better batteries, superconductors, or medical technologies.

In Summary

The paper introduces MJOLNIR, a software tool designed to tame the "data monster" created by next-generation neutron instruments. It turns a chaotic cloud of millions of measurements into clear, beautiful, and understandable 3D maps, helping scientists unlock the secrets of the microscopic world.

Technical Summary

1. Problem Statement

Modern neutron scattering research is shifting from standard triple-axis spectrometers to multiplexing triple-axis instruments (e.g., CAMEA, RITA-II, MultiFLEXX). While these instruments offer significant advantages—such as simultaneous detection of scattered neutrons at various angles and energies, leading to higher data acquisition rates and larger coverage of the dynamical structure factor $S(Q, \omega)$—they introduce a new level of data complexity.

• Data Volume and Structure: Unlike standard instruments that measure isolated cuts, multiplexing instruments generate "point clouds" of data across multiple $(Q, \omega)$ positions simultaneously.
• Processing Challenges: The post-processing requirements increase significantly. Standard reduction algorithms are insufficient because:
  • Data must be converted from detector coordinates to reciprocal space for multiple simultaneous scattering angles.
  • The resolution function changes across the probed volume due to out-of-plane scattering geometries (90° rotation of scattering direction).
  • There is a need to handle raw integer neutron counts for accurate statistical fitting (Poisson/Multinomial statistics) rather than pre-normalized values.
• Lack of Unified Tools: Prior to this work, there was no unified software framework capable of handling data reduction, visualization, and analysis across different multiplexing facilities.

2. Methodology

The authors developed MJOLNIR, a software package written in Python (open-source) to address these challenges. The methodology focuses on a modular architecture designed for flexibility, user-friendliness, and cross-platform compatibility.

A. Coordinate Systems and Data Conversion

MJOLNIR defines a hierarchy of coordinate systems to facilitate data treatment:

1. Instrument Coordinate System ($C_{instr}$): An orthonormal system $(Q_x, Q_y, Q_z, \Delta E)$ independent of the sample.
2. Sample Coordinate System ($C_{sample}$): Rotated to align with the sample's reciprocal lattice vectors.
3. Reciprocal Lattice Unit (RLU) System ($C_{RLU}$): Defined by specific reciprocal lattice vectors $(P_1, P_2, \Delta E)$.

The conversion process involves:

• Virtual Instrument Modeling: Using a Geometry module to create virtual representations of instruments (monochromators, analyzers, detectors) via XML files or scripts. This supports various configurations, including prismatic analyzers (multiple energies from one analyzer) and position-sensitive detectors.
• Parameter Extraction: Instrument parameters (scattering angles, final energies) can be derived from design specifications or measured experimentally to account for deviations.
• Transformation: Data is transformed from detector positions to RLU using the formalism of TasUBlib (adapted from C++ to Python).

B. Data Structure and Processing

• Raw Data Preservation: MJOLNIR keeps data in its original, un-normalized form (integer neutron counts, monitor counts, and efficiency factors) throughout the process. This is crucial for accurate statistical fitting later.
• Binning and Visualization: The software bins raw data into equi-sized voxels in reciprocal space. It supports 1D, 2D, and 3D visualizations.
• Masking: A dedicated module allows users to mask extrinsic signals (spurious peaks, unwanted phonons) using Boolean algebraic operations before analysis.

C. Interfaces

To accommodate different user expertise levels, MJOLNIR provides three interfaces:

1. Scripting Interface (Python): The core backbone, allowing full control over instrument modeling, conversion, and analysis.
2. Command Line Interface (CLI): Four specific commands (MJOLNIR3DView, MJOLNIRConvert, MJOLNIRCalibrationInspector, MJOLNIRHistory) for quick data checks and conversions during experiments.
3. Graphical User Interface (GUI): Built with Qt bindings, allowing point-and-click data inspection, figure generation, and the ability to export the actions as Python scripts.

3. Key Contributions

• Unified Framework: MJOLNIR is the first software package generalized to handle data from any multiplexing triple-axis spectrometer, not just the CAMEA instrument for which it was initially developed. It supports instruments like MultiFLEXX, FlatCone, and Bambus.
• Statistical Rigor: By separating normalization from data storage, the software enables advanced fitting routines that respect Poisson statistics (using integer counts), which is often lost in standard reduction pipelines.
• Modular Architecture: The code is split into logical modules:
  • Geometry: Virtual instrument modeling.
  • Data: Data handling, conversion, and masking.
  • Statistics: Fitting and analysis (currently integrating with uFit).
  • _tools: Utility functions and conversion libraries.
• Future-Proofing: The design anticipates future instruments like BIFROST (at the European Spallation Source), which will use time-of-flight methods, by supporting additional dimensions in data handling.

4. Results and Validation

• Commissioning Success: The software was successfully utilized during the commissioning phase of the CAMEA spectrometer at the Paul Scherrer Institut (PSI).
• Data Treatment: It was used to process and analyze real experimental data, specifically spin wave measurements on a large MnF2 single crystal at 10 K.
• Visualization Capabilities: The paper demonstrates the software's ability to:
  • Generate interactive 3D views of data in instrument and sample coordinates.
  • Create complex 2D cuts (intensity vs. Energy and Q) along arbitrary paths in reciprocal space.
  • Visualize constant energy planes and 3D scattering intensity maps with transparency for low-intensity regions.
• Usability: The GUI and CLI were shown to effectively streamline data inspection and conversion, bridging the gap between raw data acquisition and scientific analysis.

5. Significance

The development of MJOLNIR is significant for the neutron scattering community for several reasons:

• Enabling New Science: It unlocks the full potential of multiplexing spectrometers by solving the data reduction bottleneck, allowing researchers to exploit the high data acquisition rates of these instruments.
• Collaboration and Standardization: By providing a uniform framework, it facilitates collaboration across different facilities (e.g., PSI, HZB, ESS) and standardizes data treatment workflows.
• Transition to Python: It aligns with the broader trend in scientific computing toward Python, ensuring compatibility with modern libraries (NumPy, SciPy, Pandas, Matplotlib) and future large-scale facility systems.
• Advanced Analysis: The preservation of raw data counts enables more sophisticated modeling and fitting of multi-dimensional data, leading to more accurate extraction of microscopic material properties.

In summary, MJOLNIR is a critical infrastructure tool that transforms the complex, high-volume data output of next-generation neutron spectrometers into accessible, analyzable scientific results.