remage: a Geant4-based simulation framework for low-background experiments

In the quiet corners of the universe, where the noise of everyday matter fades, scientists search for the faintest whispers of new physics. To hear these whispers, they build experiments deep underground, shielding their detectors from the constant rain of cosmic rays and natural radioactivity that bombards the Earth's surface. These experiments rely on a simple but profound idea: if you can make your detector clean enough and quiet enough, you might finally see a rare event that has never been observed before. But to know if you have truly found something new, you must first know exactly what your detector would see if nothing new were happening. This requires a perfect understanding of the background noise itself. To achieve this, researchers use powerful computer simulations that act as virtual laboratories, tracking how billions of invisible particles move, collide, and interact with the materials of their detectors. These simulations are essential for designing the experiments and for distinguishing a genuine discovery from a trick of the light.

For decades, creating these simulations has been a laborious task, often requiring each research team to write their own complex computer programs from scratch. This process is slow, prone to errors, and difficult to share, as every group reinvents the wheel to model their unique detector shapes and materials. A new framework called remage aims to change this by providing a shared, open-source toolkit that standardizes how these simulations are built. Instead of forcing scientists to write thousands of lines of difficult code for every new experiment, remage separates the description of the detector's physical shape from the heavy lifting of the particle tracking. The shape is defined using simple, flexible tools that anyone can edit, while the core engine handles the complex physics. This approach allows researchers to focus on the science rather than the software engineering, ensuring that the simulations are reliable, fast, and consistent across different experiments.

The heart of this new system is its ability to handle the geometry of an experiment with remarkable ease. In a typical simulation, the detector is described as a hierarchy of nested volumes, like a set of Russian dolls, each made of specific materials. Previously, changing even a small part of this structure required rewriting and recompiling the entire computer program. With remage, the geometry is described in a standard file format that can be created and modified using Python, a programming language known for its readability. This allows scientists to build complex detector models, complete with sensitive parts and optical properties, without needing to be experts in the underlying C++ code that runs the simulation. The system also includes built-in tools to check for errors, such as overlapping parts that shouldn't exist, and to benchmark how fast the simulation runs, ensuring that the virtual detector behaves exactly as the real one would.

Once the virtual detector is built, the next challenge is to simulate the particles that might enter it. In low-background physics, the sources of these particles are often radioactive decays happening within the materials of the detector itself or cosmic rays that have managed to penetrate deep underground. remage provides specialized tools to generate these starting points, or "vertices," with high precision. It can place these starting points anywhere within a volume, on a surface, or even just a few micrometers beneath a surface to model contamination that might have occurred during manufacturing. It can also generate the particles themselves, including rare events like double-beta decay or the showers of particles created when a cosmic ray hits the atmosphere. The system is flexible enough to accept input from other tools, allowing researchers to study exotic scenarios that no standard generator could produce.

The physics engine behind remage is carefully tuned to the specific needs of these experiments. It simulates the interactions of particles with matter, from the high-energy collisions of cosmic rays to the subtle energy deposits of radioactive decay. A key feature is its ability to handle the production of light within the detector. Many modern experiments use materials that glow when struck by a particle, and the system tracks these photons as they bounce off surfaces and are eventually caught by sensors. To make this process faster and more efficient, the system includes smart shortcuts. For instance, it can group together tiny steps that a particle takes within a fraction of a millimeter, treating them as a single event if they are too small to matter for the final measurement. It can also decide to ignore certain particles that are unlikely to reach the detector, saving vast amounts of computing power. These optimizations allow the simulation to run on large computer clusters, generating the massive amounts of data needed to study rare background sources.

The output of these simulations is designed to be immediately useful for analysis. Instead of dumping raw, unstructured data, remage organizes the results into clear, hierarchical files that record exactly what happened in the detector. It groups interactions that happen close together in time into "hits," mimicking how a real detector would record a signal. This structure makes it easy for scientists to apply models of how their detectors respond to energy, such as how the signal might blur or change shape depending on where the interaction occurred. The system has been rigorously tested against real-world data. In one validation study, researchers simulated a radioactive source placed near a germanium detector and compared the results to actual measurements taken in an underground laboratory. The simulation reproduced the energy spectrum of the data with high accuracy, confirming that the framework correctly models the complex interplay of particles and materials.

This framework is already being adopted by major experiments, including the LEGEND collaboration, which is searching for a rare nuclear decay that could help explain why the universe is made of matter rather than antimatter. By moving away from custom-built, isolated software to a shared, validated platform, these experiments can accelerate their design and analysis. The system is built to evolve, with continuous updates and automated testing ensuring that it remains compatible with the latest physics models. For the broader community of low-background physicists, remage offers a way to bypass the years of software development that once stood between a new idea and a testable simulation. It provides a robust, transparent foundation upon which the next generation of discoveries can be built, turning the complex task of simulating the invisible into a manageable and collaborative effort.