35.1. Project Overview

Table of Contents

Build and analyze a complete Geant4 detector simulation

A complete Geant4 detector simulation is a small scientific project that follows the same structure as real research work. In this final project you will use everything you learned in the course to design a simple but realistic detector, simulate a beam or source, record what the detector would measure, and analyze the results with external tools such as ROOT.

At the highest level, your goal is to connect three parts: a well defined physics question, a Geant4 implementation that answers it, and an analysis that shows and interprets the results. The project chapters that follow this overview will guide you step by step, but here you should start thinking about the whole workflow and the choices you must make.

From question to simulation

You should begin with a clear, specific question. For example, you might want to know how much energy a gamma ray deposits in a scintillator of given size, what the depth dose of a proton beam in water looks like, or how efficient a small detector is for detecting 511 keV photons. The geometry, materials, particle source, and physics list must all reflect that question.

Your simulation question should be narrow enough that you can implement it with the tools from this course and run it on a normal computer in a reasonable time. You are not building a full experimental facility. Instead, you create a minimal but meaningful setup that captures the key physics of interest.

A useful project starts from a clear, quantitative question such as "What is the energy spectrum measured by this detector for particles of a given type and energy?" or "What is the detection efficiency as a function of energy or geometry?". Vague goals are very hard to validate.

When you define your goal, think in terms of observables: what quantity will you read from the simulation that answers the question? Examples include energy deposited in a volume per event, number of hits above threshold, time of flight between detectors, or angular distribution of outgoing particles. You will later map these observables to Geant4 objects such as hits, steps, or analysis histograms.

Planning the detector model

Once you know what you want to measure, you can sketch a conceptual detector. At this stage you should think about simple shapes and clear hierarchies. Ask yourself which parts are absolutely necessary: a world volume, one or more detector volumes, and possibly extra structures such as shielding or support material, if they are relevant to your question.

The next chapters will guide you through creating the geometry, but already in the overview you should decide the level of detail. For this project, moderate simplification is recommended. For example, a cylindrical scintillator can be represented by a single $G4Tubs$ volume, a water phantom by a $G4Box$, and a segmented detector by a regular array of repeated volumes.

Keep in mind that each detector element you want to distinguish in the analysis must usually correspond to a separate logical volume instance or a parameterized copy with a unique identifier. You should plan these identifiers in advance because they will be important when you record and analyze hits.

Choosing and defining materials

Materials determine how particles interact with your detector. In this final project you should use a small set of well defined materials. Many common choices such as air, water, silicon, or lead can be taken from the NIST database, while special detector materials like specific scintillators or biological tissue can be created or approximated as needed.

Before you start coding, decide for each volume in your geometry which material it should contain. Also decide if material properties, such as density or composition, are important for the observable you want to study. If the exact microstructure does not matter for your observable, a simpler material model is usually better. It keeps the simulation easier to understand, validate, and modify.

Configuring the particle source

The particle source is where your simulated particles begin. It must be consistent with your project goal. You will choose the particle type, energy or energy spectrum, initial position, and direction distribution. For example, a pencil beam entering a phantom, an isotropic source at the center of a detector, or a planar source illuminating a shield.

You can implement your source in C++ through a primary generator class or configure it with macro commands using a general particle source. In both cases, you should already think about how many events you need and what range of energies or angles is physically interesting. This will influence the run time of your simulation and the statistical precision of your final results.

Selecting suitable physics

A complete detector simulation must use a physics list that models the relevant interactions. For this project you are not expected to build a custom physics list from scratch, but you must choose a suitable reference physics list and, if needed, enable or disable certain components.

Your choice depends on the particles and energies in your problem. For example, if you simulate medical photons and electrons at MeV energies, you will prefer an electromagnetic physics configuration with good low energy performance. If you simulate protons in a water phantom, you must ensure that hadronic and electromagnetic interactions for protons are included. The physics list chapter in this course explains the options, and here you apply that knowledge by making and justifying one specific choice.

Implementing sensitive detectors and scoring

To extract meaningful information from your simulation, you need a way to record what happens inside chosen volumes. This is the role of sensitive detectors and hits collections. In this project you will implement at least one sensitive detector class that records the quantities linked to your project goal.

You must decide which information is really necessary. Typical choices include total energy deposition per event in a detector, position of an interaction, or time information for timing studies. For segmented detectors, you will also need a clear mapping from volume indices or copy numbers to detector IDs so that you can reconstruct which part of the detector was hit.

Think of the sensitive detector as the bridge between Geant4 tracking and your analysis code. It translates many low level steps and interactions into higher level detector signals that can be stored in histograms or ntuples.

Recording and organizing simulation data

The Geant4 analysis system will be your main tool to record results. In this final project you will plan in advance which histograms and ntuples you need. For instance, you might want an energy spectrum, an efficiency as a function of energy, or a depth dose profile.

Each quantity to be studied should have a clear place in your analysis output. This planning prevents you from cluttering the code with many unused data structures and also avoids missing important observables that would require rerunning long simulations. You will connect the analysis manager to your run, event, or stepping actions so that data is filled consistently.

Before running large simulations, define all needed histograms and ntuple columns and verify that they are correctly filled for a small test run. Changing analysis structure later often requires rerunning the simulation.

You will also choose an output format, such as ROOT or CSV, and a naming scheme for your output files. This makes it easier to analyze and compare different runs, for example when you change geometry, beam energy, or physics settings.

Running, validating, and analyzing

The last part of the project is to run the simulation, check that it behaves as expected, and analyze the output. You will first perform small test runs to verify that the geometry loads correctly, that there are no obvious errors such as overlapping volumes or missing energy deposition, and that your hits and analysis outputs are populated.

Then you will increase the number of events until the statistical uncertainties are acceptable for your question. You will compare key observables to simple expectations or reference data when available. For example, you might check rough absorption lengths, mean free paths, or peak positions in an energy spectrum. Even qualitative agreement with basic physics expectations is important for building trust in your model.

Finally, you will load your output files into an external analysis tool such as ROOT. There you will create plots, apply cuts, compute derived quantities such as efficiencies or depth dose curves, and generate figures that clearly communicate your results. You will link these figures back to your original question and describe what the simulation teaches you.

Putting everything together

By the end of this final project you will have built a full chain: from a physics question, through a Geant4 implementation with geometry, materials, sources, and physics, to sensitive detectors, data recording, and external analysis. Each of the following chapters in the Final Project section takes one of these pieces and guides you through its implementation. In this overview you have seen how they connect and how to think about the project as a coherent, scientific simulation, not just a collection of code fragments.

The main outcome is not only a working example, but also a structured approach that you can adapt to future detector simulations of your own.