Skip to content

uLib

DOI

uLib is the base toolkit library for the CMT (Cosmic Muon Tomography) project, developed at the University of Padova and INFN Sezione di Padova, Italy.

It provides:

  • Core – object model, timers, configuration, UUID utilities.
  • Math – linear algebra (Eigen3), structured grids, voxel images, ray-tracing, image filters.
  • Python bindings – full pybind11 interface for scripting and analysis workflows.
  • Optional CUDA acceleration for voxel filtering (transparent RAM ↔ VRAM management).

Quick Start

# Activate your conda/micromamba environment first
micromamba activate mutom

poetry install            # CPU build
USE_CUDA=ON poetry install # GPU build

conan install . --output-folder=build --build=missing
cmake --preset conan-release
cmake --build build --target uLib_python -j$(nproc)

export PYTHONPATH="$(pwd)/build/src/Python:$(pwd)/src/Python"

Then in Python:

import uLib

# Core
timer = uLib.Core.Timer()
timer.Start()

# Math
grid = uLib.Math.StructuredGrid([10, 10, 10])
grid.SetSpacing([1.0, 1.0, 1.0])

img = uLib.Math.VoxImage([10, 10, 10])
img.SetValue(0, 3.14)
print(img.GetValue(0))

Documentation Sections

Section Description
Python – Installation Environment setup, user install, developer build
Python – API Usage Full API reference with examples
Python – Developer Guide Adding bindings, running tests, build details
Python – Streaming Between Recipes Stream ports, reading and writing them, a DataBuffer in between, live updates
Python – Agent Context (pyagent) Tools and resources generated from the classes themselves, and the backend that calls them
Python – Agent Panel (agentchat) gcompose's chat with Claude, Gemini or OpenCode, acting on the open scene
C++ Build – Usage & CUDA CMake build, CUDA configuration
Geant4 integration: status and plan What of Geant4 is integrated, what is missing, and the order to do the rest
Geant4 features as uLib objects Sensitive detectors, scoring with filters, fields and limits, replicas and assemblies, parallel worlds and reflections -- with five runnable examples and ready-made scenes
Histograms and plots Histograms and point sets, the ROOT proxies that draw them as curves and scatters, grouping several onto one pad and what may not share one, and arranging pads on a canvas
uprops – Property & Metadata Engine Standalone properties, rich metadata, signaling, encoders/decoders, and proxy representations
DataSource – Raw & Embedded Data Storage Universal data source abstraction, file linking, stream access, and self-contained scene packing (.gcx)
DataSink – Output Streams & Export Packaging The write counterpart of DataSource: output port, format-identified encoding, embedded payloads, and context-wide flushing
DataSource/DataSink as Encoder–Decoder Components Stage 1 implemented. TypeId, EncodingId, codec registry, native codec, self-describing envelope and object transfer to files and other processes; schema archives and FlatBuffers/protobuf still planned