qupid

English | 한국어

Quasi-Signal Generation Model for Pixelized Detector.

qupid models the charge a traversing particle deposits on a pixelized silicon sensor as an analytic 2D distribution, rather than running a full Geant4/TCAD simulation. It integrates that distribution over a pixel matrix, applies a firing threshold, and studies the resulting cluster size by Monte Carlo. Geometry and units follow ALPIDE-like sensors: lengths in µm (pixel pitch 29.24 × 26.88), signal amplitude in electrons (3.6 eV per electron-hole pair), a typical firing threshold of 250 e⁻.

The model

The signal is an amplitude times a 2D probability density. QPSignalGaussian takes the density from a mean vector and a covariance matrix, so σx, σy, and a rotation are all free. A digitizer sums the injected signals and integrates the analog field over each cell of the pixel matrix; QPDigitizerSquare is a uniform rectangular grid. Pixels whose integrated signal exceeds the threshold are the fired pixels, and their count is the cluster size. QPMonteCarlo runs this chain over randomized events and emits one record per event. Random numbers come through the QPRandomBase interface; the default implementation, QPRandom, uses std::mt19937_64.

The supporting linear algebra and math (QPVector, QPMatrix, QPMath — Gaussian variants, Woods-Saxon, rotation matrices, ellipse geometry) are hand-written and use nothing beyond the standard library.

Dependencies

• Core (libqupid): a standard C++17 toolchain, nothing else. The signal model, digitizer, pixel integration, RNG, and Monte Carlo driver all live here.
• Visualization (libqupidVis, from vis/): ROOT. QPGraphFactory renders a digitizer's analog, digital, and thresholded views as ROOT TH2D histograms. This module is optional and built only when ROOT is present.

ROOT's TRandom can drive the Monte Carlo in place of QPRandom through the header-only adapter vis/include/QPRandomTRandom.hpp; only an executable that includes it links ROOT.

Build

CMake ≥ 3.16.

cmake -B build
cmake --build build -j

By default (QUPID_WITH_ROOT=ON) CMake looks for ROOT and builds libqupidVis and the ROOT-dependent tests when it is found; if ROOT is missing it builds the core alone and reports that it did. To leave ROOT out entirely:

cmake -B build -DQUPID_WITH_ROOT=OFF
cmake --build build -j

Either way libqupid and the core tests are built. Executables land in the build directory and write their output (.csv, .root, .pdf) to the current directory, so run them from there.

Tests and examples

The programs in test/ double as usage examples, ordered from primitives to the full pipeline. test0_VectorMatrix, test3_digitizer, and test7_montecarlo_csv build and run without ROOT (test7 writes the Monte Carlo output as CSV). test1, test2, test4, test5, and test6 use ROOT for plotting or tree output and are built only in the ROOT configuration.

Tutorials

tutorials/ collects short, commented programs, built through the same core / ROOT split as the tests:

1. defining a Gaussian signal and integrating its density (core only);
2. a digitizer and the fired-pixel cluster size at a threshold (core only);
3. a Monte Carlo run to CSV (core only);
4. visualization with QPGraphFactory and the TRandom adapter (ROOT).

Documentation

The API documentation is generated with Doxygen (doc/Doxyfile) and published to GitHub Pages on each push to master: https://isaac-kwon.github.io/qupid/. For a ROOT interactive session, rootlogon.C loads libqupid and libqupidVis and applies the plotting style.

Provenance

Ported from the Python version, pyqupid.