KI-Glossar
Das vollständige Wörterbuch der Künstlichen Intelligenz
Electromagnetic Calorimeter
A detector that measures the energy of light electrically charged particles (electrons, photons) by stopping them and causing a shower of secondary electrons and photons.
Track Reconstruction (Tracking)
The process of reconstructing the paths (trajectories) of charged particles through the internal detectors of an accelerator, essential for determining their momentum and charge.
Jet Identification (Jet Tagging)
Algorithmic classification of particle jets to determine their origin (quarks, gluons, or heavy particles like the top quark or the Higgs boson) using deep neural networks.
Pile-up Rejection
A set of techniques aimed at separating the signals from an interesting proton-proton collision from the background noise originating from other simultaneous collisions within the same proton bunch.
Graph Neural Networks (GNN) in Physics
Application of graph neural networks where nodes are detectors or particle tracks and edges are their relationships, optimizing the reconstruction of complex events.
Fast Simulation
The use of AI models (like GANs or VAEs) to generate realistic detector data much faster than traditional physics-based Monte Carlo simulation.
Collision Anomaly
Detection of collision events that do not match any known theoretical model, often performed by unsupervised learning algorithms to discover new particles or phenomena.
AI-Based Detector Calibration
The process of adjusting a detector's parameters using machine learning algorithms to correct for drifts and improve the accuracy of energy and position measurements.
Physical Object Reconstruction
Algorithm that assembles raw detector signals (tracks, energy deposits) into coherent physical objects like muons, electrons, photons, and jets.
Intelligent Trigger (Smart Trigger)
Real-time trigger system that uses lightweight AI models to filter and select the most relevant collision events among billions of collisions per second.
Jet Momentum Estimation
Prediction of a jet's four-momentum by correcting measured energies with neural networks that learn calorimeter losses and non-uniform responses.
Neutrino Detection
Inference of neutrino presence and energy, which interact very little, by analyzing missing transverse energy in a collision event via supervised learning models.
High Granularity Data Analysis
Processing of data from high-granularity detectors (like next-generation calorimeters) where AI is crucial for interpreting the massive volume of spatial and energy information.
Detector Physics Optimization
Use of AI to design and optimize the geometry and materials of future detectors by simulating millions of configurations to maximize sensitivity to new particles.
Energy Regression
Application of regression models (neural networks, decision trees) to more accurately estimate a particle's energy by correcting detector non-linearities and saturation effects.
Primary Vertexing
Reconstruction of the initial collision point (primary vertex) using AI algorithms to fit particle trajectories originating from it with high precision.
Weak Signal Search
Development of AI classifiers capable of distinguishing a rare and expected physical signal (e.g., decay of a new particle) from a background several orders of magnitude higher.