Unfolding Particle Detector Effects and solving QCD Inverse Problem with Generative AI

Advancements in artificial intelligence (AI) have revolutionized high-energy physics by enabling generative models to address key detector-related Challenges. This work explores the generative model to mitigate smearing, acceptance, and inefficiency in particle detectors, enhancing experimental precision. We present a generative model-based framework to model and correct detector distortions. Using the Jefferson Lab CLAS g11 experiment as a case study, our approach successfully unfolds detector effects in multi-particle final states while preserving multidimensional correlations despite complex reaction mechanisms. A key focus is addressing the acceptance problem—accurately modeling detector acceptance without computationally expensive simulations. By training generative model-based framework on simulated detector data, we generate synthetic distributions that match both measured and unmeasured data, providing an efficient alternative for acceptance corrections. Our approach extends to multiple topologies, shifting from reconstructing unmeasured events to generating events across the full phase space, ensuring accurate reproduction even in regions with restricted acceptance. We further address detector inefficiency, which arises when detector components fail to register particle attributes correctly. Our framework integrates all detector effects—smearing, acceptance, and inefficiency—into a single model, unfolding them to recover vertex-level distributions. Additionally, we develop generative models-based surrogate modeling framework to overcome the computational and analytical challenges of traditional theory-driven approaches in event-level analysis. By replacing complex quantum correlation function (QCF) models with generative AI based surrogates, including both Generative Adversarial Networks (GANs) and Diffusion Models (DMs), the proposed framework enables efficient generation of synthetic event-level data directly from QCD parameters. Moreover, this approach facilitates parameter inference by learning the mappings between theoretical parameters and observable data, bypassing the gradient calculation challenges of directly incorporating the QCF model into solving QCD inverse problems. Our results demonstrate the feasibility of generative models in improving detector performance, the capability of generative models-based surrogates to accurately replicate theoretical models and reconstructing QCFs from experimental data, and the enhancing of data accuracy in high-energy physics, opening avenues for broader applications.