Speaker
Description
The Deep Underground Neutrino Experiment (DUNE) is a next-generation neutrino oscillation experiment with near and far detectors. The DUNE Near Detector employs a pixelated liquid argon Time Projection Chamber (ND-LAr). Its pixelated readout system, LArPix, provides direct imaging of charge deposition through zero-suppressed, data-driven sampling. While this design enables high-rate operation with controlled power consumption, the large number of pixels poses challenges for scalable simulation. We present a GPU-accelerated framework for ND-LAr based on analytic signal modeling and sparse representations, enabling efficient and scalable forward simulation of detector response. Building on this framework, we introduce an improved charge unfolding approach that operates directly on non-equally spaced LArPix outputs, allowing reconstruction of continuous charge information and enhanced data-simulation comparisons. GPU parallelization delivers significant performance gains, making the framework suitable for large-scale ND-LAr simulation and validation studies.