Session Q08: Detector Simulation and Computational Physics

Live

The PEnergy fitter was developed for SNO$+$ to provide more accurate event energy reconstruction, achieving energy resolution comparable to the Poisson limit while automatically accounting for detector configuration changes and position-dependent variations in detector response. However, the simulation-based approach that leads to these advantages also adds significant additional computation time. Here, we present PEnergyC, an improved version of PEnergy that utilizes Chroma, a GPU accelerated photon Monte Carlo to speed up the photon tracking process in PEnergy simulations. This approach retains the basic design and functionality of the original fitter, while significantly improves its run time from \textasciitilde 50s / 5MeV event to \textasciitilde 2s / 5MeV event. The same approach can be adopted to improve other important processes that involve significant photon propagation tracing. The design and usage of the new fitter is described, and results of the simulated events fits in LAP/PPO and in water fill of SNO$+$ are presented and compared with those of the original PEnergy fitter and other conventional energy reconstruction fitters.   

Live

The rapid development of general-purpose computing on graphics processing units (GPGPU) is allowing the implementation of highly-parallelized Monte Carlo simulation chains for high-energy physics experiments. This technique is particularly suitable for the simulation of a 3D pixelated charge readout for liquid argon time projection chambers, given the large number of channels that this technology employs. This solution is actively being explored by the DUNE collaboration for the design of the near detector. Here we present the first implementation of a full Monte Carlo simulator for a 3D pixelated charge readout using a set of GPU-optimized algorithms. The algorithms have been written in Python and translated into CUDA kernels using Numba, a just-in-time compiler for a subset of Python and NumPy instructions. The results of the simulation are compared against data from prototype LArTPCs employing 3D pixelated charge readout. This implementation will also allow taking full advantage of Perlmutter, the next-generation NERSC supercomputer, which will feature dedicated NVIDIA GPU nodes.   

Live

A Geant4 framework has been initialized to simulate the operation and performance characteristics of polarized targets at the CLAS12 spectrometer in Hall B of Jefferson Lab. The simulation is aimed at defining the parameters of the experiment such as beam size and position, optimal shielding from gamma radiation and Moller electrons, the vertex and momentum reconstruction, and beamline configuration to minimize accidentals in the CLAS12 detector. This simulation framework has been utilized in various polarized experiments such as the proposed experiment of spin-dependent electron scattering from a polarized $^3$He target at CLAS12\footnote{JLab Conditionally Approved Proposal PR12-20-002: {\it A Program of Spin-Dependent Electron Scattering from a Polarized $^3$He Target in CLAS12}, Co-Spokespeople: Harut Avakian, James Maxwell, Richard Milner, Dien Nguyen.}. In this talk, we will present the details of the Geant4 framework, the previous and recent development and the future plan of this simulation effort on polarized targets.   

Live

The High Luminosity Large Hadron Collider (HL-LHC) upgrade will require use of new computing resources such as HPCs in their production workflows. Preparing MadGraph, an event generator, for use of these resources yields an opportunity to improve the computational efficiency of MadGraph. We have studied computational hotspots to gain insight as to where MadGraph is using its computational resources. The Intel VTune Profiling tool is used to assess the "hot-spots" in MadGraph for Leading Order (LO) and Nest-to-Leading Order (NLO) processes. The identification of the time-consuming parts of the MadGraph workflow provides valuable insight for the parallelisation of parts of MadGraph. Event generation was expected to take the most amount of time in MadGraph, as it generates millions of events. The most CPU-intensive parts of event generation with MadGraph were found to be the matrix-element evaluation and LHAPDF.   

Live

A variable ranking system utilizing dense neural networks trained for binary categorization is presented in the context of the search for four-top production in the single-lepton final state. The ranking framework, Variable Importance, defines the importance metric as the difference in the Area Under the Curve scores between two trained networks where one network has N inputs and the other has N-1 inputs. The quantity used for ranking is the significance, defined as the ratio between the mean and RMS of the importance metric distributions. Variable Importance is characterized by its large-scale computing approach and incorporates a method for accounting for correlation between variables. In addition to determining a ranking order, the Variable Importance framework includes steps for hyper parameter optimization and k-fold cross validation to obtain a final trained model. Tests on the predictive ability of the significance for model performance are shown as well as the resulting four-top, single-lepton final state simulation-based significance and limits derived from the discriminator of the final trained model.   

Live

One of the biggest challenges in the High-Luminosity LHC (HL-LHC) era will be significantly more data to be recorded and analyzed from the collisions at the ATLAS and CMS experiments. ServiceX is a software R\&D project in the area of Data Organization, Management and Access (DOMA) of the IRIS- HEP to investigate new computational models for the HL-LHC era. ServiceX is an experiment-agnostic service to enable on-demand data delivery specifically tailored for nearly-interactive vectorized analyses. It is capable of retrieving data from the data lake, on-the-fly data transformation, and delivering user-selected data in a variety of different formats which can be streamed to the analysis system using an efficient wire protocol that can be readily consumed by a variety of computational frameworks. We will briefly discuss the current status of ServiceX and introduce few practical use cases. An integration of ServiceX and TRExFitter, a popular framework to perform profile likelihood fits in ATLAS, will be discussed with an emphasis on the delivery of data.   

Live

The Fisher-Bingham distribution (FB$_8$) is an eight-parameter family of probability density functions (PDF) on the unit sphere that, under certain conditions, reduce to spherical analogues of bivariate normal PDFs. Due to difficulties in its interpretation and estimation, applications have been mainly restricted to subclasses of FB$_8$, such as the Kent (FB$_5$) or von Mises-Fisher (vMF) distributions. However, these subclasses often do not adequately describe directional data that are not symmetric along great circles. In this talk I will outline a series calculation of the FB$_8$ normalization constant, which allows for some speed improvements. I will then discuss some applications for directional reconstruction in IceCube.   

Live

Spin crossover (SCO) materials are a possible solution to challenges faced in developing new nonvolatile voltage-controlled micro and nanoelectric devices. The design and implementation of devices benefits from ab initio calculations determining device characteristics and behaviors. Previous work by Bousseksou et al. shows that the phase transition between the high-spin (HS) and low-spin (LS) states of a SCO material may be described by a 2D Ising-type model, where the Ising ``spin-up'' and ``spin-down'' states correspond to the HS and LS states respectively. 2D Ising-type models have successfully captured the behavior of 2D thin films such as graphene and CrSiTe, and 2D Monte Carlo Ising Model simulations already provide insight into the switching mechanism for experimentally observed thin film behaviors. By expanding the established 2D model to a 3D model for thin-film heterostructures, we capture the behavior of a SCO-based device by simulating the phase transition of a Fe(II) ion centered octahedral coordination compound adsorbed to a ferroelectric polymer in the presence of temperature change and electric field excitation. Furthermore, by varying parameters in the Ising-like Hamiltonian for the SCO molecules, we better direct the choice of potential SCO molecular candidates for viable spintronic device designs.