Computer Physics Communications

This paper presents uniGasFoam, an open-source particle-based solver for multiscale rarefied gas flow simulations, which has been developed within the well-established OpenFOAM framework, and is an extension of the direct simulation Monte Carlo (DSMC) solver dsmcFoam+. The developed solver addresses the coupling challenges inherent in hybrid continuum-particle methods, originating from the disparate nature of finite-volume (FV) solvers found in computational fluid dynamics (CFD) software and DSMC particle solvers. This is achieved by employing alternative stochastic particle methods, resembling DSMC, to tackle the continuum limit. The uniGasFoam particle-particle coupling produces a numerical implementation that is simpler and more robust, faster in many steady-state flows, and more scalable for transient flows compared to conventional continuum-particle coupling. The presented framework is unified and generic, and can couple DSMC with stochastic particle (SP) and unified stochastic particle (USP) methods, or be employed for pure DSMC, SP, and USP gas simulations. To enhance user experience, reduce required computational resources and minimise user error, advanced adaptive algorithms such as transient adaptive sub-cells, non-uniform cell weighting, and adaptive global time stepping have been integrated into uniGasFoam. In this paper, the hybrid USP-DSMC module of uniGasFoam is rigorously validated through multiple benchmark cases, consistently showing excellent agreement with pure DSMC, hybrid CFD-DSMC, and literature results. Notably, uniGasFoam achieves significant computational gains compared to pure dsmcFoam+ simulations, rendering it a robust computational tool well-suited for addressing multiscale rarefied gas flows of engineering importance.

This paper introduces the MeshAC package, which generates three-dimensional adaptive meshes tailored for the efficient and robust implementation of multiscale coupling methods. While Delaunay triangulation is commonly used for mesh generation across the entire computational domain, generating meshes for multiscale coupling methods is more challenging due to intrinsic discrete structures such as defects, and the need to match these structures to the continuum domain at the interface. The MeshAC package tackles these challenges by creating hierarchical mesh structures linked through a novel modified interface region. It also incorporates localized modification and reconstruction operations specifically designed for interfaces. These enhancements improve both the implementation efficiency and the quality of the coupled mesh. Furthermore, MeshAC introduces a novel adaptive feature that utilizes gradient-based a posteriori error estimation, which automatically adjusts the atomistic region and continuum mesh, striving for an appropriate trade-off between accuracy and efficiency. This package can be directly applied to the geometry optimization problems of a/c coupling in static mechanics [1], [2], [3], [4], [5], with potential extensions to many other scenarios. Its capabilities are demonstrated for complex material defects, including straight edge dislocation in BCC W and double voids in FCC Cu. These results suggest that MeshAC can be a valuable tool for researchers and practitioners in computational mechanics.

We present the new version of the FeynGrav. The package provides tools to operate with Feynman rules for quantum gravity within FeynCalc. The latest version improves package efficiency and implements new physical models. We discover recurrent relations between metric factors that enhance computational efficiency. We discuss gravitational interaction with Horndeski gravity, quadratic gravity, and the simplest axion-like coupling. We implemented the massive graviton propagator and discussed the possibility of implementing massive gravity within the package.

ERMES 20.0 is an open-source software which solves the Maxwell's equations in frequency domain with the Finite Element Method (FEM). The new ERMES 20.0 is a significant upgrade from the previous ERMES 7.0 [1]. It introduces new features, modules, and FEM formulations to address the challenging problems commonly encountered in the design and analysis of nuclear fusion reactors [2]. Key additions are the electrostatic and cold plasma module, along with new FEM formulations as the stabilized double-curl edge element formulation [3] and the local L^2 projection method with nodal and bubble elements [4,5]. Furthermore, all the formulations now include an A-V potentials version. The ample set of methods available in the new ERMES 20.0 allows the user to select the most suitable FEM formulation to generate the best possible conditioned matrix for each specific problem. ERMES 20.0 operates in the static, quasi-static and the high-frequency regimens, making it a versatile tool which can be used in a wide variety of situations. For instance, it had been applied to microwave engineering, bioelectromagnetics, and electromagnetic compatibility. Now, thanks to the new electrostatic and cold plasma modules, the range of applications has been extended to relevant nuclear fusion engineering problems as: the computation of induced forces, plasma control, probability estimation of electric arc initiation, current distribution in arbitrary geometries, and the study of electromagnetic wave-plasma-wall interactions inside a fusion reactor. ERMES 20.0 is available for Windows and Linux systems and it has improved its capabilities to solve large problems on High Performance Computing (HPC) infrastructures thanks to its new interface with the solver libraries PETSc [6] and Python NumPy [7]. As in previous versions, ERMES 20.0 features a graphical user-friendly interface integrated into the pre- and post-processor GiD [8]. GiD handles geometrical modeling, data input, meshing, and result visualization. ERMES 20.0 is licensed under the open-source software 2-clause BSD license. This document is accompanied by a comprehensive manual that provides a step-by-step installation guide, a detailed description of all the new features and formulations, as well as the executables, user interface, examples, and source code of ERMES 20.0.

We present KinetiX, a software toolkit to generate computationally efficient fuel-specific routines for the chemical source term, thermodynamic and mixture-averaged transport properties for use in combustion simulation codes. The C++ routines are designed for high-performance execution on both CPU and GPU architectures. On CPUs, chemical kinetics computations are optimized by eliminating redundant operations and using data alignment and loops with trivial access patterns that enable auto-vectorization, reducing the latency of complex mathematical operations. On GPUs, performance is improved by loop unrolling, reducing the number of costly exponential evaluations and limiting the number of live variables for better register usage. The accuracy of the generated routines is checked against reference values computed using Cantera and the maximum relative errors are below 10^-7. We evaluate the performance of the kernels on some of the latest CPU and GPU architectures from AMD and NVIDIA, i.e., AMD EPYC 9653, AMD MI250X, and NVIDIA H100. The routines generated by KinetiX outperform the general-purpose Cantera library, achieving speedups of up to 2.4x for species production rates and 3.2x for mixture-averaged transport properties on CPUs. Compared to the routines generated by PelePhysics (CEPTR), KinetiX achieves speedups of up to 2.6x on CPUs and 1.7x on GPUs for the species production rates kernel on a single-threaded basis.

Neural Networks (NNs) are effective models for refining the accuracy of molecular dynamics, opening up new fields of application. Typically trained bottom-up, atomistic NN potential models can reach first-principle accuracy, while coarse-grained implicit solvent NN potentials surpass classical continuum solvent models. However, overcoming the limitations of costly generation of accurate reference data and data inefficiency of common bottom-up training demands efficient incorporation of data from many sources. This paper introduces the framework chemtrain to learn sophisticated NN potential models through customizable training routines and advanced training algorithms. These routines can combine multiple top-down and bottom-up algorithms, e.g., to incorporate both experimental and simulation data or pre-train potentials with less costly algorithms. chemtrain provides an object-oriented high-level interface to simplify the creation of custom routines. On the lower level, chemtrain relies on JAX to compute gradients and scale the computations to use available resources. We demonstrate the simplicity and importance of combining multiple algorithms in the examples of parametrizing an all-atomistic model of titanium and a coarse-grained implicit solvent model of alanine dipeptide.

The double wall-ring (DWR) rotational configuration is nowadays the instrument of choice regarding interfacial shear rheometers (ISR) in rotational configurations. Complex numerical schemes must be used in the analysis of the output data in order to appropriately deal with the coupling between interfacial and bulk fluid flows, and to separate viscous and elastic contribution or the interfacial response. We present a second generation code for analyzing the interfacial shear rheology experimental results of small amplitude oscillatory measurements made with a DWR rotational rheometer. The package presented here improves significantly the accuracy and applicability range of the previous available software packages by implementing: i) a physically motivated iterative scheme based on the probe's equation of motion, ii) an increased user selectable spatial resolution, and iii) a second order approximation for the velocity gradients at the ring surfaces. Moreover, the optimization of the computational effort allows, in many cases, for on-the-fly execution during data acquisition in real experiments.

Parton And-hadron China Institute of Atomic Energy (PACIAE) is a multipurpose Monte Carlo event generator developed to describe a wide range of high-energy collisions, including lepton-lepton, lepton-hadron, lepton-nucleus, hadron-hadron, hadron-nucleus, and nucleus-nucleus collisions. It is built based on the PYTHIA program, and incorporates parton and hadron cascades to address the nuclear medium effects. PACIAE 4.0 is the new generation of PACIAE model surpassing the version 3.0. In PACIAE 4.0, the old fixed-format FORTRAN 77 code has been refactored and rewritten by the free-format modern Fortran and C++ languages. The C++-based PYTHIA 8.3 is interfaced in, while previous versions connected to the Fortran-based PYTHIA 6.4 only. Several improvements are also introduced, which enable PACIAE 4.0 to contain more physics and features to model the high-energy collisions. This is the first attempt to transition PACIAE from Fortran to C++.

Machine learning has been revolutionizing our world over the last few years and is also increasingly exploited in several areas of physics, including quantum dynamics and control. The need for a framework that brings together machine learning models and quantum simulation methods has been quite high within the quantum control field, with the ultimate goal of exploiting these powerful computational methods for the efficient implementation of modern quantum technologies. The existing frameworks for quantum system simulations, such as QuTip and QuantumOptics.jl, even though they are very successful in simulating quantum dynamics, cannot be easily incorporated into the platforms used for the development of machine learning models, like for example PyTorch. The TorchQC framework introduced in the present work comes exactly to fill this gap. It is a new library written entirely in Python and based on the PyTorch deep learning library. PyTorch and other deep learning frameworks are based on tensors, a structure that is also used in quantum mechanics. This is the common ground that TorchQC utilizes to combine quantum physics simulations and deep learning models. TorchQC exploits PyTorch and its tensor mechanism to represent quantum states and operators as tensors, while it also incorporates all the tools needed to simulate quantum system dynamics. All necessary operations are internal in the PyTorch library, thus TorchQC programs can be executed in GPUs, substantially reducing the simulation time. We believe that the proposed TorchQC library has the potential to accelerate the development of deep learning models directly incorporating quantum simulations, enabling the easier integration of these powerful techniques in modern quantum technologies.

Due to the similarities between electron-nucleus (eA) and neutrino-nucleus scattering (νA), eA data can contribute key information to improve cross-section modeling in eA and hence in νA event generators. However, to compare data and generated events, either the data must be radiatively corrected or radiative effects need to be included in the event generators. We implemented a universal radiative corrections program that can be used with all reaction mechanisms and any eA event generator. Our program includes real photon radiation by the incident and scattered electrons, and virtual photon exchange and photon vacuum polarization diagrams. It uses the “extended peaking” approximation for electron radiation and neglects charged hadron radiation. This method, validated with GENIE, can also be extended to simulate νA radiative effects. This work facilitates data-event-generator comparisons used to improve νA event generators for the next-generation of neutrino experiments.