Efficient Simulation Code Research Articles

Unified framework based parallel FEM code for simulating marine seismoacoustic scattering

The simulation of seismic wave propagation in marine areas, which belongs to seismoacoustic scattering problem, is complicated due to the fluid-solid interaction between seawater and seabed, especially when the seabed is saturated with fluid. Meanwhile, huge computation resources are required for large-scale marine seismic wave simulation. In the paper, an efficient parallel simulation code is developed to solve the near-field seismoacoustic scattering problem. The method and technologies used in this code includes: 1) Unified framework for acoustic-solid-poroelastic interaction analysis, in which seawater and dry bedrock are considered as generalized saturated porous media with porosity equals to one and zero, respectively, and the coupling between seawater, saturated seabed and dry bedrock can be analyzed in the unified framework of generalized saturated porous media and avoid interaction between solvers of different differential equation; 2) Element-by-element strategy and voxel finite-element method (VFEM), with these strategies, it only needs to calculate several classes of element matrix and avoid assembling and storing the global system matrices, which significantly reduces the amount of memory required; 3) Domain-partitioning procedure and parallel computation technology, it performs 3D and 2D model partitioning for the 3D and 2D codes respectively, sets up the velocity structure model for the partitioned domain on each CPU or CPU core, and calculates the seismic wave propagation in the domain using Message Passing Interface data communication at each time step; 4) Local transmitting boundary condition, we adopt multi-transmitting formula, which is independent of specific wave equations, to minimize reflections from the boundaries of the computational model. A horizontal layered model with the plane P-wave incident vertically from bottom is used to demonstrate the computational efficiency and accuracy of our code. Then, the code is used to simulate the wave propagation in Tokyo Bay. All codes were written following to the standards of Fortran 95.

Investigation of novel weight window methods in Serpent 2 for fusion neutronics applications

Released in 2009, the Serpent Monte Carlo code has established itself as a highly efficient and powerful simulation code for nuclear systems analysis. Originally developed for reactor physics applications, the scope of the code now extends to coupled multi-physics simulations and photon transport. The latter has allowed adoption of the code by the fusion neutronics community following developments of a coupled neutron-photon capability in 2014 and the ability to handle complex geometry types in 2016. The code is well validated for the energy regimes and geometry types one can expect in fission reactor analysis. Over the course of recent years a benchmarking effort has been undertaken for application of the code to nuclear fusion. Compared to nuclear fission, or accelerator based applications, the underlying particle interaction phenomena differ greatly at the energies expected in a fusion reactor as well as the specific responses that are of interest. In this paper, a novel weight window generation implementation in Serpent is investigated. The applicability of this method is demonstrated for the Frascati Neutron Generator (FNG) bulk blanket and shield experiment, part of the SINBAD database, and a DEMO helium cooled pebble bed (HCPB) computational model. A comparison is performed against MCNP using weight windows generated with ADVANTG. Excellent agreement is found for the specified tallies and the significant efficiency gain using weight windows generated using both methods is comparable. A robust variance reduction method implementation is fundamental to applications to fusion neutronics and as such, this work is an important step in deployment of Serpent for this type of analysis.

Simulation of Formation of Binary Compact Objects in Globular Cluster

Binary compact objects are the special binary systems, which were composed of compact objects (i.e., white dwarf, neutron star and black hole). They contribute a lot to the sources of gravitational waves. The study of binary compact objects in star clusters and galaxies can provide a theoretical guidance for gravitational wave detection and improve the probability of detection effectively. We simulate the formation and fraction of binary compact objects in a small globular cluster, via NBODY6++GPU, an efficient N-body simulation code. We obtain the fractions of white dwarf binaries and neutron star binaries at different ages. The results show that the dynamic interactions among stars can also result in some binary compact objects besides binary evolution.

PyGeNN: A Python Library for GPU-Enhanced Neural Networks.

More than half of the Top 10 supercomputing sites worldwide use GPU accelerators and they are becoming ubiquitous in workstations and edge computing devices. GeNN is a C++ library for generating efficient spiking neural network simulation code for GPUs. However, until now, the full flexibility of GeNN could only be harnessed by writing model descriptions and simulation code in C++. Here we present PyGeNN, a Python package which exposes all of GeNN's functionality to Python with minimal overhead. This provides an alternative, arguably more user-friendly, way of using GeNN and allows modelers to use GeNN within the growing Python-based machine learning and computational neuroscience ecosystems. In addition, we demonstrate that, in both Python and C++ GeNN simulations, the overheads of recording spiking data can strongly affect runtimes and show how a new spike recording system can reduce these overheads by up to 10×. Using the new recording system, we demonstrate that by using PyGeNN on a modern GPU, we can simulate a full-scale model of a cortical column faster even than real-time neuromorphic systems. Finally, we show that long simulations of a smaller model with complex stimuli and a custom three-factor learning rule defined in PyGeNN can be simulated almost two orders of magnitude faster than real-time.

An open-source 2D finite difference based transient electro-thermal simulation model for three-phase concentric superconducting power cables

Advances in superconductor technology make the prospect of economical operation of high temperature superconducting (HTS) power cables a practical concept for grid applications in urban centers. With more advanced cable designs being developed and commercialized, their geometrical features and dynamic behavior are becoming increasingly complicated to be modeled. This brings new challenges as the complex structure of HTS power cables significantly increases the computation power needed to perform simulations. In this paper we develop a two-dimensional open source simulation code based on the finite difference method which is solved by means of the alternating direction implicit routine. The algorithm has been written in MatLab programming language. The method improves computational performance and simulation time. In addition, this enables the creation of open-source simulation codes. A three-phase concentric HTS cable design has been chosen for the development of the code, nevertheless the model can be employed for any cable design. The results indicate an efficient, stable and powerful simulation code. During the development no numerical instabilities have been found. Besides that, the model is able to deliver quantities that are experimentally difficult to access. Simulation files are available for the scientific community on the HTS Modeling Workgroup webpage. 1 1Simulation files can be downloaded.here.

Impact of different compilers and build types on Geant4 simulation execution time

Experimental observations and advanced computer simulations in High Energy Physics (HEP) paved the way for the recent discoveries at the Large Hadron Collider (LHC) at CERN. Currently, Monte Carlo simulations account for a very significant amount of computational resources of the Worldwide LHC Computing Grid (WLCG). The current growth in available computing performance will not be enough to fulfill the expected demand for the forthcoming High Luminosity run (HL-LHC). More efficient simulation codes are therefore required. This study focuses on evaluating the impact of different build methods on the simulation execution time. The Geant4 toolkit, the standard simulation code for the LHC experiments, consists of a set of libraries which can be either dynamically or statically linked to the simulation executable. Dynamic libraries are currently the preferred build method. In this work, three versions of the GCC compiler, namely 4.8.5, 6.2.0 and 8.2.0 have been used. In addition, a comparison between four optimization levels (Os, O1, O2 and O3) has also been performed. Static builds for all the GCC versions considered, exhibit a reduction in execution times of about 10%. Switching to newer GCC version results in an average of 30% improvement in the execution time regardless of the build type. In particular, a static build with GCC 8.2.0 leads to an improvement of about 34% with respect to the default configuration (GCC 4.8.5, dynamic, O2). The different GCC optimization flags do not affect the execution times.

Simulation of motion of many ions in a linear Paul trap

The quadrupole linear Paul trap is one of the key instruments in building highly stable atomic clocks. However, a frequency reference based on a single trapped ion is limited in stability due to the time needed for the interrogation cycle which cannot be further shortened. A promising strategy is the utilization of multiple trapped ions. The ions of the same kind then repulse each other with the Coulomb force, which is countered by the ponderomotive force of the time depended field in the trap. A few ions form a chain along the axis of a linear Paul trap. Adding more ions (a few tens or hundreds) gives rise to Coulomb crystals. We created an efficient simulation code which calculates the motion of such collections of ions in quasi-static radiofrequency fields of real linear quadrupole traps (including the micromotion). We include a model for laser cooling of the ions. The simulation tool can be used to study the formation and the dynamics of Coulomb crystals under conditions corresponding to various experimental setups.

Bayesian Dynamic Finite‐Fault Inversion: 1. Method and Synthetic Test

AbstractDynamic earthquake source inversions aim to determine the spatial distribution of initial stress and friction parameters leading to dynamic rupture models that reproduce observed ground motion data. Such inversions are challenging, particularly due to their high computational burden; thus, so far, only few attempts have been made. Using a highly efficient rupture simulation code, we introduce a novel method to generate a representative sample of acceptable dynamic models from which dynamic source parameters and their uncertainties can be assessed. The method assumes a linear slip‐weakening friction law and spatially variable prestress, strength, and characteristic slip‐weakening distance along the fault. The inverse problem is formulated in a Bayesian framework, and the posterior probability density function is sampled using the Parallel Tempering Monte Carlo algorithm. The forward solver combines a 3‐D finite difference code for dynamic rupture simulation on a simplified geometry to compute slip rates and precalculated Green's functions to compute ground motions. We demonstrate the performance of the proposed method on a community benchmark test for source inversion. We find that the dynamic parameters are resolved well within the uncertainty, especially in areas of large slip. The overall relative uncertainty of the dynamic parameters is rather large, reaching ~50% of the averaged values. In contrast, the kinematic rupture parameters (rupture times, rise times, and slip values), also well resolved, have relatively lower uncertainties of ~10%. We conclude that incorporating physics‐based constraints, such as an adequate friction law, may serve also as an effective constraint on the rupture kinematics in finite‐fault inversions.

A New View on Auger Data and Cosmogenic Neutrinos in Light of Different Nuclear Disintegration and Air-shower Models

Abstract We study the implications of ultrahigh-energy cosmic-ray (UHECR) data from the Pierre Auger Observatory for potential accelerator candidates and cosmogenic neutrino fluxes for different combinations of nuclear disintegration and air-shower models. We exploit the most recent spectral and mass composition data (2017) with a new, computationally efficient simulation code, PriNCe. We extend a systematic framework, which has been previously applied in a combined fit by the Pierre Auger Collaboration, with the cosmological source evolution as an additional free parameter. In this framework, an ensemble of generalized UHECR accelerators is characterized by a universal spectral index (equal for all injection species), a maximal rigidity, and the normalizations for five nuclear element groups. We find that the 2017 data favor a small but constrained contribution of heavy elements (iron) at the source. We demonstrate that the results moderately depend on the nuclear disintegration (Puget–Stecker–Bredekamp, Peanut, or Talys) model and more strongly on the air-shower (EPOS-LHC, Sibyll 2.3, or QGSjetII-04) model. Variations of these models result in different source evolution and spectral indices, limiting the interpretation in terms of a particular class of cosmic accelerators. Better-constrained parameters include the maximal rigidity and the mass composition at the source. Hence, the cosmogenic neutrino flux can be robustly predicted. Depending on the source evolution at high redshifts, the flux is likely out of reach of future neutrino observatories in most cases, and a minimal cosmogenic neutrino flux cannot be claimed from data without assuming a cosmological distribution of the sources.

A scalable interface-resolved simulation of particle-laden flow using the lattice Boltzmann method

Abstract We examine the scalable implementation of the lattice Boltzmann method (LBM) in the context of interface-resolved simulation of wall-bounded particle-laden flows. Three distinct aspects relevant to performance optimization of our lattice Boltzmann simulation are studied. First, we optimize the core sub-steps of LBM, the collision and the propagation (or streaming) sub-steps, by reviewing and implementing five different published algorithms to reduce memory loading and storing requirements to boost performance. For each, two different array storage formats are benchmarked to test effective cache utilization. Second, the vectorization of the multiple-relaxation-time collision model is discussed and our vectorized collision and propagation algorithm is presented. We find that careful use of Intel’s Advance Vector Extensions and appropriate array storage formats can significantly enhance performance. Third, in the presence of many finite-size, moving solid particles within the flow field, three different communication schemes are proposed and compared in order to optimize the treatment of fluid-solid interactions. These efforts together lead to a very efficient LBM simulation code for interface-resolved simulation of particle-laden flows. Overall, the optimized scalable code of particle-laden flow is a factor of 4.0-to-8.5 times faster than our previous implementation.

Updated core libraries of the ALPS project

The open source ALPS (Algorithms and Libraries for Physics Simulations) project provides a collection of physics libraries and applications, with a focus on simulations of lattice models and strongly correlated systems. The libraries provide a convenient set of well-documented and reusable components for developing condensed matter physics simulation code, and the applications strive to make commonly used and proven computational algorithms available to a non-expert community. In this paper we present an updated and refactored version of the core ALPS libraries geared at the computational physics software development community, rewritten with focus on documentation, ease of installation, and software maintainability. Program summaryProgram Title: ALPS Core librariesProgram Files doi:http://dx.doi.org/10.17632/fckj5d7wtr.1Programming language: C++Licensing provisions: GNU GPLv3Nature of problem: Need for modern, lightweight, tested and documented libraries covering the basic requirements of rapid development of efficient physics simulation codes, especially for modeling strongly correlated electron systems.Solution method: We present a C++ open source computational library that provides a convenient set of components for developing parallel physics simulation code. The library features a short development cycle and up-to-date user documentation.External routines/libraries:CMake, MPI, Boost, HDF5.

Computational Simulations of Thermoelectric Transport Properties

This review examines computational simulations of thermoelectric properties, such as electrical conductivity, Seebeck coefficient, and thermal conductivity. With increasing computing power and the development of several efficient simulation codes for electronic structure and transport properties calculations, we can evaluate all the thermoelectric properties within the first-principles calculations with the relaxation time approximation. This review presents the basic principles of electrical and thermal transport equations and how they evaluate properties from the first-principles calculations. As a model case, this review presents results on Bi2Te3 and Si. Even though there is still an unsolved parameter such as the relaxation time, the effectiveness of the computational simulations on the transport properties will provide much help to experimental scientist researching novel thermoelectric materials. Key words: Thermoelectric properties, Computational simulations, Seebeck coefficient, Electrical conductivity, Thermal conductivity

On the Gröbner basis triangularization of constraint equations in natural coordinates

Efficient dynamic simulation code is essential in many situations (including hardware-in-the-loop and model-predictive control applications), and highly beneficial in others (such as design optimization, sensitivity analysis, parameter identification, and controller tuning tasks). When the number of modeling coordinates n exceeds the degrees-of-freedom of the system f, as is often the case when closed kinematic chains are present, the governing dynamic equations consist of n second-order ordinary differential equations (ODEs) coupled with m=n−f algebraic constraint equations. This set of n+m index-3 differential-algebraic equations can be difficult to solve in an efficient yet accurate manner. Embedding (or generalized coordinate partitioning) can be used to obtain f ODEs (one for each independent acceleration), which are generally more amenable to numerical integration; however, the dependent positions are typically computed from the independent positions at each time step. Newton–Raphson iteration is often used for solving the position-level kinematics, but only provides solutions to within a specified tolerance, and can require several iterations to converge. In this work, Grobner bases are used to obtain recursively solvable symbolic solutions for the dependent positions, which can then be evaluated to within machine precision using a fixed number of arithmetic operations. Natural coordinates are particularly attractive in this context, since the resulting constraint equations are maximally quadratic polynomials and are, therefore, easily triangularized. The proposed approach is suitable for use in an automated formulation procedure and, as demonstrated by three examples, is capable of generating highly efficient simulation code with minimal additional effort required at the formulation stage.

Self-amplified coherent spontaneous emission in a free electron laser with “quiet” bunches

For a planar FEL configuration we study stimulated coherent spontaneous emission driven by a gradient of the bunch current in the presence of different levels of noise in bunches. To perform a vast amount of simulations required for obtaining statistically valid results, we developed a memory and time efficient one-dimensional simulation code based on the integral solution to a Klein-Gordon equation describing the field evolution. The longitudinal granularity of the electron bunch density originating from shot noise is maintained throughout the analysis. Three-dimensional effects like transverse emittance and diffraction are taken into account in simulations via an effective FEL parameter calculated from Xie's fitting formula. Calculations are performed for an FEL model with the SwissFEL injector bunch parameters. It turns out that a reduction of noise by several orders of magnitude below the level of shot noise is required to mitigate the noise effect. We propose a novel scheme that allows for formation of electron bunches with a reduced level of noise and a high gradient of the current at the bunch tail to enhance coherent spontaneous emission. The presented scheme uses effects of noise reduction and controlled microbunching instability and consists of a laser heater, a shot noise suppression section as well as a bunch compressor. The noise factor and microbunching gain with and without laser heater are estimated. We found that shot noise reduction by three orders of magnitude can be achieved for a finite transverse size electron bunch.

An efficient parallel simulation of interacting inertial particles in homogeneous isotropic turbulence

This study has conducted parallel simulations of interacting inertial particles in statistically-steady isotropic turbulence using a newly-developed efficient parallel simulation code. Flow is computed with a fourth-order finite-difference method and particles are tracked with the Lagrangian method. A binary-based superposition method has been developed and implemented in the code in order to investigate the hydrodynamic interaction among many particles. The code adopts an MPI library for a distributed-memory parallelization and is designed to minimize the MPI communication, which leads to a high parallel performance. The code has been run to obtain collision statistics of a monodisperse system with St=0.4 particles, where St is the Stokes number representing the particle relaxation time relative to the Kolmogorov time. The attained Taylor-microscale based Reynolds number Rλ ranges from 54.9 to 527. The largest simulation computed the flow on 20003 grids and 10003 (one billion) particles. Numerical results have shown that the collision kernel increases for Rλ<100 then decreases as Rλ increases. This Reynolds dependency is attributed to that of the radial distribution function at contact, which measures the contribution of particle clustering to the collision kernel. The results have also shown that the hydrodynamic interaction for St=0.4 particles decreases both the radial relative velocity and radial distribution function at contact, leading the collision efficiency less than unity. The collision efficiency increases from 0.65 to 0.75 as Rλ increases for Rλ<200 and then saturates.

Triangularizing kinematic constraint equations using Gröbner bases for real-time dynamic simulation

Real-time simulation is an essential component of hardware- and operator-in-the-loop applications, such as driving simulators, and can greatly facilitate the design, implementation, and testing of dynamic controllers. Such applications may involve multibody systems containing closed kinematic chains, which are most readily modeled using a set of redundant generalized coordinates. The governing dynamic equations for such systems are differential-algebraic in nature—that is, they consist of a set of ordinary differential equations coupled with a set of nonlinear algebraic constraint equations—and can be difficult to solve in real time. In this work, the equations of motion are formulated symbolically using linear graph theory. The embedding technique is applied to eliminate the Lagrange multipliers from the dynamic equations and obtain one ordinary differential equation for each independent acceleration. The theory of Grobner bases is then used to triangularize the kinematic constraint equations, thereby producing a recursively solvable system for calculating the dependent generalized coordinates given values of the independent coordinates. The proposed approach can be used to generate computationally efficient simulation code that avoids the use of iteration, which makes it particularly suitable for real-time applications.

Efficient Simulation of the Spatial Transmission Dynamics of Influenza

Early data from the 2009 H1N1 pandemic (H1N1pdm) suggest that previous studies over-estimated the within-country rate of spatial spread of pandemic influenza. As large spatially resolved data sets are constructed, the need for efficient simulation code with which to investigate the spatial patterns of the pandemic becomes clear. Here, we present a significant improvement to the efficiency of an individual-based stochastic disease simulation framework commonly used in multiple previous studies. We quantify the efficiency of the revised algorithm and present an alternative parameterization of the model in terms of the basic reproductive number. We apply the model to the population of Taiwan and demonstrate how the location of the initial seed can influence spatial incidence profiles and the overall spread of the epidemic. Differences in incidence are driven by the relative connectivity of alternate seed locations. The ability to perform efficient simulation allows us to run a batch of simulations and take account of their average in real time. The averaged data are stable and can be used to differentiate spreading patterns that are not readily seen by only conducting a few runs.

Ontology-based simulation in agricultural systems modeling

A methodology and applications of ontology-based simulation are presented. An environment for building simulations based on the Lyra ontology management system is described which includes web-based visual design tools for constructing models and automatically generating simulation code. The ontology is used for representing all equations and all symbols appearing in these equations that are needed to describe a model. The example applications presented are models of soil, water, and nutrient management in citrus and sugarcane. Results thus far show that the ontology-based approach has advantages for representing the model structure, equations, and symbols, that complex models can be described in this format, and that efficient simulation code can be generated automatically from the ontology definition of the model. Potential applications, not yet fully explored, include ability to automatically connect models and data sources, using the ontology to organize model bases containing many models and model components, and using ontology reasoners to search for models, automatically discover model similarities and differences, and generate model instances from general principles.

Fast and memory‐efficient Monte Carlo‐based image reconstruction for whole‐body PET

Several studies have shown the benefit of an accurate system modeling using Monte Carlo techniques. For state-of-the-art whole-body positron emission tomography (PET) scanners, Monte Carlo-based image reconstruction is associated with a significant computational cost to calculate the system matrix as well as a large memory capacity to store it. In this article, the authors present a simulation-reconstruction framework to solve these problems on the Philips Gemini GS PET scanner. A fast, realistic system matrix simulation module was developed using egs_pet, which is an efficient PET simulation code based on EGSnrc. The generated system matrix was then used in a rotator-based ordered subset expectation maximization (OS-EM) algorithm, which exploits the rotational symmetry of a cylindrical PET scanner. The system matrix was further compressed by using sparse storage techniques. The system matrix simulation took five days on 50 cores of Xeon 2.66 GHz, resulting in a system matrix of 2.01 GB. The entire system matrix could be stored in the main memory of a standard personal computer. The image quality in terms of contrast-noise trade-offs was considerably improved compared to a standard OS-EM algorithm. The image quality was also compared to the clinical software on the scanner using routine parameter settings. The contrast recovery coefficient of small hot spheres and cold spheres was significantly improved. The results indicated that the proposed framework could be used for this PET scanner with improved image quality. This method could also be applied to other state-of-the-art whole-body PET scanners and preclinical PET scanners with a similar shape.

Efficient simulation of the spatial transmission dynamics of influenza

Early data from the 2009 H1N1 pandemic (H1N1pdm) suggest that previous studies over-estimated the within-country rate of spatial spread of pandemic influenza. As large spatially-resolved data sets are constructed, the need for efficient simulation code with which to investigate the spatial patterns of the pandemic becomes clear. Here, we describe a significant improvement in the efficiency of an individual-based stochastic disease simulation framework that has been used for multiple previous studies. We quantify the efficiency of the revised algorithm and present an alternative parameterization of the model in terms of the basic reproductive number. We apply the model to the population of Taiwan and demonstrate how the location of the initial seed can influence spatial incidence profiles and the overall spread of the epidemic. Differences in incidence are driven by the relative connectivity of alternate seed locations.