Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallelization efficiency”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,513 records · Page 84

Bulk ion acceleration and particle heating during magnetic reconnection in a laboratory plasma

Bulk ion acceleration and particle heating during magnetic reconnection are studied in the collisionless plasma of the Magnetic Reconnection Experiment (MRX). The plasma is in the two-fluid regime, where the motion of the ions is decoupled from that of the electrons within the ion diffusion region. The reconnection process studied here is quasi-symmetric since plasma parameters such as the magnitude of the reconnecting magnetic field, the plasma density, and temperature are compatible on each side of the current sheet. Our experimental data show that the in-plane (Hall) electric field plays a key role in ion heating and acceleration. The electrostatic potential that produces the in-plane electric field is established by electrons that are accelerated near the electron diffusion region. The in-plane profile of this electrostatic potential shows a “well” structure along the direction normal to the reconnection current sheet. This well becomes deeper and wider downstream as its boundary expands along the separatrices where the in-plane electric field is strongest. Since the in-plane electric field is 3–4 times larger than the out-of-plane reconnection electric field, it is the primary source of energy for the unmagnetized ions. With regard to ion acceleration, the Hall electric field causes ions near separatrices to be ballistically accelerated toward the outflow direction. Ion heating occurs as the accelerated ions travel into the high pressure downstream region. This downstream ion heating cannot be explained by classical, unmagnetized transport theory; instead, we conclude that ions are heated by re-magnetization of ions in the reconnection exhaust and collisions. Two-dimensional (2-D) simulations with the global geometry similar to MRX demonstrate downstream ion thermalization by the above mechanisms. Electrons are also significantly heated during reconnection. The electron temperature sharply increases across the separatrices and peaks just outside of the electron diffusion region. Unlike ions, electrons acquire energy mostly from the reconnection electric field, and the energy gain is localized near the X-point. However, the increase in the electron bulk flow energy remains negligible. These observations support the assertion that efficient electron heating mechanisms exist around the electron diffusion region and that the heat generated there is quickly transported along the magnetic field due to the high parallel thermal conductivity of electrons. Classical Ohmic dissipation based on the perpendicular Spitzer resistivity is too small to balance the measured heat flux, indicating the presence of anomalous electron heating.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Hybrid algorithms in quantum Monte Carlo

With advances in algorithms and growing computing powers, quantum Monte Carlo (QMC) methods have become a leading contender for high accuracy calculations for the electronic structure of realistic systems. The performance gain on recent HPC systems is largely driven by increasing parallelism: the number of compute cores of a SMP and the number of SMPs have been going up, as the Top500 list attests. However, the available memory as well as the communication and memory bandwidth per element has not kept pace with the increasing parallelism. This severely limits the applicability of QMC and the problem size it can handle. OpenMP/MPI hybrid programming provides applications with simple but effective solutions to overcome efficiency and scalability bottlenecks on large-scale clusters based on multi/many-core SMPs. We discuss the design and implementation of hybrid methods in QMCPACK and analyze its performance on current HPC platforms characterized by various memory and communication hierarchies.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

The artificial retina processor for track reconstruction at the LHC crossing rate

We present results of an R&D study for a specialized processor capable of precisely reconstructing, in pixel detectors, hundreds of charged-particle tracks from high-energy collisions at 40 MHz rate. We apply a highly parallel pattern-recognition algorithm, inspired by studies of the processing of visual images by the brain as it happens in nature, and describe in detail an efficient hardware implementation in high-speed, high-bandwidth FPGA devices. This is the first detailed demonstration of reconstruction of offline-quality tracks at 40 MHz and makes the device suitable for processing Large Hadron Collider events at the full crossing frequency.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Accelerating Subsurface Transport Simulation on Heterogeneous Clusters

Reactive transport numerical models simulate chemical and microbiological reactions that occur along a flowpath. These models have to compute reactions for a large number of locations. They solve the set of ordinary differential equations (ODEs) that describes the reaction for each location through the Newton-Raphson technique. This technique involves computing a Jacobian matrix and a residual vector for each set of equation, and then solving iteratively the linearized system by performing Gaussian Elimination and LU decomposition until convergence. STOMP, a well known subsurface flow simulation tool, employs matrices with sizes in the order of 100x100 elements and, for numerical accuracy, LU factorization with full pivoting instead of the faster partial pivoting. Modern high performance computing systems are heterogeneous machines whose nodes integrate both CPUs and GPUs, exposing unprecedented amounts of parallelism. To exploit all their computational power, applications must use both the types of processing elements. For the case of subsurface flow simulation, this mainly requires implementing efficient batched LU-based solvers and identifying efficient solutions for enabling load balancing among the different processors of the system. In this paper we discuss two approaches that allows scaling STOMP's performance on heterogeneous clusters. We initially identify the challenges in implementing batched LU-based solvers for small matrices on GPUs, and propose an implementation that fulfills STOMP's requirements. We compare this implementation to other existing solutions. Then, we combine the batched GPU solver with an OpenMP-based CPU solver, and present an adaptive load balancer that dynamically distributes the linear systems to solve between the two components inside a node. We show how these approaches, integrated into the full application, provide speed ups from 6 to 7 times on large problems, executed on up to 16 nodes of a cluster with two AMD Opteron 6272 and a Tesla M2090 per node.

GPU↗

Demand Driven Energy Management for PIPO Auxiliary Power Supply Architecture

In this paper, a demand driven energy management (DDEM) technique is implemented on modular parallel input parallel output DC/DC auxiliary power supply (APS) structure. In this method, modules are brought in/out of the operation making them operate under high loading conditions, avoiding low efficiency under light load. Such a method finds its application in power conservative systems with variable load. Implementation of DDEM and the energy saved is shown using APS comprised of two 15W commercially off-the-shelf DC-DC modules, with 100–1000 VDC input and 24 V output voltage. Along with the DDEM, the lifetime of the power stage devices is controlled minimizing their junction temperature swing through control of cooling system. This control method uses digital twin as interface to virtualize the power devices loading operation/ health condition. Accordingly, no extra sensor or closed loop control required to measure the junction temperature of the devices or feedback the losses to the cooling flow.

Azadeh, Yalda↗

Analyzing the Impact of System Reliability Events on Applications in the Titan Supercomputer

Extreme-scale computing systems employ Reliability, Availability and Serviceability (RAS) mechanisms and infrastructure to log events from multiple system components. In this paper, we analyze RAS logs in conjunction with the application placement and scheduling database, in order to understand the impact of common RAS events on application performance. This study conducted on the records of about 2 million applications executed on Titan supercomputer provides important insights for system users, operators and computer science researchers. Specifically, we investigate the impact of RAS events on application performance and its variability by comparing cases where events are recorded with corresponding cases where no events are recorded. Such a statistical investigation is possible since we observed that system users tend to execute their applications multiple times. Our analysis reveals that most RAS events do impact application performance, although not always. We also find that different system components affect application performance differently. In particular, our investigation includes the following components: parallel file system processor, memory, graphics processing units, system and user software issues. Our work establishes the importance of providing feedback to system users for increasing operational efficiency of extreme-scale systems.

Ashraf, Rizwan↗

Shear Rolling of Magnesium Sheet for Automotive, Defense, and Energy Applications

Magnesium is the lightest structural metal known: at approximately 1/5 the density of steel, 1/2 the density of titanium and 2/3 the density of aluminum. Hence magnesium alloys represent potential weight savings across the entire transportation industry. The major hurdle to the deployment of magnesium products by the transportation industry is the price barrier that exists due to the current high cost of producing magnesium alloy sheet on a volume basis. Proven technology (e.g., twin roll sheet casting and hot reversing coil mill technology) exists which could lower the cost of magnesium alloy sheet by as much as 50%, but needs to be demonstrated and implemented in high volume production to achieve benefits. In addition, the predominant basal texture (alignment of basal planes parallel to the sheet surface) that exists in magnesium alloy sheet results in poor low temperature formability resulting in added fabrication costs. Cost reduction achieved through energy efficiency, coupled with even greater energy savings by deployment of this “lightest of metals”, will help the United States achieve its goal to eliminate dependence on foreign fossil fuel. Oak Ridge National Laboratory (ORNL), Magnesium Elektron North America (MENA), and FATA Hunter collaborated on this project to develop shear rolling technology of magnesium sheet to enable improvement of the formability of magnesium sheet while addressing cost and lower energy consumption.

Muralidharan, Govindarajan↗

Neutron and X-ray Detectors

The Basic Energy Sciences (BES) X-ray and neutron user facilities attract more than 12,000 researchers each year to perform cutting-edge science at these state-of-the-art sources. While impressive breakthroughs in X-ray and neutron sources give us the powerful illumination needed to peer into the nano- to mesoscale world, a stumbling block continues to be the distinct lag in detector development, which is slowing progress toward data collection and analysis. Urgently needed detector improvements would reveal chemical composition and bonding in 3-D and in real time, allow researchers to watch “movies” of essential life processes as they happen, and make much more efficient use of every X-ray and neutron produced by the source The immense scientific potential that will come from better detectors has triggered worldwide activity in this area. Europe in particular has made impressive strides, outpacing the United States on several fronts. Maintaining a vital U.S. leadership in this key research endeavor will require targeted investments in detector R&D and infrastructure. To clarify the gap between detector development and source advances, and to identify opportunities to maximize the scientific impact of BES user facilities, a workshop on Neutron and X-ray Detectors was held August 1-3, 2012, in Gaithersburg, Maryland. Participants from universities, national laboratories, and commercial organizations from the United States and around the globe participated in plenary sessions, breakout groups, and joint open-discussion summary sessions. Sources have become immensely more powerful and are now brighter (more particles focused onto the sample per second) and more precise (higher spatial, spectral, and temporal resolution). To fully utilize these source advances, detectors must become faster, more efficient, and more discriminating. In supporting the mission of today’s cutting-edge neutron and X-ray sources, the workshop identified six detector research challenges (and two computing hurdles that result from the corresponding increase in data volume) for the detector community to overcome in order to realize the full potential of BES neutron and X-ray facilities. Resolving these detector impediments will improve scientific productivity both by enabling new types of experiments, which will expand the scientific breadth at the X-ray and neutron facilities, and by potentially reducing the beam time required for a given experiment. These research priorities are summarized in the table below. Note that multiple, simultaneous detector improvements are often required to take full advantage of brighter sources. High-efficiency hard X-ray sensors: The fraction of incident particles that are actually detected defines detector efficiency. Silicon, the most common direct-detection X-ray sensor material, is (for typical sensor thicknesses) 100% efficient at 8 keV, 25%efficient at 20 keV, and only 3% efficient at 50 keV. Other materials are needed for hard X-rays. Replacement for 3He for neutron detectors: 3He has long been the neutron detection medium of choice because of its high cross section over a wide neutron energy range for the reaction 3He + n —> 3H + 1H + 0.764 MeV. 3He stockpiles are rapidly dwindling, and what is available can be had only at prohibitively high prices. Doped scintillators hold promise as ways to capture neutrons and convert them into light, although work is needed on brighter, more efficient scintillator solutions. Neutron detectors also require advances in speed and resolution. Fast-framing X-ray detectors: Today’s brighter X-ray sources make time-resolved studies possible. For example, hybrid X-ray pixel detectors, initially developed for particle physics, are becoming fairly mature X-ray detectors, with considerable development in Europe. To truly enable time-resolved studies, higher frame rates and dynamic range are required, and smaller pixel sizes are desirable. High-speed spectroscopic X-ray detectors: Improvements in the readout speed and energy resolution of X-ray detectors are essential to enable chemically sensitive microscopies. Advances would make it possible to take images with simultaneous spatial and chemical information. Very high-energy-resolution X-ray detectors: The energy resolution of semiconductor detectors, while suitable for a wide range of applications, is far less than what can be achieved with X-ray optics. A direct detector that could rival the energy resolution of optics could dramatically improve the efficiency of a multitude of experiments, as experiments are often repeated at a number of different energies. Very high-energy-resolution detectors could make these experiments parallel, rather than serial. Low-background, high-spatial-resolution neutron detectors: Low-background detectors would significantly improve experiments that probe excitations (phonons, spin excitations, rotation, and diffusion in polymers and molecular substances, etc.) in condensed matter. Improved spatial resolution would greatly benefit radiography, tomography, phase-contrast imaging, and holography. Improved acquisition and visualization tools: In the past, with the limited variety of slow detectors, it was straightforward to visualize data as it was being acquired (and adjust experimental conditions accordingly) to create a compact data set that the user could easily transport. As detector complexity and data rates explode, this becomes much more challenging. Three goals were identified as important for coping with the growing data volume from high-speed detectors: Facilitate better algorithm development. In particular, algorithms that can minimize the quantity of data stored. Improve community-driven mechanisms to reduce data protocols and enhance quantitative, interactive visualization tools. Develop and distribute community-developed, detector-specific simulation tools. Aim for parallelization to take advantage of high-performance analysis platforms. Improved analysis work flows: Standardize the format of metadata that accompanies detector data and describes the experimental setup and conditions. Develop a standardized user interface and software framework for analysis and data management. The diversity of detector improvements required is necessarily as broad as the range of scientific experimentation at BES facilities. This workshop identified a variety of avenues by which detector R&D can enable enhanced science at BES facilities. The Research Directions listed above will be addressed by focused R&D and detector engineering, both of which require specialized infrastructure and skills. While U.S. leadership in neutron and X-ray detectors lags behind other countries in several areas, significant talent exists across the complex. A forum of technical experts, facilities management, and BES could be a venue to provide further definition.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Residential Cold Climate Heat Pump (CCHP) w/Variable Speed Technology

This report summarizes the results of a three year program awarded to Unico, Inc. to commercialize a residential cold climate heat pump. Several designs were investigated. Compressors were selected using analysis from Oakridge National Laboratories followed by prototype construction and lab testing in a specially built environmental chamber capable of reaching -30°F. The initial design utilized two variable speed compressors in series with very good capacity results and acceptable efficiency at very cold temperatures. The design was then modified to reduce cost and complexity by redesigning the system using three dual-stage compressors: two in parallel followed by one in series. Extensive testing found significant challenge with oil management, reliability, weight and cost which prevented the system from being fully commercialized. Further analysis of other conceptual designs indicated that these challenges could be overcome in the future.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Transient and Steady-State Inverse Problems in Sierra/Aria

Inverse problems arise in a wide range of applications, whenever unknown model parameters cannot be measured directly. Instead, the unknown parameters are estimated using experimental data and forward simulations. Thermal inverse problems, such as material characterization problems, are often large-scale and transient. Therefore, they require intrusive adjoint-based gradient implementations in order to be solved efficiently. The capability to solve large-scale transient thermal inverse problems using an adjoint-based approach was recently implemented in SNL Sierra Mechanics, a massively parallel capable multiphysics code suite. This report outlines the theory, optimization formulation, and path taken to implement thermal inverse capabilities in Sierra within a unit test framework. The capability utilizes Sierra/Aria and Sierra/Fuego data structures, the Rapid Optimization Library, and an interface to the Sierra/InverseOpt library. The existing Sierra/Aria time integrator is leveraged to implement a time-dependent adjoint solver.

97 MATHEMATICS AND COMPUTING↗

Implementation of a cell-wise Block-Gauss-Seidel iterative method for SN transport on a hybrid parallel computer architecture

We have implemented a cell-wise, block-Gauss-Seidel (bGS) iterative algorithm, for the solution of the S{sub n} transport equations on the Roadrunner hybrid, parallel computer architecture. A compute node of this massively parallel machine comprises AMD Opteron cores that are linked to a Cell Broadband Engine{trademark} (Cell/B.E.). LAPACK routines have been ported to the Cell/B.E. in order to make use of its parallel Synergistic Processing Elements (SPEs). The bGS algorithm is based on the LU factorization and solution of a linear system that couples the fluxes for all S{sub n} angles and energy groups on a mesh cell. For every cell of a mesh that has been parallel decomposed on the higher-level Opteron processors, a linear system is transferred to the Cell/B.E. and the parallel LAPACK routines are used to compute a solution, which is then transferred back to the Opteron, where the rest of the computations for the S{sub n} transport problem take place. Compared to standard parallel machines, a hundred-fold speedup of the bGS was observed on the hybrid Roadrunner architecture. Numerical experiments with strong and weak parallel scaling demonstrate the bGS method is viable and compares favorably to full parallel sweeps (FPS) on two-dimensional, unstructured meshes when it is applied to optically thick, multi-material problems. As expected, however, it is not as efficient as FPS in optically thin problems.

97 MATHEMATICS AND COMPUTING↗

Hybrid algorithms in quantum Monte Carlo

With advances in algorithms and growing computing powers, quantum Monte Carlo (QMC) methods have become a leading contender for high accuracy calculations for the electronic structure of realistic systems. The performance gain on recent HPC systems is largely driven by increasing parallelism: the number of compute cores of a SMP and the number of SMPs have been going up, as the Top500 list attests. However, the available memory as well as the communication and memory bandwidth per element has not kept pace with the increasing parallelism. This severely limits the applicability of QMC and the problem size it can handle. OpenMP/MPI hybrid programming provides applications with simple but effective solutions to overcome efficiency and scalability bottlenecks on large-scale clusters based on multi/many-core SMPs. We discuss the design and implementation of hybrid methods in QMCPACK and analyze its performance on current HPC platforms characterized by various memory and communication hierarchies.

Esler, Kenneth P↗

Influence of Ionization and Beam Quality on Interaction of TW-Peak CO2 Laser with Hydrogen Plasma

3D numerical simulations of the interaction of a powerful CO2 laser with hydrogen jets demonstrating the role of ionization and laser beam quality are presented. Simulations are performed in support of the plasma wakefield accelerator experiments being conducted at the BNL Accelerator Test Facility (ATF). The CO2 laser at BNL ATF has several potential advantages for laser wakefield acceleration compared to widely used solid-state lasers. SPACE, a parallel relativistic Particle-in-Cell code, developed at SBU and BNL, has been used in these studies. A novelty of the code is its set of efficient atomic physics algorithms that compute ionization and recombination rates on the grid and transfer them to particles. The primary goal of the initial BNL experiments was to characterize the plasma density by measuring the sidebands in the spectrum of the probe laser. Simulations, that resolve hydrogen ionization and laser spectra, help explain several trends that were observed in the experiments.

97 MATHEMATICS AND COMPUTING↗

A Case Study of LLVM-Based Analysis for Optimizing SIMD Code Generation

This paper presents a methodology for using LLVM-based tools to tune the DCA++ (dynamical cluster approximation) application that targets the new ARM A64FX processor. The goal is to describe the changes required for the new architecture and generate efficient single instruction/multiple data (SIMD) instructions that target the new Scalable Vector Extension instruction set. During manual tuning, the authors used the LLVM tools to improve code parallelization by using OpenMP SIMD, refactored the code and applied transformation that enabled SIMD optimizations, and ensured that the correct libraries were used to achieve optimal performance. By applying these code changes, code speed was increased by 1.98× and 78 GFlops were achieved on the A64FX processor. The authors aim to automatize parts of the efforts in the OpenMP Advisor tool, which is built on top of existing and newly introduced LLVM tooling.

Huber, Joseph↗

The Optimized Parallel MOC module in Code cosLATC

A new neutron transport lattice physics code, cosLATC, which is an essential part in COSINE (Core and System Integrated Engine) code package, has been developed by SPICRI. It is mainly used for 2-D transport calculations on Pressurized Water Reactor (PWR) assemblies or pin cells. cosLATC generates few group constants for the core physics code cosCORE in COSINE package. With regards to cross-section data, WIMS-D 69 and 281 energy groups library is employed currently. cosLATC also has capability of resonance treatment, depletion and fuel temperature calculation. cosLATC uses modular tracking and parallelization technique in MOC transport calculation, which can save lots of memory and running time. In this paper, MOC module in cosLATC are introduced, the accuracy and efficiency of the module are verified by C5G7 benchmark and a PWR typical assembly. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Enabling parallel simulation of large-scale HPC network systems

Here, with the increasing complexity of today’s high-performance computing (HPC) architectures, simulation has become an indispensable tool for exploring the design space of HPC systems—in particular, networks. In order to make effective design decisions, simulations of these systems must possess the following properties: (1) have high accuracy and fidelity, (2) produce results in a timely manner, and (3) be able to analyze a broad range of network workloads. Most state-of-the-art HPC network simulation frameworks, however, are constrained in one or more of these areas. In this work, we present a simulation framework for modeling two important classes of networks used in today’s IBM and Cray supercomputers: torus and dragonfly networks. We use the Co-Design of Multi-layer Exascale Storage Architecture (CODES) simulation framework to simulate these network topologies at a flit-level detail using the Rensselaer Optimistic Simulation System (ROSS) for parallel discrete-event simulation. Our simulation framework meets all the requirements of a practical network simulation and can assist network designers in design space exploration. First, it uses validated and detailed flit-level network models to provide an accurate and high-fidelity network simulation. Second, instead of relying on serial time-stepped or traditional conservative discrete-event simulations that limit simulation scalability and efficiency, we use the optimistic event-scheduling capability of ROSS to achieve efficient and scalable HPC network simulations on today’s high-performance cluster systems. Third, our models give network designers a choice in simulating a broad range of network workloads, including HPC application workloads using detailed network traces, an ability that is rarely offered in parallel with high-fidelity network simulations

97 MATHEMATICS AND COMPUTING↗

Refining HPCToolkit for application performance analysis at exascale

As part of the US Department of Energy’s Exascale Computing Project (ECP), Rice University has been refining its HPCToolkit performance tools to better support measurement and analysis of applications executing on exascale supercomputers. To efficiently collect performance measurements of GPU-accelerated applications, HPCToolkit employs novel non-blocking data structures to communicate performance measurements between tool threads and application threads. To attribute performance information in detail to source lines, loop nests, and inlined call chains, HPCToolkit performs parallel analysis of large CPU and GPU binaries involved in the execution of an exascale application to rapidly recover mappings between machine instructions and source code. To analyze terabytes of performance measurements gathered during executions at exascale, HPCToolkit employs distributed-memory parallelism, multithreading, sparse data structures, and out-of-core streaming analysis algorithms. To support interactive exploration of profiles up to terabytes in size, HPCToolkit’s hpcviewer graphical user interface uses out-of-core methods to visualize performance data. The result of these efforts is that HPCToolkit now supports collection, analysis, and presentation of profiles and traces of GPU-accelerated applications at exascale. These improvements have enabled HPCToolkit to efficiently measure, analyze and explore terabytes of performance data for executions using as many as 64K MPI ranks and 64K GPU tiles on ORNL’s Frontier supercomputer. HPCToolkit’s support for measurement and analysis of GPU-accelerated applications has been employed to study a collection of open-science applications developed as part of ECP. This paper reports on these experiences, which provided insight into opportunities for tuning applications, strengths and weaknesses of HPCToolkit itself, as well as unexpected behaviors in executions at exascale.

Adhianto, Laksono↗

Hard Disk/Solid State Drive Synergy in Support of Data-Intensive Computing [Slides]

Data-intensive applications are becoming increasingly common in high-performance computing. Examples include combustion simulation, human genome analysis, and satellite image processing. Efficient access of data sets is critical to the performance of these applications. Because of the size of the data today's economically feasible approach is to store the data files on an array of hard disks or data servers equipped with hard disks and managed by a parallel file system such as PVFS or Lustre wherein the data is striped over a (large) number of disks for high aggregate I/O throughout. With file striping, a request for a segment of logically contiguous file space is decomposed into multiple sub-requests, each to a different server. While the data unit for this striping is usually reasonably large to benefit disk efficiency, the first and/or last sub-requests can be much smaller than the striping unit if the request does not align with the striping pattern, severely compromising hard disk efficiency and thus application performance. We propose to exploit solid state drives (SSD), whose efficiency is much less sensitive to small random accesses, to enable the alignment of requests to disk with the data striping pattern. In this scheme hard disks mainly serve large, aligned, sequential requests, with SSDs serving small or unaligned requests, thus respecting the relative cost, performance, and durability characteristics of the two media, and thereby achieving synergy in performance/cost. We will describe the design of the proposed scheme, its implementation on CCS-7's Darwin cluster, and performance results.

97 MATHEMATICS AND COMPUTING↗