Simulation Runtime Research Articles

In silico screening for oligopeptides useful as capture and reporting probes for interleukin-6 biosensing.

IL-6 is an important interleukin associated with inflammation and several diseases such as cancer. Evaluation of its levels in human blood sera is a critical step for an accurate diagnosis of the diseases. Our goal is to design peptides that can selectively bind in different poses with good affinities to IL-6. For this purpose, we started from the crystal structures of different IL-6/protein complexes available in the Protein Data Bank (PDB) to select short peptides in the interaction zones, in which we intentionally introduced point mutations to increase their stability and affinity. To examine their usefulness as capture and reporting probes for the IL-6 biosensing, the five peptides and their interaction with IL-6 were studied in saline aqueous solution. Molecular docking, MD, and MM-PBSA were used to investigate the affinity and stability of these complexes. The conformational changes, the distance between the mass centers, the gyration radii, and the numbers of hydrogen bonds were analyzed to select the most suitable candidates. Three peptides, namely CTE17, CAY15 and CSE25, have the highest affinities presenting significant numbers of residues that have contact frequencies greater than 50% of simulation run time and are the most promising candidates. CTE17 and CSE25 showed they can form a stable sandwich with the target protein. For sake of comparison, we examined the previously known peptides (FND20, INL19 and CEK17) having affinity to IL-6 and the affinity of the lead i.e. CSE25 to two other interleukin family members (IL-4 and to IL-10).

A machine learning approach to predict the stress results of quadratic tetrahedral elements

Industries are always looking for an effective and efficient way to reduce the computation time of simulation because of the huge expenditure involved. From basics of Finite Element Method (FEM), it is known that, linear order finite elements consume less computation time and are less accurate compared to higher order finite elements say quadratic elements. An approach to get the benefit of less computation cost of linear elements and the good accuracy of quadratic elements can be of a good thought. The methodology to get the accurate results of quadratic elements with the advantage of less simulation run time of linear elements is presented here. Machine Learning (ML) algorithms are found to be effective in making predictions based on some known data set. The present paper discusses a methodology to implement ML model to predict the results equivalent to that of quadratic elements based on the solutions obtained from the linear elements. Here, a ML model is developed using python code, the stress results from Finite Element (FE) model of linear tetrahedral elements is given as the input to it to predict the stress results of quadratic tetrahedral elements. Abaqus is used as the FEM tool to develop the FE models. A python script is used to extract the stresses and the corresponding node numbers. The results showed that the developed ML model is successful in prediction of the accurate stress results for the set of test data. The scatter plots showed that the Z-score method was effective in removing the singularities. The proposed methodology is effective to reduce the computation time for simulation.

Model-based autotuning of discretization methods in numerical simulations of partial differential equations

We present an autotuning approach for compile-time optimization of numerical discretization methods in simulations of partial differential equations. Our approach is based on data-driven regression of performance models for numerical methods. We use these models at compile time to automatically determine the parameters (e.g., resolution, time step size, etc.) of numerical simulations of continuum spatio-temporal models in order to optimize the tradeoff between simulation accuracy and runtime. The resulting autotuner is developed for the compiler of a Domain-Specific Language (DSL) for numerical simulations. The abstractions in the DSL enable the compiler to automatically determine the performance models and know which discretization parameters to tune. We demonstrate that this high-level approach can explore a large space of possible simulations, with simulation runtimes spanning multiple orders of magnitude. We evaluate our approach in two test cases: the linear diffusion equation and the nonlinear Gray-Scott reaction–diffusion equation. The results show that our model-based autotuner consistently finds configurations that outperform those found by state-of-the-art general-purpose autotuners. Specifically, our autotuner yields simulations that are on average 4.2x faster than those found by the best generic exploration algorithms, while using 16x less tuning time. Compared to manual tuning by a group of researchers with varying levels of expertise, the autotuner was slower than the best users by not more than a factor of 2, whereas it was able to significantly outperform half of them.

Numerical investigations of 2-D planar magnetic nozzle effects on pulsed plasma plumes

AbstractThis paper presents a study of a novel type of magnetic nozzle that allows for three-dimensional (3-D) steering of a plasma plume. Numerical simulations were performed using Tech-X’s USim® software to quantify the nozzle’s capabilities. A 2-D planar magnetic nozzle was applied to plumes of a nominal pulsed inductive plasma (PIP) source with discharge parameters similar to those of Missouri S&T’s Missouri Plasmoid Experiment (MPX). Argon and xenon plumes were considered. Simulations were verified and validated through a mesh convergence study as well as comparison with available experimental data. Periodicity was achieved over the simulation run time and phase angle samples were taken to examine plume evolution over pulse cycles. The resulting pressure, velocity, and density fields were analysed for nozzle angles from 0° to 14°. It was found that actual plume divergence was small compared to the nozzle angle. Even with an offset angle of 14° for the magnetic nozzle, the plume vector angle was only about 2° for argon and less than 1° for xenon. The parameters that had the most effect on the vectoring angle were found to be the coil current and inlet velocity.

Beam Pointing Optimization Based Downlink Interference Mitigation Technique Between NGSO Satellite Systems

This letter proposes a novel interference mitigation technique for spectral coexistence between large-scale non-geostationary orbit (NGSO) satellite systems by optimizing satellite antenna beam pointing. Firstly, a dynamic exclusion zone of NGSO satellite antenna beam pointing is defined. Secondly, we transfer frequency compatibility problem between NGSO systems into a constrained nonlinear multivariate optimization problem considering the joint effect of link distances as well as angles between links and solve it using Genetic Algorithm (GA). Then we select actual working parameters of Starlink and OneWeb satellite systems to verify the effectiveness of proposed interference mitigation technique in the worldwide. The available time of the interfered system is up to 98.5% of total simulation run time after interference mitigation. The results show that the proposed interference mitigation technique can effectively mitigate the interference between NGSO satellite systems and guarantee signal to interference plus noise ratio (SINR) at the interfered earth station.

Bridging TCAD and AI: Its Application to Semiconductor Design

There is a growing consensus that the physics-based model needs to be coupled with machine learning (ML) model relying on data or vice versa in order to fully exploit their combined strengths to address scientific or engineering problems that cannot be solved separately. We propose several methodologies of bridging technology computer-aided design (TCAD) simulation and artificial intelligence (AI) with its application to the tasks for which traditional TCAD faces challenges in terms of simulation runtime, coverage, and so on. AI-emulator that learns fine-grained information from rigorous TCAD enables simulation of process technologies and device in real-time as well as large-scale simulation such as full-pattern analysis of stress without high demand on computational resource. To accelerate atomistic molecular dynamics (MD) simulation, we have done a comparison study of descriptor-based and graph-based neural net potential, and also show their capability with large-scale and long-time simulation of silicon oxidation. Finally, we discuss the use of hybrid modeling of AI- and physics-based model for the case where physical equations are either fully or partially unknown.

Hybrid strategy for routing, modulation and spectrum assignment in elastic optical networks

In this paper, we propose a novel hybrid strategy that combines mixed integer linear programming (MILP) formulation with different alternative routing approaches that is capable of solving, simultaneously, the routing, modulation format and spectrum assignment problem in the design of elastic optical networks. We extend our proposal to a path-link formulation that, although it is not guaranteed to find the optimal solution, it has shown reasonable solutions within inferior simulation time, allowing it to be used in not very short networks. We also compare our proposal with another MILP formulation and three adapted heuristics, all available in the literature. The results show the benefits of our proposal considering diverse simulation scenarios and different number of modulation formats over 19 realistic networks in terms of simulation run time and maximum number of used set of slots.

A simplified model to assess the thermal performance of pavement solar collectors

The assessment of thermal performance and efficiency of the Pavement Solar Collector (PSC) systems is essential in designing such systems. In this paper, a simplified simulation model is developed to predict PSC's performance at the design stage. Due to its simplicity, accuracy, and extremely quick simulation run-time, the model output can be linked with other thermal models and renewable heat sources to apply real-time changes in the PSC systems and used in the automation of a PID-controller. The output results show good compatibility between the predicted and simulated data, where the relative error values for model parametrization and validation are less than 0.21% and 0.5%. In terms of computational cost, a yearly long-term performance run time of the present model was simply 3 s (without parametrization), 300,000 times faster than a FEM model for only one month. A systematic sensitivity analysis on different design parameters shows that the efficiency index increases with an increase in pipe spacing, pipe embedment depth, and flow rate, while it decreases with an increase in the asphalt thermal conductivity in the wintertime. Also, the variation of the heat transfer coefficient UA* and correlation parameter k in different PSC configurations was established.

Inclusion of Narrow Flow Paths between Buildings in Coarser Grids for Urban Flood Modeling: Virtual Surface Links

Urban flood modeling benefits from new remote sensing technologies, which provide high-resolution data and allow the consideration of small-scale urban key features. Since high-resolution data often result in large simulation runtimes, coarsening of the 2D grid via resampling techniques can be used to achieve a good balance between accuracy and computation time. However, the representation of urban features and topographical properties degrades, since small-scale features are blurred. Therefore, narrow flow paths between buildings are often not considered, building’s sizes are overestimated, and their arrangement in the grid changes. Thus, flow paths change and waterways are blocked, leading to incorrect inundations around buildings. This paper develops a method to improve the simulation results of coarser grids by adding virtual surface links (VSL) between buildings. The VSL mimic the flow paths of a high-resolution model in the areas of interest. The approach is developed for dual-drainage 1D/2D models. The approach shows a visible improvement at the localized level where the VSL are applied, in terms of under/overestimating flooding and a moderate overall improvement of the simulation results. Relatively to the model resolution of 2 m, the computational time, by applying this method, is reduced by 93.6% when using a 5 m grid and by 99% when using a 10 m grid. For a small test case, where the local effects are investigated, the error in the maximum water volume, relative to a grid size of 2 m, is reduced from 69.63% to 5.03% by using a 5 m grid and from 152.75% to 22.92% for a 10 m grid.

Fast and differentiable simulation of driven quantum systems

The controls enacting logical operations on quantum systems are described by time-dependent Hamiltonians that often include rapid oscillations. In order to accurately capture the resulting time dynamics in numerical simulations, a very small integration time step is required, which can severely impact the simulation run time. Here, we introduce a semianalytic method based on the Dyson expansion that allows us to time-evolve driven quantum systems much faster than standard numerical integrators. This solver, which we name Dysolve, efficiently captures the effect of the highly oscillatory terms in the system Hamiltonian, significantly reducing the simulation's run time as well as its sensitivity to the time-step size. Furthermore, this solver provides the exact derivative of the time-evolution operator with respect to the drive amplitudes. This key feature allows for optimal control in the limit of strong drives and goes beyond common pulse-optimization approaches that rely on rotating-wave approximations. As an illustration of our method, we show results of the optimization of a two-qubit gate using transmon qubits in the circuit QED architecture.

Building the road network for city-scale active transport simulation models

City-scale simulation modelling of active modes of transportation (i.e., walking and cycling) is becoming increasingly popular in recent years. The heterogeneous and complex behaviour of these transportation modes, however, indicates the need for a shift from the traditional car and public transport centred modelling approaches towards incorporating the requirements for walking and cycling behaviour, while maintaining the run-time efficiency of the models. In this paper, we introduce and test our algorithm to create road network representations, designed and optimised to be used in city-scale active transportation modelling. The algorithm relies on open and universal data. In addition to the major roads and attributes typically used in transport modelling (e.g., speed limit, number of lanes, permitted travel modes), the algorithm also captures minor roads usually favoured by pedestrians and cyclists, along with road attributes such as bicycle-specific infrastructure, traffic signals, road gradient and road surface type. Furthermore, it simplifies the complex geometries of the network and merges parallel roads, if applicable, to make it suitable for large-scale simulations.To examine the utility and performance of the algorithm, we used it to create a network representation for Greater Melbourne, Australia, and compared the output with a network created using an existing simulation toolkit along with another network from an existing city-scale transport model from the Victorian government. Through simulation experiments with these networks, we illustrated that for routed trips on our network for walking and cycling, it is of comparable accuracy to the common network conversion tools in terms of travel distance of the shortest paths while being more than two times faster when used for simulating different sample sizes. Therefore, our algorithm offers a flexible and adjustable solution for users to create road networks for city-scale active transport modelling while balancing between their desired simulation accuracy and run-time.

Methods—PETLION: Open-Source Software for Millisecond-Scale Porous Electrode Theory-Based Lithium-Ion Battery Simulations

This article presents PETLION, which is an open-source, high-performance computing implementation of the porous electrode theory (PET) model in Julia. A typical runtime for a dynamic simulation of full charge or discharge is 3 ms on a laptop while allocating about 1 MB of total memory, and the software is seen to be two orders of magnitude faster than comparable software for some applications. At moderate spatial resolutions, the computation times are similar to those of reduced-order and reformulated models in the literature. Multiple numerical solvers and methods for their initialization are compared in terms of numerical convergence and computational times, for a wide variety of operating conditions. PETLION is shown to quickly and robustly simulate complex battery protocols such as the Galvanostatic Intermittent Titration Technique (GITT), and to achieve high performance when incorporated into real-time PET-based nonlinear model predictive control.

Image semantic segmentation method based on GAN network and ENet model

Currently, image semantic segmentation has problems such as low accuracy and long running time. This paper proposes an image semantic segmentation method based on generative adversarial network and ENet model combined with deep neural network. This method first improves the network model of generative adversarial network. Ensure the high resolution of the generated image and achieve high similarity with the real image. While ensuring the high accuracy of image semantic segmentation, it effectively improves the real-time performance of network processing. The proposed method is verified based on public data sets. The experimental results show that the segmentation accuracy of this method can reach more than 93%, and the simulation running time is less than 0.171 s, which shows good high accuracy and high real-time performance. A feasible strategy is proposed for the further productization of semantic segmentation.

Multi-Dimensional evacuation risk evaluation in standard subway station

The performance-based approach using Required Safety Egress Time / Available Safety Egress Time concept is the most common method in fire safety evaluation. However, this one-dimension approach cannot fully evaluate the performance of a complex scenario with thousands of occupants in subway stations. Therefore, this paper develops a four-dimension parameter system that includes Required Safety Egress Time, Average Evacuation Time, Average Waiting Time, and Average Moving Distance to quantify the evacuation performance from four aspects. The convergence test approach is adopted to validate the accuracy of simulation results. A dimensionless parameter Risk Index (RI) is proposed to support multi-dimensional risk evaluation and comparison. One realistic standard subway station in Guangzhou metro system is modeled to testify the applicability of the proposed method. Totally twelve evacuation scenarios are designed and simulated, results indicate that: 1) Repeated simulations are necessary. Up to 629 simulation run times are required to meet the acceptance criterion of 0.01% for mean and 1% for standard deviation. 2) The Risk Index of Average Waiting Time is distinctly higher than other Risk Indexes. A long waiting time at bottlenecks increases the risk level in all scenarios. 3) The risk evaluation result of comprehensive RI is consistent with but not the same as the evaluation conclusion of RSET. The difference between RI and RSET shows that RI integrates more aspects of evacuation risk than RSET. The method proposed could support safety evaluation capacity improvement and standard revision in metro system.

Model Input and Optimization: Improving the Speed and Accuracy of Our COVID-19 Hospitalization Forecasts

Background: During the COVID-19 pandemic, public-health decision makers have increasingly relied on hospitalization forecasts that are routinely provided, accurate, and based on timely input data to inform pandemic planning. In North Carolina, we adapted an existing agent-based model (ABM) to produce 30-day hospitalization forecasts of COVID-19 and non–COVID-19 hospitalizations for use by public-health decision makers. We sought to continually improve model speed and accuracy during forecasting. Methods: The geospatially explicit ABM included movement of agents (ie, patients) among 104 short-term acute-care hospitals, 10 long-term acute-care hospitals, 421 licensed nursing homes, and the community in North Carolina. Agents were based on a synthetic population of North Carolina residents (ie, >10.4 million agents). We assigned SARS-CoV-2 infections to agents according to county-level susceptible, exposed, infectious, recovered (SEIR) models informed by reported COVID-19 cases by county. Agents’ COVID-19 severity and probability of hospitalization were determined using agent-specific characteristics (eg, age, comorbidities). During May 2020–December 2020, we produced weekly 30-day forecasts of intensive care unit (ICU) and non-ICU bed occupancy for COVID-19 agents and non–COVID-19 agents statewide and by region under a range of SARS-CoV-2 effective reproduction numbers. During the reporting period, we identified optimizations for faster results turnaround. We evaluated the incorporation of real-time hospital-level occupancy data at model initialization on forecast accuracy using mean absolute percent error (MAPE). Results: During May 2020–December 2020, we provided 31 weekly reports of 30-day hospitalization forecasts with a 1-day turnaround time. Reports included (1) raw and smoothed 7-day average values for 42 model output variables; (2) static visuals of ICU and non-ICU bed demand and capacity; and (3) an interactive Tableau workbook of hospital demand variables. Identifying code efficiencies reduced a single model runtime from ~100 seconds to 28 seconds. The use of cloud computing reduced simulation runtime from ~20 hours to 15 minutes. Across forecasts, the average MAPEs were 21.6% and 7.1% for ICU and non-ICU bed demand, respectively. By incorporating hospital-level occupancy data, we reduced the average MAPE to 6.5% for ICU bed demand and 3.9% for non-ICU bed demand, indicating improved accuracy. Conclusions: We adapted an ABM and continually improved it during COVID-19 forecasting by optimizing code and computing resources and including real-time hospital-level occupancy data. Planned SEIR model updates for enhanced forecasts include the addition of compartments for undocumented infections and recoveries as well as permission of reinfection from recovered compartments.Funding: NoDisclosures: None

Efficient, High-Fidelity Modeling and Simulation of Global Navigation Satellite System Constellations

This paper presents a method for modeling and simulation of Global Navigation Satellite System (GNSS) constellations by the use of Chebyshev polynomials fit to publicly available precision data. The Global Positioning System (GPS) is used to demonstrate the method, but the method applies to all GNSS with precise ephemeris data including Galileo and GLONASS. The method facilitates highly accurate satellite orbit and clock modeling, including relativistic effects, with mean fit differences of the order in position, in velocity, and microseconds in clock offset. The method also improves simulation run time by using a fast technique for evaluating the Chebyshev polynomials. Runtime comparisons demonstrate the polynomial method reduces runtime by up to 72% of more traditional methods. The speed of the method makes it well suited for practical, accurate Monte Carlo simulation techniques. The method can be used in simulations for the verification and validation of GNSS receivers embedded in autonomous aerospace navigation systems, for uncertainty analysis, and for other purposes. Details for validating the precision data are given. Pseudocode algorithms are provided for fitting and evaluating Chebyshev polynomials. Also provided are high-level concepts of one approach for integrating the method into larger, multisystem aerospace simulations, including the effects of signal propagation delay and Earth blockage.

2D particle-in-cell simulations of the electron temperature anisotropy driven whistler instability in plasmas having kappa velocity distributions

This study presents the results of 2D particle-in-cell (PIC) simulations of the electron temperature anisotropy driven whistler instability for plasmas in which the electron species is modeled by a bi-kappa velocity distribution. These simulations utilize our previously developed method to generate the initial multi-dimensional kappa velocity distributions. The use of multi-dimensional kappa loadings in PIC simulations provides insights into the non-linear regime of wave evolution in plasmas having non-equilibrium velocity distributions. Three cases are considered, corresponding to κe=2,3, and ∞ (Maxwellian case). Owing to the use of a large value of electron anisotropy required for reasonable simulation run times, the Maxwellian electron run has the fastest growth rate, reaching saturation earliest. The κe=2 case exhibits the slowest growth rate. Spectral analysis of the fluctuating fields reveals considerable wave intensity at frequencies and wavenumbers that satisfy the linear whistler wave dispersion relation. In the runs with kappa distributions, the regions of most intense fluctuations comprise frequencies and wavenumbers that agree only qualitatively with linear whistler wave theory. The results suggest that after saturation, there is some degree of Landau damping of the oblique whistler modes, which returns energy to the electron species. The rate of damping of the oblique modes is highest in the Maxwellian case and lowest for κe=2. Evidence of significant superthermal acceleration of electrons in the direction parallel to the ambient magnetic field is also observed. Thus, the power-law index of the electron distribution is reduced by the anisotropic whistler turbulence produced.

Modelling Casuarina junghuhniana dispersal in Tengger sea of sands of Bromo Tengger Semeru National Park using cellular automata

The growing of population of a pioneer species may show a unique behavior. Casuarina junghuhniana is reported as a pioneer species, this is because of its tolerance to adverse environment and transform it to become inhabitable for other organism. A comprehensive method involving field observation, the use of mathematical model and computation to understand plants behavior still needs to be developed. This paper is attempting to show a method to build a model for showing unique dispersal behavior of Casuarina junghuhniana. A relatively straightforward use of Fisher’s reaction diffusion equation, cellular automata based modeling and programming can show behavior easier to understand on a grid space. Some assumption must be used to represent the C. junghuhniana dispersal. As it is known that C. junghuhniana can reproduce both vegetatively and generatively, and both were assumed to have different dispersal behavior. Vegetative reproduction using shoots emerging from roots has pattern to spread close to each other. While generative reproduction by seeds which is lightweight has possibility spread further with the help of wind. The total model area of 1km 2 was divided into 300 × 300 cells which represents 1 km 2 area. Cellular automata apply the same rule to each cell and the result was observed from the pattern of cells interaction. A simple program was developed using Python with Jupyter Notebook an interactive computing environment. The number of iteration was limited before observed simulation reached one of area boundaries. The simulation running time was greatly reduced using Graphical Processing Unit (GPU) parallel processing. NUMBA was used as an interface between python and Compute Unified Device Architecture (CUDA), the GPU programming API. The simulation showed behavior of the trees spread downward along the slope. The wind direction simulation with additional data from wind rose available from internet was also demonstrated in the simulation. Modeling dispersal with influence from wind showed more sporadic pattern at further distance from originating parents cluster. The trees in the simulation showed clusters in dispersal pattern when rely more on generative reproduction. When the trees rely on vegetative reproduction by roots shoots, the population showed more like continuous cover. When concerning real 3 dimensional aspect, this simulation still can not distinguish vertical (z axis) aspect. It means two points would be seen as close to each other although they actually have far vertical distance. At this stage, the simulation has not yet been optimized to easily convert input from the wind rose. The wind information still needs to be written as code into the program.

Machine learning for accelerating 2D flood models: Potential and challenges

AbstractTwo‐dimensional hydrodynamic models numerically solve full Shallow Water Equations (SWEs). Despite their high accuracy, these models have long simulation run times and therefore are of limited use for exploratory or real‐time flood predictions. We investigated the possibility of improving flood modelling speed using Machine Learning (ML). We propose a new method that replaces the computationally expensive parts of the hydrodynamic models with simple and efficient data‐driven approximations. Our hypothesis is that by integrating ML with physics‐based numerical methods, we can achieve improved generalization performance: that is, the trained model for one case study can be used in other studies without the need for new training. We tested two ML approaches: for the first, we integrated curve fitting, and, for the second, artificial neural networks (ANN) with a finite volume scheme to solve the local inertial approximation of the SWEs. The data‐driven models approximated the Momentum Equation, which explicitly solved the time derivative of flow rates. Water depths were then updated by applying a water balance equation. We also tested two different training datasets: the simulated dataset, generated from the results of hydrodynamic model, and the random dataset, generated by directly solving the momentum equation on randomly sampled input data. Various combinations of input features, for example, water slope and depth, were explored. The proposed models were trained in a small hypothetical case and tested in a different hypothetical and in two real case studies. Results showed that the curve‐fitting method can be implemented successfully, given sufficient training and input data. The ANN model trained with a random dataset was substantially more accurate than that of the model trained with the simulated dataset. However, it was not successful in the real case studies. The curve‐fitting method resulted in better generalization performance and increased the simulation speed of the local inertial model by 23%. Future research should test the performance of ML in terms of an increase in stable time step size and approximation of the full SWEs.

In Situ Visualization With Temporal Caching

In situ visualization is a technique in which plots and other visual analyses are performed in tandem with numerical simulation processes in order to better utilize HPC machine resources. Especially with unattended exploratory engineering simulation analyses, events may occur during the run, which justify supplemental processing. Sometimes though, when the events do occur, the phenomena of interest includes the physics that precipitated the events and this may be the key insight into understanding the phenomena that is being simulated. In situ temporal caching is the temporary storing of produced data in memory for possible later analysis including time varying visualization. The later analysis and visualization still occurs during the simulation run but not until after the significant events have been detected. In this article, we demonstrate how temporal caching can be used with in-line in situ visualization to reduce simulation run-time while still capturing essential simulation results.