SEARCH · Engineering Papers

Results for “Adaptation models”

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.

SourceResearch connectionFrom yearThrough year

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

At least 991 records · Page 55

All sources NASA NTRS DOE OSTI

DOE OSTI2024

Foundations of Off-Grid Solar in Haiti: 8. Financial Modeling for Off-Grid Solar

This training lends insight into key information and concepts for off-grid solar in Haiti, building foundational knowledge and an ability to catalyze interest in off-grid solar for rural electrification in Haiti. This self-paced course is offered in both English and French and covers a variety of topics related to energy access in Haiti including off-grid solar products, market potential in Haiti, supply and demand side considerations, system design, installation and maintenance, off-grid solar business models, financial modeling, gender and energy access, productive use of energy, and climate adaptation. The training is intended for university students, government partners, NGOs, funders, development partners, professionals, and more. No prior knowledge of off-grid solar or Haiti is required to benefit from this training. See NREL/PR-7A40-89262 for the French translation of this document.

ENERGY PLANNING, POLICY, AND ECONOMY,SOLAR ENERGY↗

DOE OSTI2023

Understanding local plant extinctions before it is too late: bridging evolutionary genomics with global ecology

Understanding evolutionary genomic and population processes within a species range is key to anticipating the extinction of plant species before it is too late. However, most models of biodiversity risk under global change do not account for the genetic variation and local adaptation of different populations. Population diversity is critical to understanding extinction because different populations may be more or less susceptible to global change and, if lost, would reduce the total diversity within a species. Here, two new modeling frameworks advance our understanding of extinction from a population and evolutionary angle: Rapid climate change-driven disruptions in population adaptation are predicted from associations between genomes and local climates. Furthermore, losses of population diversity from global land-use transformations are estimated by scaling relationships of species' genomic diversity with habitat area. Overall, these global eco-evolutionary methods advance the predictability – and possibly the preventability – of the ongoing extinction of plant species.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2020

Adaptive Hybridizable Discontinuous Galerkin discretization of the Grad–Shafranov equation by extension from polygonal subdomains

In this work, we propose a high-order adaptive numerical solver for the semilinear elliptic boundary value problem modeling magnetic plasma equilibrium in axisymmetric confinement devices. In the fixed boundary case, the equation is posed on curved domains with piecewise smooth curved boundaries that may present corners. The solution method we present is based on the hybridizable discontinuous Galerkin method and sidesteps the need for geometry-conforming triangulations thanks to a transfer technique that allows to approximate the solution using only a polygonal subset as computational domain. Moreover, the solver features automatic mesh refinement driven by a residual-based a posteriori error estimator. As the mesh is locally refined, the computational domain is automatically updated in order to always maintain the distance between the actual boundary and the computational boundary of the order of the local mesh diameter. Numerical evidence is presented of the suitability of the estimator as an approximate error measure for physically relevant equilibria with pressure pedestals, internal transport barriers, and current holes on realistic geometries.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2024

Foundations of Off-Grid Solar in Haiti: 5. Designing and Modeling Off-Grid Solar Systems

designing↗

DOE OSTI2025

Adaptive Interface-PINNs (AdaI-PINNs): An Efficient Physics-Informed Neural Networks Framework for Interface Problems

Here, we present an efficient physics-informed neural networks (PINNs) framework, termed Adaptive Interface-PINNs (AdaI-PINNs), to improve the modeling of interface problems with discontinuous coefficients and/or interfacial jumps. This framework is an enhanced version of its predecessor, Interface PINNs or I-PINNs (Sarma et al.; https://doi.org/10.1016/j.cma.2024.117135), which involves domain decomposition and assignment of different predefined activation functions to the neural networks in each subdomain across a sharp interface, while keeping all other parameters of the neural networks identical. In AdaI-PINNs, the activation functions vary solely in their slopes, which are trained along with the other parameters of the neural networks. This makes the AdaI-PINNs framework fully automated without requiring preset activation functions. Comparative studies on one-dimensional, two-dimensional, and three-dimensional benchmark elliptic interface problems reveal that AdaI-PINNs outperform I-PINNs, reducing computational costs by 2-6 times while producing similar or better accuracy.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

Adapting the Electron Beam from SEM as a Quantitative Heating Source for Nanoscale Thermal Metrology

The electron beam (e-beam) in the scanning electron microscopy (SEM) provides an appealing mobile heating source for thermal metrology with spatial resolution of ~1 nm, but the lack of systematic quantification of the e-beam heating power limits such application development. Here, we systemically study e-beam heating in LPCVD silicon nitride (SiN x ) thin-films with thickness ranging from 200 to 500 nm from both experiments and complementary Monte Carlo simulations using the CASINO software package. There is good agreement about the thickness-dependent e-beam energy absorption of thin-film between modeling predictions and experiments. Using the absorption results, we then demonstrate adapting the e-beam as a quantitative heating source by measuring the thickness-dependent thermal conductivity of SiN x thin-films, with the results validated to within 7% by a separate Joule heating experiment. The results described here will open a new avenue for using SEM e-beams as a mobile heating source for advanced nanoscale thermal metrology development.

42 ENGINEERING↗

DOE OSTI2025

Accelerating phase field simulations through a hybrid adaptive Fourier neural operator with U-net backbone

Prolonged contact between a corrosive liquid and metal alloys can cause progressive dealloying. For one such process as liquid-metal dealloying (LMD), phase field models have been developed to understand the mechanisms leading to complex morphologies. However, the LMD governing equations in these models often involve coupled non-linear partial differential equations (PDE), which are challenging to solve numerically. In particular, numerical stiffness in the PDEs requires an extremely refined time step size (on the order of 10 -12 s or smaller). This computational bottleneck is especially problematic when running LMD simulation until a late time horizon is required. This motivates the development of surrogate models capable of leaping forward in time, by skipping several consecutive time steps at-once. In this paper, we propose a U-shaped adaptive Fourier neural operator (U-AFNO), a machine learning (ML) based model inspired by recent advances in neural operator learning. U-AFNO employs U-Nets for extracting and reconstructing local features within the physical fields, and passes the latent space through a vision transformer (ViT) implemented in the Fourier space (AFNO). We use U-AFNOs to learn the dynamics of mapping the field at a current time step into a later time step. We also identify global quantities of interest (QoI) describing the corrosion process (e.g., the deformation of the liquid-metal interface, lost metal, etc.) and show that our proposed U-AFNO model is able to accurately predict the field dynamics, in spite of the chaotic nature of LMD. Most notably, our model reproduces the key microstructure statistics and QoIs with a level of accuracy on par with the high-fidelity numerical solver, while achieving a significant 11, 200 × speed-up on a high-resolution grid when comparing the computational expense per time step. Finally, we also investigate the opportunity of using hybrid simulations, in which we alternate forward leaps in time using the U-AFNO with high-fidelity time stepping. We demonstrate that while advantageous for some surrogate model design choices, our proposed U-AFNO model in fully auto-regressive settings consistently outperforms hybrid schemes.

36 MATERIALS SCIENCE↗

DOE OSTI2022

Plasma Wall Interaction with 3-D Plasma Boundaries

The interaction of the edge plasma and the material surfaces is one of the most critical challenges on the path to harness fusion power as new, fundamental energy source. This challenge typically combines the thrust to reach high density, low temperature (detached) plasmas in front of the divertor target plates as well as understanding the plasma material interaction (PMI) in particular in this regime. The combination of both research thrusts represents an extraordinarily challenging subject encompassing spatial scales spanning nanometers to meters in all states of matter and across a broad energy range. Modeling capabilities, which help to interpret data from nowadays experiments and enable extrapolation to future devices are urgently required. This is in particular true for toroidal magnetic confinement devices with three-dimensional (3D) plasma boundaries. Such plasma boundary geometries occur in tokamaks, when small amplitude magnetic perturbations are used to stabilize the unruly edge plasma or in stellarators, that are inherently 3D plasma confinement devices. In this project, the impact of 3D plasma boundaries on the plasma material interaction (PMI) was assessed. This work focused on plasma boundary conditions, in which high-density conditions at the material surfaces yield mitigation of the otherwise immense heat and particle loads that these materials would see. These so-called high recycling and eventually detached plasma regimes are of great interest for future reactor operation. In the project, key features that are unique to 3D boundaries were explored in comparison to canonically assumed axisymmetric plasma edge situations in tokamaks. In particular, the relevance of the 3D boundary situation in the extrapolation to the plasma boundary solution at ITER, the next step fusion energy experiment under construction as a multi-national, world-wide large-science experiment in southern France, has been explored. The EMC3-EIRENE plasma edge fluid and kinetic neutral transport code has been advanced to cope with the challenging and unprecedented conditions in the ITER boundary plasma including 3D fields that are planned to be used to suppress harmful edge instabilities, the so-called edge localized modes. This is a vital integration challenge for ITER and the results from this grant have provide a leading capability for this assessment. It was shown that the detachment process in a 3D edge solution for ITER follows the recycling regimes that are known from axisymmetric solutions, but that multiple plasma exhaust channels connected to the material surfaces are established which feature individual recycling characteristics. Because these channels touch the material surfaces in the divertor in a 3D geometry, the compatibility with the plasma material interaction (PMI), including erosion and impurity generation has been found to be an important part of the integration challenge. To address this, the fully 3D plasma material interaction code ERO2 has been adapted to these ITER specific geometries and a homogeneous mixing model was implemented, that allows to consider the mixing of Be and Was used at ITER in the PMI modeling. This model enhancement has been used to study non-local migration of Be in the JET ITER like wall configuration and it has been shown that with this model such complex migration processes in ITER relevant plasma shapes and with ITER relevant plasma boundary conditions can be addressed. The combined modeling approach using EMC3-EIRENE as a plasma boundary transport code and the ERO2 specialized PMI model will be an asset for the continued preparations of ITER operation as well as for Fusion Pilot Plant efforts that have emerged in the U.S. during the evolution of this grant. The predictive capability of this numerical tool has been validated at the DIII-D US national fusion facility. Here, dedicated plasma edge diagnostics were implemented to measure the impurity household around a 3D edge plasma during ELM suppression by 3D fields. Dedicated experiments with local material probes using these diagnostics and the state-of-the-art suite of boundary measurements at DIII-D have shown that the 3D perturbation of the plasma edge that is excreted by such 3D control fields yield a perturbation of the plasma boundary flux structure and hence also of the resulting PMI. The 3D boundary plasma is composed out of helical magnetic flux channels that intersect the divertor targets at an angle relative to the main guiding field, i.e., the toroidal magnetic field component of the tokamak. A similar effect has been measured as well on limiter surfaces during the startup campaign at the new stellarator experiment Wendelstein 7-X. These experiments ad initial analysis with the ERO plasma material interaction model, suggested that the place of erosion for a given particle from the surface and its re-deposition can be different in such 3D field geometries yielding potentially a significant level of net-erosion. This is not the case for axisymmetric solutions, where it was shown in the past that the eroded particles are effectively re-deposited into gaps produced by erosion at the same position and hence the net-erosion levels are small. For ITER, the quest to suppress the ELMs and at the same time maintain the integrity of the divertor is an issue, which these fundamental findings will help to resolve. The coupling of this work to the extrapolation in the ITER program has been addressed by both the PI and the lead numerical scientist being ITER Science Fellows in the duration of the contract and forward. A second focus in the exploration of 3D boundary effects on tokamaks and stellarators has been set on the measurement of helium exhaust features with such 3D fields. This is important because He represents the ash of the fusion process and needs to be exhausted. It was shown that 3D field application compatible with suppression of ELMs yields an increase of the helium exhaust performance. The ratio of the effective helium confinement time over the energy confinement time was reduced by almost 50% which demonstrated that the impact of helium accumulation in the plasma core with respect to the confinement of energy to sustain the fusion reaction is significantly improved with such 3D control fields. It was shown that this is the case for tokamaks as well as stellarators. At the Large helical Device in Japan, a similar enhancement of the helium exhaust features when small amplitude additional 3D fields were applied was measured. This is an important additional function of 3D field application and its impact on ITER is presently being studied in combination with investigations of helium exhaust in 3D field geometries of stellarator devices.

3D plasma edge transport↗

DOE OSTI2021

Accelerated basis adaptation in homogeneous chaos spaces

Polynomial chaos expansions (PCE) provide an efficient approach to uncertainty quantification (UQ) and have been adapted to diverse applications across the spectrum of science and engineering. For situations involving large stochastic parameterizations, the curse of dimensionality renders PCE-based methods computationally prohibitive. A basis adaptation approach for PCE was proposed by Tipireddy and Ghanem (2014) which transforms the input random variables through an isometry such that sufficient probabilistic characterization of specific quantities of interest (QoI) is concentrated in an algebraic manifold embedded in the linear span of the dominant transformed variables. While quite versatile, that original version of the PCE basis adaptation exhibited slow convergence for a number of problems of practical significance. In the present paper, we propose two novel methods to accelerate the convergence of the original basis adaptation approach, thus expanding its reach while also providing insight into its performance. In the first method, information gained from a pilot PCE representation is used to correct the mean and Gaussian coefficients in the adapted space. By taking advantages of probabilistic information in higher dimensional adaptation gleaned from an initial adaptation, the second method updates the rotation matrix used to identify the dominant transformed variables. In this manner, the new rotation matrix concentrates even more probabilistic information in its first few dimensions. These two method can be combined to achieve even better performance, the combined method is referred to as sequentially optimized adaptation method. Here, the methods are demonstrated on an analytical test function and a model of a space structure with several sub-components and a non-smooth quantity of interest representing the maximum acceleration over time. Both methods achieve accelerated convergence of the basis adaptation approach with negligible additional costs.

Accelerated convergence↗

DOE OSTI2025

The Importance of Being Adaptable: An Exploration of the Power and Limitations of Domain Adaptation for Simulation-Based Inference with Galaxy Clusters

The application of deep machine learning methods in astronomy has exploded in the last decade, with new models showing remarkably improved performance on benchmark tasks. Not nearly enough attention is given to understanding the models' robustness, especially when the test data are systematically different from the training data, or "out of domain." Domain shift poses a significant challenge for simulation-based inference, where models are trained on simulated data but applied to real observational data. In this paper, we explore domain shift and test domain adaptation methods for a specific scientific case: simulation-based inference for estimating galaxy cluster masses from X-ray profiles. We build datasets to mimic simulation-based inference: a training set from the Magneticum simulation, a scatter-augmented training set to capture uncertainties in scaling relations, and a test set derived from the IllustrisTNG simulation. We demonstrate that the Test Set is out of domain in subtle ways that would be difficult to detect without careful analysis. We apply three deep learning methods: a standard neural network (NN), a neural network trained on the scatter-augmented input catalogs, and a Deep Reconstruction-Regression Network (DRRN), a semi-supervised deep model engineered to address domain shift. Although the NN improves results by 17% in the Training Data, it performs 40% worse on the out-of-domain Test Set. Surprisingly, the Scatter-Augmented Neural Network (SANN) performs similarly. While the DRRN is successful in mapping the training and Test Data onto the same latent space, it consistently underperforms compared to a straightforward Yx scaling relation. These results serve as a warning that simulation-based inference must be handled with extreme care, as subtle differences between training simulations and observational data can lead to unforeseen biases creeping into the results.

Ntampaka, Michelle [Baltimore, Space Telescope Sci↗

DOE OSTI2024

Analyzing inference workloads for spatiotemporal modeling

Ensuring power grid resiliency, forecasting climate conditions, and optimization of transportation infrastructure are some of the many application areas where data is collected in both space and time. Spatiotemporal modeling is about modeling those patterns for forecasting future trends and carrying out critical decision-making by leveraging machine learning/deep learning. Once trained offline, field deployment of trained models for near real-time inference could be challenging because performance can vary significantly depending on the environment, available compute resources and tolerance to ambiguity in results. Users deploying spatiotemporal models for solving complex problems can benefit from analytical studies considering a plethora of system adaptations to understand the associated performance-quality trade-offs. To facilitate the co-design of next-generation hardware architectures for field deployment of trained models, it is critical to characterize the workloads of these deep learning (DL) applications during inference and assess their computational patterns at different levels of the execution stack. In this paper, we develop several variants of deep learning applications that use spatiotemporal data from dynamical systems. We study the associated computational patterns for inference workloads at different levels, considering relevant models (Long short-term Memory, Convolutional Neural Network and Spatio-Temporal Graph Convolution Network), DL frameworks (Tensorflow and PyTorch), precision (FP16, FP32, AMP, INT16 and INT8), inference runtime (ONNX and AI Template), post-training quantization (TensorRT) and platforms (Nvidia DGX A100 and Sambanova SN10 RDU). Overall, our findings indicate that although there is potential in mixed-precision models and post-training quantization for spatiotemporal modeling, extracting efficiency from contemporary GPU systems might be challenging. Instead, co-designing custom accelerators by leveraging optimized High Level Synthesis frameworks (such as SODA High-Level Synthesizer for customized FPGA/ASIC targets) can make workload-specific adjustments to enhance the efficiency.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Adaptive changes in sexual behavior in the high–risk population in response to human monkeypox transmission in Canada can help control the outbreak: Insights from a two–group, two–route epidemic model

Monkeypox, a zoonotic disease, is emerging as a potential sexually transmitted infection/disease, with underlying transmission mechanisms still unclear. We devised a risk-structured, compartmental model, incorporating sexual behavior dynamics. We compared different strategies targeting the high-risk population: a scenario of control policies geared toward the use of condoms and/or sexual abstinence (robust control strategy) with risk compensation behavior change, and a scenario of control strategies with behavior change in response to the doubling rate (adaptive control strategy). Monkeypox's basic reproduction number is 1.464, 0.0066, and 1.461 in the high-risk, low-risk, and total populations, respectively, with the high-risk group being the major driver of monkeypox spread. Policies imposing condom use or sexual abstinence need to achieve a 35% minimum compliance rate to stop further transmission, while a combination of both can curb the spread with 10% compliance to abstinence and 25% to condom use. With risk compensation, the only option is to impose sexual abstinence by at least 35%. Adaptive control is more effective than robust control where the daily sexual contact number is reduced proportionally and remains constant thereafter, shortening the time to epidemic peak, lowering its size, facilitating disease attenuation, and playing a key role in controlling the current outbreak.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2025

Ecological acclimation: A framework to integrate fast and slow responses to climate change

Ecological responses to climate change occur across vastly different time-scales, from minutes for physiological plasticity to decades or centuries for community turnover and evolutionary adaptation. Accurately predicting the range of ecosystem trajectories will require models that incorporate both fast processes that may keep pace with climate change and slower ones likely to lag behind and generate disequilibrium dynamics. However, the knowledge necessary for this integration is currently fragmented across disciplines. We develop ‘ecological acclimation’ as a unifying framework to emphasize the similarity of dynamics driven by processes operating on dramatically different time-scales and levels of biological organization. The framework focuses on ecoclimate sensitivities, measured as the change in an ecological response variable per unit of climate change. Acclimation processes acting at different time-scales cause these sensitivities to shift in magnitude and even direction over time. We highlight shifting ecoclimate sensitivities in case studies from diverse ecosystems, including terrestrial plant communities, coral reefs and soil microbiomes. Models predicting future ecosystem states inevitably make assumptions about acclimation processes; these assumptions must be explicit for users to evaluate whether a model is appropriate for a given forecast horizon. Similarly, decision frameworks that clearly account for multiple acclimation processes and their distinct time-scales will help natural resource managers plan for ecological impacts of climate change from years to many decades into the future. We outline a synthetic research programme focused on the time-scales of ecological acclimation to reduce uncertainty in ecological forecasts.

climate adaptation↗

DOE OSTI2023

Efficient Reliability Analysis using Generalized Multifidelity Modeling and Explainable Active Learning

To assess the reliability of critical technologies like nuclear plants and infrastructure systems and improve the robustness of design, engineers have to quantify the uncertainties surrounding the system behavior accurately. However, the complexity of the problem can make standard reliability analysis algorithms prohibitively expensive, primarily due to the high computational cost of estimating the system response at each iteration. This cost can be greatly reduced by using multi-fidelity modeling and machine learning to build a surrogate model to replace the expensive response function. We propose a general and robust method for building surrogates from multiple Low Fidelity (LF) models coupled with machine learning to retain accuracy. Our framework first constructs “Corrected Low Fidelity models” (CLFs) by coupling a High Fidelity (HF) model inferred Gaussian Process correction term with each of the LF models. It then uses the correction terms to assign model probabilities to each of these CLFs in an explainable way before using them to assemble the final surrogate. No assumptions are made about the type of the LF models or their correlation with the HF model. The proposed surrogate modeling framework is used within the subset simulation algorithm (a variance-reduced MCMC-based reliability analysis algorithm) for enhanced efficiency. Additionally, an active learning step is added to the algorithm to adaptively decide when the surrogate is not sufficiently accurate, at which point the HF model is called and used to refine the surrogate. Through a frame buckling example, our method is shown to be highly efficient at reducing the expensive HF model calls while accurately estimating the failure probability.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

Turbulent field fluctuations in gyrokinetic and fluid plasmas

A key uncertainty in the design and development of magnetic confinement fusion energy reactors is predicting edge plasma turbulence. An essential step in overcoming this uncertainty is the validation in accuracy of reduced turbulent transport models. Drift-reduced Braginskii two-fluid theory is one such set of reduced equations that has for decades simulated boundary plasmas in experiment, but significant questions exist regarding its predictive ability. To this end, using a novel physics-informed deep learning framework, we demonstrate the first ever direct quantitative comparisons of turbulent field fluctuations between electrostatic two-fluid theory and electromagnetic gyrokinetic modelling with good overall agreement found in magnetized helical plasmas at low normalized pressure. This framework is readily adaptable to experimental and astrophysical environments, and presents a new technique for the numerical validation and discovery of reduced global plasma turbulence models.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2023

MtrA modulates Mycobacterium tuberculosis cell division in host microenvironments to mediate intrinsic resistance and drug tolerance

The success of Mycobacterium tuberculosis (Mtb) is largely attributed to its ability to physiologically adapt and withstand diverse localized stresses within host microenvironments. Here, we present a data-driven model (EGRIN 2.0) that captures the dynamic interplay of environmental cues and genome-encoded regulatory programs in Mtb. Analysis of EGRIN 2.0 shows how modulation of the MtrAB two-component signaling system tunes Mtb growth in response to related host microenvironmental cues. Disruption of MtrAB by tunable CRISPR interference confirms that the signaling system regulates multiple peptidoglycan hydrolases, among other targets, that are important for cell division. Further, MtrA decreases the effectiveness of antibiotics by mechanisms of both intrinsic resistance and drug tolerance. Together, the model-enabled dissection of complex MtrA regulation highlights its importance as a drug target and illustrates how EGRIN 2.0 facilitates discovery and mechanistic characterization of Mtb adaptation to specific host microenvironments within the host.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2023

Traffic Signal Optimization by Integrating Reinforcement Learning and Digital Twins

Machine learning (ML) methods, especially reinforcement learning (RL), have been widely considered for traffic signal optimization in intelligent transportation systems. Most of these ML methods are centralized, lacking in scalability and adaptability in large traffic networks. Further, it is challenging to train such ML models due to the lack of training platforms and/or the cost of deploying and training in a real traffic networks. This paper presents an approach for the integration of decentralized graph-based multi-agent reinforcement learning (DGMARL) with a Digital Twin (DT) to optimize traffic signals for the reduction of traffic congestion and network-wide fuel consumption related to stopping. Specifically, the DGMARL agents learn traffic state patterns and make decisions regarding traffic signal control with assistance from a Digital Twin module, which simulates and replicates the traffic behaviors of a real traffic network. The proposed approach was evaluated using PTV-Vissim [1], a microscopic traffic simulation platform. PTV-Vissim is also the simulation engine of the DT, enabling emulation and optimization of the traffic signals on the MLK Smart Corridor in Chattanooga, Tennessee. Compared to an actuated signal control baseline approach, experiment results show that Eco_PI, a developed performance measure capturing the impact of stops on fuel consumption, was reduced by 44.27% in a 24-hour and an average of 29.88% in a PM-peak-hour scenario.

Multi-Agent Reinforcement Learning, Digital Twin, ↗

DOE OSTI2022

Performance Analysis of Speculative Parallel Adaptive Local Timestepping for Conservation Laws

Stable simulation of conservation laws, such as those used to model fluid dynamics and plasma physics applications, requires the satisfaction of the so-called Courant-Friedrichs-Lewy condition. By allowing regions of the mesh to advance with different timesteps that locally satisfy this stability constraint, significant work reduction can be attained when compared to a time integration scheme using a single timestep size. However, parallelizing this algorithm presents considerable difficulty. Since the stability condition depends on the state of the system, dependencies become dynamic and potentially non-local. In this article, we present an adaptive local timestepping algorithm using an optimistic (Timewarp-based) parallel discrete event simulation. We introduce waiting heuristics to limit misspeculation and a semi-static load balancing scheme to eliminate load imbalance as parts of the mesh require finer or coarser timesteps. Last, we outline an interface for separating the physics of the specific conservation law from the temporal integration allowing for productive adoption of our proposed algorithm. We present a misspeculation study for three conservation laws, demonstrating both the productivity of the local timestepping API, for which 74% of the lines of code are reused across different conservation laws, and the robustness of the waiting heuristics—at most 1.5% of element updates are rolled back. Our performance studies demonstrate up to a 2.8× speedup versus a baseline unoptimized local timestepping approach, a 4x improvement in per-node throughput compared to an MPI parallelization of synchronous timestepping, and scalability up to 3,072 cores on NERSC’s Cori Haswell partition.

97 MATHEMATICS AND COMPUTING↗