Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Model initialization”

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 919 records · Page 51
All sourcesNASA NTRSDOE OSTI
NASA NTRS2004

Sensor Web Dynamic Measurement Techniques and Adaptive Observing Strategies

Sensor Web observing systems may have the potential to significantly improve our ability to monitor, understand, and predict the evolution of rapidly evolving, transient, or variable environmental features and events. This improvement will come about by integrating novel data collection techniques, new or improved instruments, emerging communications technologies and protocols, sensor mark-up languages, and interoperable planning and scheduling systems. In contrast to today's observing systems, "event-driven" sensor webs will synthesize real- or near-real time measurements and information from other platforms and then react by reconfiguring the platforms and instruments to invoke new measurement modes and adaptive observation strategies. Similarly, "model-driven" sensor webs will utilize environmental prediction models to initiate targeted sensor measurements or to use a new observing strategy. The sensor web concept contrasts with today's data collection techniques and observing system operations concepts where independent measurements are made by remote sensing and in situ platforms that do not share, and therefore cannot act upon, potentially useful complementary sensor measurement data and platform state information. This presentation describes NASA's view of event-driven and model-driven Sensor Webs and highlights several research and development activities at the Goddard Space Flight Center.

Talabac, Stephen J.↗
NASA NTRS2018

Acceleration Modes and Transitions in Pulsed Plasma Accelerators

Pulsed plasma accelerators typically operate by storing energy in a capacitor bank and then discharging this energy through a gas, ionizing and accelerating it through the Lorentz body force. Two plasma accelerator types employing this general scheme have typically been studied: the gas-fed pulsed plasma thruster and the quasi-steady magnetoplasmadynamic (MPD) accelerator. The gas-fed pulsed plasma accelerator is generally represented as a completely transient device discharging in approximately 1-10 microseconds. When the capacitor bank is discharged through the gas, a current sheet forms at the breech of the thruster and propagates forward under a j (current density) by B (magnetic field) body force, entraining propellant it encounters. This process is sometimes referred to as detonation-mode acceleration because the current sheet representation approximates that of a strong shock propagating through the gas. Acceleration of the initial current sheet ceases when either the current sheet reaches the end of the device and is ejected or when the current in the circuit reverses, striking a new current sheet at the breech and depriving the initial sheet of additional acceleration. In the quasi-steady MPD accelerator, the pulse is lengthened to approximately 1 millisecond or longer and maintained at an approximately constant level during discharge. The time over which the transient phenomena experienced during startup typically occur is short relative to the overall discharge time, which is now long enough for the plasma to assume a relatively steady-state configuration. The ionized gas flows through a stationary current channel in a manner that is sometimes referred to as the deflagration-mode of operation. The plasma experiences electromagnetic acceleration as it flows through the current channel towards the exit of the device. A device that had a short pulse length but appeared to operate in a plasma acceleration regime different from the gas-fed pulsed plasma accelerators was developed by Cheng, et al. The Coaxial High ENerGy (CHENG) thruster operated on the 10-microseconds timescales of pulsed plasma thrusters, but claimed high thrust density, high efficiency and low electrode erosion rates, which are more consistent with the deflagration mode of acceleration. Separate work on gas-fed pulsed plasma thrusters (PPTs) by Ziemer, et al. identified two separate regimes of performance. The regime at higher mass bits (termed Mode I in that work) possessed relatively constant thrust efficiency (ratio of jet kinetic energy to input electrical energy) as a function of mass bit. In the second regime at very low mass bits (termed Mode II), the efficiency increased with decreasing mass bit. Work by Poehlmann et al. and by Sitaraman and Raja sought to understand the performance of the CHENG thruster and the Mode I / Mode II performance in PPTs by modeling the acceleration using the Hugoniot Relation, with the detonation and deflagration modes representing two distinct sets of solutions to the relevant conservation laws. These works studied the proposal that, depending upon the values of the various controllable parameters, the accelerator would operate in either the detonation or deflagration mode. In the present work, we propose a variation on the explanation for the differences in performance between the various pulsed plasma accelerators. Instead of treating the accelerator as if it were only operating in one mode or the other during a pulse, we model the initial stage of the discharge in all cases as an accelerating current sheet (detonation mode). If the current sheet reaches the exit of the accelerator before the discharge is completed, the acceleration mode transitions to the deflagration mode type found in the quasi-steady MPD thrusters. This modeling method is used to demonstrate that standard gas-fed pulsed plasma accelerators, the CHENG thruster, and the quasi-steady MPD accelerator are variations of the same device, with the overall acceleration of the plasma depending upon the behavior of the plasma discharge during initial transient phase and the relative lengths of the detonation and deflagration modes of operation.

Polzin, Kurt A.↗
DOE OSTI2020

Weak Lensing And Cosmology In The Dark Energy Survey

In this PhD Thesis, we present studies which aim at understanding and testing the fundamental assumptions of the $\Lambda$CDM cosmological model. We initially show that weak gravitational lensing can be derived as a consequence of General Relativity, and that it can be applied to astronomical data from the Dark Energy Survey (DES) in order to constrain the amplitude of matter fluctuations in the universe. This constraint can be used as an end-to-end cosmological test when it is compared to inferences from the Cosmic Microwave Background. We further extend the cosmological model to account for non-standard parameters which can point to novel physics, and use DES data to assess their statistical significance. We finally show that the Cold Dark Matter (CDM) paradigm can also be fundamentally tested with observational data using morphological signatures of disk galaxies in cluster environments if CDM is self-interacting.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

Domain Aware Deep-learning Algorithms Integrated with Scientific-computing Technologies (DADAIST)

This technical report summarized the contribution of the DADAIST project funded by the Data Model Convergence Initiative via the Laboratory Directed Research and Development (LDRD) investments at Pacific Northwest National Laboratory (PNNL). Specifically, we report the development of the NeuroMANCER (Neural Modules with Adaptive Nonlinear Constraints and Efficient Regularizations), a new open-source Scientific Machine Learning library for formulating and solving parametric constrained optimization problems, physics-informed system identification, and parametric optimal control problems. NeuroMANCER is using differentiable programming to combine modern data-driven models and optimization modeling language into a coherent algorithmic and software framework. NeuroMANCER is a Pytorch-based framework and adopts much of its philosophy focused on research and development, rapid prototyping, and streamlined deployment. Strong emphasis is given to extensibility, interoperability with the PyTorch ecosystem, and quick adaptability to custom domain problems. Neuromancer repository contains a comprehensive library of differentiable modules, including custom activation functions, matrix factorizations, deep learning architectures, neural differential equations, differential equation solvers, implicit layers such as iterative solvers, high-level API for symbolic expressions, API for modeling and control of dynamical systems, and extensive set of tutorial code examples in the form of python scripts and jupyter notebooks.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Enhancing segmentation fairness through curriculum learning and progressive loss: a centralized and federated perspective on radiograph analysis

Bias in medical image segmentation can lead to unequal performance across demographic subgroups, raising concerns about fairness and reliability in clinical AI systems. While deep learning models have achieved high segmentation accuracy, ensuring equitable performance across race and gender remains a significant challenge, particularly in privacy-sensitive healthcare environments. This study investigates fairness-aware medical image segmentation for hip and knee radiographs using deep learning models evaluated in both centralized and Federated Learning (FL) settings. We introduce Curriculum Learning (CL) strategies and Progressive Loss (PL) functions to regulate sample difficulty during training. In addition, we propose two novel fairness-oriented federated learning algorithms, Federated Intersection over Union (FedIoU) and Federated Intersection over Union with Outlier Analysis (FedIoUoutlier). Experiments are conducted using multiple segmentation backbones and simulated multi-site data partitions derived from the Osteoarthritis Initiative dataset. Model performance is evaluated using Intersection over Union (IoU), IoU standard deviation, Skewed Error Ratio (SER), and Min-Max Disparity across race and gender subgroups. Statistical significance was verified using paired t-tests to compare per-sample IoU performance against baseline configurations. Across both hip and knee segmentation tasks, curriculum learning and progressive loss strategies consistently improved segmentation accuracy and reduced demographic performance disparities in centralized training. In federated settings, fairness-aware aggregation further enhanced performance. Notably, FedIoUoutlier combined with balanced curriculum learning and tiered progressive loss achieved the highest mean IoU while yielding the lowest SER and Min-Max Disparity, indicating improved fairness without sacrificing accuracy. In several configurations, federated models matched or exceeded the performance of optimized centralized models, with statistically significant improvements in per-sample IoU over baseline configurations. The results demonstrate that structured training strategies and fairness-aware federated aggregation can jointly improve accuracy, stability, and demographic fairness in medical image segmentation. By integrating curriculum learning, progressive loss, and novel FL algorithms, this work provides a practical pathway toward equitable and privacy-preserving AI systems for medical imaging.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2015

IPRT Polarized Radiative Transfer Model Intercomparison Project-Phase A

The polarization state of electromagnetic radiation scattered by atmospheric particles such as aerosols, cloud droplets, or ice crystals contains much more information about the optical and microphysical properties than the total intensity alone. For this reason an increasing number of polarimetric observations are performed from space, from the ground and from aircraft. Polarized radiative transfer models are required to interpret and analyse these measurements and to develop retrieval algorithms exploiting polarimetric observations. In the last years a large number of new codes have been developed, mostly for specific applications. Benchmark results are available for specific cases, but not for more sophisticated scenarios including polarized surface reflection and multi-layer atmospheres. The International Polarized Radiative Transfer (IPRT) working group of the International Radiation Commission (IRC) has initiated a model intercomparison project in order to fill this gap. This paper presents the results of the first phase A of the IPRT project which includes ten test cases, from simple setups with only one layer and Rayleigh scattering to rather sophisticated setups with a cloud embedded in a standard atmosphere above an ocean surface. All scenarios in the first phase A of the intercomparison project are for a one-dimensional plane-parallel model geometry. The commonly established benchmark results are available at the IPRT website

radiative transfer↗
NASA NTRSDate not supplied

Quantum Hardware-Enabled Molecular Dynamics via Transfer Learning

The ability to perform ab initio molecular dynamics simulations using potential energy surfaces provided by quantum computers would open the door to virtually exact dynamics for a variety of chemical and biochemical systems, with impacts on catalysis and biophysics. Nonetheless, performing molecular dynamics on surfaces produced by quantum hardware has been hampered by the noisy energies typically produced by quantum computers and challenges associated with computing gradients and scaling to large systems interest. A recent set of advances in machine learning, known as transfer learning, provides a new path forward for molecular dynamics simulations on quantum hardware. Transfer learning offers a workaround, where one first trains models on larger, less accurate classical datasets and then refines them on smaller, more accurate quantum datasets. We explore this approach by training machine learning models to predict a molecule's potential energy based on its geometric structure using Behler-Parrinello neural networks. When successfully trained, the model enables energy gradient predictions necessary for dynamic simulations. To reduce the quantum resources needed, the model is initially trained with data derived from classical density functional theory and subsequently refined with a smaller dataset obtained from a variational quantum eigensolver optimization of the unitary coupled cluster ansatz. We show that this approach significantly reduces the size of the needed quantum training dataset while capturing the high accuracies needed within quantum chemistry simulations. The success of this two-step training method opens more opportunities to apply machine learning models on quantum data, a significant stride towards efficient quantum-classical hybrid computational models.

quantum computing↗
NASA NTRS2010

A Combined Observational and Modeling Approach to Study Modern Dust Transport from the Patagonia Desert to East Antarctica

The understanding of present atmospheric transport processes from Southern Hemisphere (SH) landmasses to Antarctica can improve the interpretation of stratigraphic data in Antarctic ice cores. In addition, long range transport can deliver key nutrients normally not available to marine ecosystems in the Southern Ocean and may trigger or enhance primary productivity. However, there is a dearth of observational based studies of dust transport in the SH. This work aims to improve current understanding of dust transport in the SH by showing a characterization of two dust events originating in the Patagonia desert (south end of South America). The approach is based on a combined and complementary use of satellite retrievals (detectors MISR, MODIS, GLAS ,POLDER, OMI,), transport model simulation (HYSPLIT) and surface observations near the sources and aerosol measurements in Antarctica (Neumayer and Concordia sites). Satellite imagery and visibility observations confirm dust emission in a stretch of dry lakes along the coast of the Tierra del Fuego (TdF) island (approx.54deg S) and from the shores of the Colihue Huapi lake in Central Patagonia (approx.46deg S) in February 2005. Model simulations initialized by these observations reproduce the timing of an observed increase in dust concentration at the Concordia Station and some of the observed increases in atmospheric aerosol absorption (here used as a dust proxy) in the Neumayer station. The TdF sources were the largest contributors of dust at both sites. The transit times from TdF to the Neumayer and Concordia sites are 6-7 and 9-10 days respectively. Lidar observations and model outputs coincide in placing most of the dust cloud in the boundary layer and suggest significant de- position over the ocean immediately downwind. Boundary layer dust was detected as far as 1800 km from the source and approx.800 km north of the South Georgia Island over the central sub-Antarctic Atlantic Ocean. Although the analysis suggests the presence of dust at approx.1500 km SW of South Africa five days after, the limited capabilities of existing satellite platforms to differentiate between aerosol types do not permit a definitive conclusion. In addition, the model simulations show dust lifting to the free troposphere as it travels south but it could not be confirmed by the satellite observations due to cloudiness. This work demonstrates that complementary information from existing transport models, satellite and surface data can yield a consistent picture of the dust transport from the Patagonia desert to Antarctica. It also illustrates the limitation of using any of these approaches individually to characterize the transport of dust in a heavily cloudy area.

Gasso, S.↗
NASA NTRS1982

Finite-element analysis of initiation, stable crack growth and instability using a crack-tip-opening displacement criterion

An elastic-plastic (incremental and small strain) finite element analysis was used with a crack growth criterion to study crack initiation, stable crack growth, and instability under monotonic loading to failure of metallic materials. The crack growth criterion was a critical crack-tip-opening displacement (CTOD) at a specified distance from the crack tip, or equivalently, a critical crack-tip-opening angle (CTOA). Whenever the CTOD (or CTOA) equaled or exceeded a critical value, the crack was assumed to grow. Single values of critical CTOD were used in the analysis to model crack initiation, stable crack growth, and instability for 7075-T651 and 2024-T351 aluminum alloy compact specimens. Calculated and experimentally measured CTOD values at initiation agreed well for both aluminum alloys. These critical CTOD values were also used to predict failure loads on center-crack tension specimens and a specially-designed three-hole-crack tension specimen made of the two aluminum alloys and of 304 stainless steel. All specimens were 12.7 mm thick. Predicted failure loads for 7075-T651 aluminum alloy and 304 stainless steel specimens were generally within + or - 15 percent of experimental failure loads, whereas the predicted failure loads for 2024-T351 aluminum alloy specimens were generally within + or - 5 percent of the experimental loads.

Newman, J. C., Jr.↗
DOE OSTI2023

Joint velocity and density reconstruction of the Universe with nonlinear differentiable forward modeling

Reconstructing the initial conditions of the Universe from late-time observations has the potential to optimally extract cosmological information. Due to the high dimensionality of the parameter space, a differentiable forward model is needed for convergence, and recent advances have made it possible to perform reconstruction with nonlinear models based on galaxy (or halo) positions. In addition to positions, future surveys will provide measurements of galaxies' peculiar velocities through the kinematic Sunyaev-Zel'dovich effect (kSZ), type Ia supernovae, the fundamental plane relation, and the Tully-Fisher relation. Here we develop the formalism for including halo velocities, in addition to halo positions, to enhance the reconstruction of the initial conditions. We show that using velocity information can significantly improve the reconstruction accuracy compared to using only the halo density field. Here we study this improvement as a function of shot noise, velocity measurement noise, and angle to the line of sight. We also show how halo velocity data can be used to improve the reconstruction of the final nonlinear matter overdensity and velocity fields. We have built our pipeline into the differentiable Particle-Mesh FlowPM package, paving the way to perform field-level cosmological inference with joint velocity and density reconstruction. This is especially useful given the increased ability to measure peculiar velocities in the near future.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS2020

Multidisciplinary Systems Analysis of a Six Passenger Quadrotor Urban Air Mobility Vehicle Powertrain

There is a high degree of research interest in the design space for electric vertical takeoff and landing (eVTOL) vehicles, because these vehicles are seen as key enablers for urban air mobility (UAM). This work further explores the eVTOL design space, by presenting analysis of a six-passenger eVTOL quadrotor powertrain, with integrated power, propulsion, and thermal management systems modeled using the Numerical Propulsion System Simulation (NPSS) and the NPSS Power System Library. Four architectures are modeled at the same design point and compared over a design mission. Results from an architecture comparison study show that a hybrid architecture performs best in terms of cruise range, however, the hybrid needs batteries with high specific energy and specific power in order to obtain a benefit vs turboelectric architectures. Sensitivity studies are conducted to show correlations between component parameters and system metrics. This data provides useful indicators for further technological improvement. Lastly, an initial TMS model is presented, and sensitivity studies on TMS design parameters are presented as well, to show high level TMS design trends.

electrified propulsion↗
NASA NTRSDate not supplied

Multidisciplinary Systems Analysis of a Six Passenger Quadrotor Urban Air Mobility Vehicle Powertrain

There is a high degree of research interest in the design space for electric vertical takeoff and landing (eVTOL) vehicles, because these vehicles are seen as key enablers for urban air mobility (UAM). This work further explores the eVTOL design space, by presenting analysis of a six-passenger eVTOL quadrotor powertrain, with integrated power, propulsion, and thermal management systems modeled using the Numerical Propulsion System Simulation (NPSS) and the NPSS Power System Library. Four architectures are modeled at the same design point and compared over a design mission. Results from an architecture comparison study show that a hybrid architecture performs best in terms of cruise range, however, the hybrid needs batteries with high specific energy and specific power in order to obtain a benefit vs turboelectric architectures. Sensitivity studies are conducted to show correlations between component parameters and system metrics. This data provides useful indicators for further technological improvement. Lastly, an initial TMS model is presented, and sensitivity studies on TMS design parameters are presented as well, to show high level TMS design trends.

evtol↗
NASA NTRS1992

An intracloud lightning parameterization scheme for a storm electrification model

The parameterization of an intracloud lightning discharge has been implemented in the present storm electrification model. The initiation, propagation direction, and termination of the discharge are computed using the magnitude and direction of the electric field vector as the determining criteria. The charge redistribution due to the lightning is approximated assuming the channel to be an isolated conductor with zero net charge over its entire length. Various simulations involving differing amounts of charge transferred and distribution of charges have been done. Values of charge transfer, dipole moment change, and electrical energy dissipation computed in the model are consistent with observations. The effects of the lightning-produced ions on the hydrometeor charges and electric field components depend strongly on the amount of charge transferred. A comparison between the measured electric field change of an actual intracloud flash and the field change due to the simulated discharge shows favorable agreement. Limitations of the parameterization scheme are discussed.

Helsdon, John H., Jr.↗
NASA NTRS1984

Atmosphere behavior in gas-closed mouse-algal systems - An experimental and modelling study

A NASA-sponsored research program initiated using mathematical modelling and laboratory experimentation aimed at examining the gas-exchange characteristics of artificial animal/plant systems closed to the ambient atmosphere is studied. The development of control techniques and management strategies for maintaining the atmospheric levels of carbon dioxide and oxygen at physiological levels is considered. A mathematical model simulating the behavior of a gas-closed mouse-algal system under varying environmental conditions is described. To verify and validate the model simulations, an analytical system with which algal growth and gas exchange characteristics can be manipulated and measured is designed, fabricated, and tested. The preliminary results are presented.

Averner, M. M.↗
NASA NTRS2011

Comprehensive Forced Response Analysis of J2X Turbine Bladed-Discs with 36- Degree Variation in CFD Loading

At NASA/MSFC, Structural Dynamics personnel continue to perform advanced analysis for the turbomachinery in the J2X Rocket Engine, which is under consideration for the new Space Launch System. One of the most challenging analyses in the program is predicting turbine blade structural capability. Resonance was predicted by modal analysis, so comprehensive forced response analyses using high fidelity cyclic symmetric finite element models were initiated as required. Analysis methodologies up to this point have assumed the flow field could be fully described by a sector, so the loading on every blade would be identical as it travelled through it. However, in the J2X the CFD flow field varied over the 360 deg of a revolution because of the flow speeds and tortuous axial path. MSFC therefore developed a complex procedure using Nastran Dmap's and Matlab scripts to apply this circumferentially varying loading onto the cyclically symmetric structural models to produce accurate dynamic stresses for every blade on the disk. This procedure is coupled with static, spin, and thermal loading to produce high cycle fatigue safety factors resulting in much more accurate analytical assessments of the blades.

Elrod, David↗
DOE OSTI2025

Evaluation of historical precipitation interannual variability in CMIP6 over the United States

Interannual precipitation variability profoundly influences society via its effects on agriculture, water resources, infrastructure, and disaster risks. In this study, we use daily in situ precipitation observations from the global historical climatology network-daily (GHCN-D) to assess the ability of 21 Coupled Model Intercomparison Project Phase 6 (CMIP6) models, including the 50-member fifth-generation Canadian Earth System Model single model initial-condition large ensemble (CanESM5_SMILE), to realistically simulate historical interannual precipitation variability trends within 17 regions of the contiguous United States (CONUS). We assess how accurately the CMIP6 simulations align with observational data across annual, summer, and winter periods, focusing on four key hydrometeorological metrics, including interannual precipitation variability, relative interannual precipitation variability (coefficient of variation), annual mean precipitation, and annual wet day frequency. Our findings reveal that CMIP6 ensemble members generally reproduce the spatial patterns of observed trends in annual mean precipitation. In most regions, models agree well with the signs of observed changes in annual mean precipitation, though discrepancies in trend magnitude are evident. Further, observed trends in winter mean precipitation broadly exhibit a spatial pattern similar to that of the observed annual mean. However, analysis of the CanESM5_SMILE shows that trends in precipitation variability may primarily be the result of model-simulated internal variability, suggesting caution in interpreting multi-model single-realization ensemble results. Challenges in accurately simulating interannual precipitation variability underscore the need for ongoing model refinement and validation to enhance climate projections, especially in regions vulnerable to extreme precipitation events.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

A Data-Agnostic, Continuous Machine Learning Framework for Application in High Energy Physics and Beyond: Phase 1 Final Scientific/Technical Report

This Phase 1 effort has focused on the development of continual learning frameworks for use in machine learning, specifically in the applied context of High Energy Physics (HEP). Machine learning (ML) is a transformative technology by which computers, typically through the use of neural networks, are able to perform tasks with proficiency that rivals or surpasses that of human users. Model Degradation & Catastrophic Forgetting are two undesired phenomena which can occur in ML where the performance of a model degrades when either deployed on novel data streams, or trained on novel data which are sufficiently different than the data the models were initially trained on. A natural example where these sorts of effects can be observed is in the performance of detectors in harsh environments, where the detector signature may change over the lifetime of the detector as it ages and deteriorates — precisely what occurs in the experiments conducted in HEP. Real world HEP data is therefore an excellent test-ground and use-case for Continual Learning paradigms, which are techniques used in ML to counteract these problems. Ensemble learning is one such technique, where multiple smaller models are trained on subsets of the overall data and are ensembled together during inference. The intuition behind this technique is that, although there are shifts in the distributions which govern the incoming data streams, these shifts are not expected to be homogeneous or global. If a sufficient diversity in solutions within the various sub-models has been achieved, then at least one sub-model is expected to retain its performance within the overall ensemble. One further strength of this approach is that the architectures of the various models do not need to be identical, and in fact even different modalities of data can naturally be combined in this way. This work focused on applying ensemble learning techniques to derive results using two main datasets, anomaly detection in HEP data & time-series forecasting in semiconductor manufacturing data. Semiconductor manufacturing involves data with surprising similarity to that of HEP (e.g. wafer maps look very similar to digi-occupancy maps) and Cerium Lab’s prominence within the semiconductor industry makes semiconductor manufacturing a natural opportunity for commercialization of this work. Our efforts have led to two strong results. The first is that we evaluated the proposed ensembling techniques using previously proposed machine learning architectures for use in anomaly detection, namely AutoEncoder based models and their derivatives. We also developed new architectures which have not been evaluated in this context before. In fact, this work marks the first use of Vision Transformers for anomaly detection in HEP. Second, we demonstrated that ensemble learning significantly improves model performance in scenarios prone to degradation, validating its effectiveness across both HEP and semiconductor datasets. These results further support ensemble learning as a powerful strategy for mitigating catastrophic forgetting and maintaining robust performance in evolving data environments.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Fixing Amdahl's Law within the Limits of Accelerated Systems: FALLACY

Closeout report for FALLACY project. The performance of Data Model Convergence Initiative (DMC) applications on parallel machines is far below the limit set by Amdahl’s law. Whether the machine is based on many-core, GPUs, FPGAs, or a heterogeneous combination, usually the most significant bottleneck is accessing data from the memory system. Aligning with DMC’s HW/architecture thrust, this project developed a set of memory-centric tools called ‘MemGaze’ that inform the HW/SW stack about an application’s memory behavior, including data access latency and diagnosing poor data layout and data composition. Our approach uses architectural modeling and analysis of workload data accesses.

97 MATHEMATICS AND COMPUTING↗