Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “memory effects”

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

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

At least 1,603 records · Page 89
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Indicator-directed Dynamic Power Management for Iterative Workloads on GPU-Accelerated Systems

Modern high-performance and warehouse computing centers show strong interest in minimizing system power consumption while satisfying customers’ quality of service (QoS). Dynamic voltage and frequency scaling (DVFS) is effective for achieving this goal. Nevertheless, automating the process online and making it transparent to users must address three major challenges: (1) Complexity — today’s hardware components (e.g., CPUs, GPUs, memory, network, etc.) can be configured in several or dozens of frequency/voltage states for satisfying divergent system demands. Given their combination and the emergence of heterogeneity, searching the optimal configuration in the design space online can be timing consuming. (2) QoS guarantee — user-defined objectives such as power constraint and performance target must be monitored, predicted and ensured at the best effort. (3) Adaptability — various known and unknown workloads run on systems. Workloads characteristics should be quickly determined and configurations dynamically adjusted in accord with workloads and QoS. In this work, we focus on applications exhibiting an interesting feature – iterative or periodic, which is common among conventional HPC and emerging machine learning workloads. We propose an online dynamic power-performance (ODPP) management framework to dynamically adjust GPU DVFS configurations to meet performance and power objectives and constraints, without any code annotation or intrusion. Particularly, ODPP extracts the performance and power indicators for applications from their resources utilization profiles in a short episode. It further automatically constructs an accurate model that infers from the indicators how the application's performance and power vary with GPU core and memory frequencies. Aided with the model, for both seen and unseen applications, ODPP can quickly determine the most appropriate DVFS configuration for their execution. We evaluate ODPP on an NVIDIA GPU using multiple exascale computing (ECP) and deep learning applications.

Zou, Pengfei↗
DOE OSTI2016

CrossSim

CrossSim is a simulator for modeling neural-inspired machine learning algorithms on analog hardware, such as resistive memory crossbars. It includes noise models for reading and updating the resistances, which can be based on idealized equations or experimental data. It can also introduce noise and finite precision effects when converting values from digital to analog and vice versa. All of these effects can be turned on or off as an algorithm processes a data set and attempts to learn its salient attributes so that it can be categorized in the machine learning training/classification context. CrossSim thus allows the robustness, accuracy, and energy usage of a machine learning algorithm to be tested on simulated hardware.

Plimpton, StevenJ.↗
DOE OSTI2020

Sage: parallel semi-asymmtric graph algorithms for NVRAMs

Non-volatile main memory (NVRAM) technologies provide an attractive set of features for large-scale graph analytics, including byte-addressability, low idle power, and improved memory-density. NVRAM systems today have an order of magnitude more NVRAM than traditional memory (DRAM). NVRAM systems could therefore potentially allow very large graph problems to be solved on a single machine, at a modest cost. However, a significant challenge in achieving high performance is in accounting for the fact that NVRAM writes can be much more expensive than NVRAM reads. In this paper, we propose an approach to parallel graph analytics using the Parallel Semi-Asymmetric Model (PSAM), in which the graph is stored as a read-only data structure (in NVRAM), and the amount of mutable memory is kept proportional to the number of vertices. Similar to the popular semi-external and semi-streaming models for graph analytics, the PSAM approach assumes that the vertices of the graph fit in a fast read-write memory (DRAM), but the edges do not. In NVRAM systems, our approach eliminates writes to the NVRAM, among other benefits. To experimentally study this new setting, we develop Sage, a parallel semi-asymmetric graph engine with which we implement provably-efficient (and often work-optimal) PSAM algorithms for over a dozen fundamental graph problems. We experimentally study Sage using a 48--core machine on the largest publicly-available real-world graph (the Hyperlink Web graph with over 3.5 billion vertices and 128 billion edges) equipped with Optane DC Persistent Memory, and show that Sage outperforms the fastest prior systems designed for NVRAM. Importantly, we also show that Sage nearly matches the fastest prior systems running solely in DRAM, by effectively hiding the costs of repeatedly accessing NVRAM versus DRAM.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2018

An ergodic approach to polarization in eRHIC electron storage ring

Electron polarization in a storage ring is subject to two long term effects: Sokolov-Ternov polarization and depolarization by diffusion, which over a long time scale lead to an equilibrium state. Simulation-wise, this can be highly CPU time and memory consuming. Simulations aimed at determining optimal ring storage energy in an electron-ion collider use to track thousand particle bunches, for a long time - yet still short compared to depolarization time scales, due to HPC limitations. Based on considerations of ergodicity of electron bunch dynamics in the presence of synchrotron radiation, tracking a single particle instead is investigated. This allows substantial saving in teh required HPC volume, “CPU-time×Memory-allocation”. The concept is illustrated with polarization lifetime and equilibrium polarization simulations at the eRHIC electron-ion collider.

43 PARTICLE ACCELERATORS↗
NASA NTRS1966

MOSFET analog memory circuit achieves long duration signal storage

Memory circuit maintains the signal voltage at the output of an analog signal amplifier when the input signal is interrupted or removed. The circuit uses MOSFET /Metal Oxide Semiconductor Field Effect Transistor/ devices as voltage-controlled switches, triggered by an external voltage-sensing device.

Source record↗
NASA NTRS1993

NASA Electronic Library System (NELS) optimization

This is a compilation of NELS (NASA Electronic Library System) Optimization progress/problem, interim, and final reports for all phases. The NELS database was examined, particularly in the memory, disk contention, and CPU, to discover bottlenecks. Methods to increase the speed of NELS code were investigated. The tasks included restructuring the existing code to interact with others more effectively. An error reporting code to help detect and remove bugs in the NELS was added. Report writing tools were recommended to integrate with the ASV3 system. The Oracle database management system and tools were to be installed on a Sun workstation, intended for demonstration purposes.

Pribyl, William L.↗
NASA NTRS2016

Evaluation of Synthetic Vision Display Concepts for Improved Awareness in Unusual Attitude Recovery Scenarios

A recent study conducted by the Commercial Aviation Safety Team (CAST) determined 40 percent of all fixed-wing fatal accidents, between 2001 and 2011, were caused by Loss-of-Control (LOC) in flight (National Transportation Safety Board, 2015). Based on their findings, CAST recommended manufacturers develop and implement virtual day-visual meteorological conditions (VMC) display systems, such as synthetic vision or equivalent systems (CAST, 2016). In a 2015 simulation study conducted at NASA Langley Research Center (LaRC), researchers gathered to test and evaluate virtual day-VMC displays under realistic flight operation scenarios capable of inducing reduced attention states in pilots. Each display concept was evaluated to determine its efficacy to improve attitude awareness. During the experiment, Evaluation Pilots (EPs) were shown the following three display concepts on the Primary Flight Display (PFD): Baseline, Synthetic Vision (SV) with color gradient, and SV with texture. The baseline configuration was a standard, conventional 'blue over brown' display. Experiment scenarios were simulated over water to evaluate Unusual Attitude (UA) recovery over 'featureless terrain' environments. Thus, the SV with color gradient configuration presented a 'blue over blue' display with a linear blue color progression, to differentiate attitude changes between sky and ocean. The SV with texture configuration presented a 'blue over blue' display with a black checkerboard texture atop a synthetic ocean. These displays were paired with a Background Attitude Indicator (BAI) concept. The BAI was presented across all four Head-Down Displays (HDDs), displaying a wide field-of-view blue-over-blue attitude indicator. The BAI aligned with the PFD and showed through the background of the navigation displays with opaque transparency. Each EP participated in a two-part experiment series with a total seventy-five trial runs: Part I included a set of twenty-five Unusual Attitude Recovery (UAR) scenarios; Part II included a set of fifty Attitude Memory Recall Tasks (AMRT). At the conclusion of each trial, EPs were asked to complete a set post-run questionnaires. Quantitative results showed that there were no significant statistical effects on UA recovery times when utilizing SV with or without the presence of a BAI. Qualitative results show the SV displays (color, texture) with BAI On are most preferred for both UA recognition and recovery when compared with the baseline display. When only comparing SV display concepts, EPs performed better when using the SV with texture, BAI On, than any other display configuration. This is an interesting find considering most EPs noted their preference towards the SV with color gradient when the BAI was on.

Nicholas, Stephanie↗
DOE OSTI2017

Nurr1 overexpression exerts neuroprotective and anti-inflammatory roles via down-regulating CCL2 expression in both in vivo and in vitro Parkinson's disease models

The abnormality of nuclear receptor-related protein 1 (Nurr1) in expression and function can contribute to neurodegeneration of dopaminergic neurons and occurrence of Parkinson's disease (PD). However, its related mechanism in PD is still unknown. In this study, we found that Nurr1 was down-regulated and CCL2 was up-regulated in PD patients and PD mice. CCL2 promoted apoptosis and secretion of TNF-α and IL-1β in SH-SY5Y cells and inhibited cell viability while knockdown of CCL2 exerted the opposite effects. Nurr1 overexpression inhibited apoptosis, the release of TNF-α and IL-1β and promoted viability in α-Syn-treated SH-SY5Y cells, which was markedly promoted by CCL2 antibody and dramatically reversed by CCL2. Nurr1 overexpression negatively regulated CCL2 expression in vivo and in vitro. Furthermore, Nurr1 overexpression remarkably relieved MPTP-induced movement disorder and spatial memory deficits and played neuroprotective and anti-inflammatory roles in MPTP-induced PD mice by down-regulating CCL2 in vivo. In conclusion, Nurr1 overexpression exerts neuroprotective and anti-inflammatory roles via down-regulating CCL2 in both in vivo and in vitro PD models, contributing to developing mechanism-based and neuroprotective strategies against PD. - Highlights: • Nurr1 was down-regulated and CCL2 was up-regulated in PD patients and PD mice. • Nurr1 overexpression inhibited apoptosis, release of TNF-α and IL-1β and promoted viability in α-Syn-treated SH-SY5Y cells. • CCL2 reversed the effect of Nurr1 overexpression on apoptosis, inflammatory cytokines secretion and viability. • Nurr1 overexpression negatively regulated CCL2 expression in vivo and in vitro. • Nurr1 overexpression remarkably relieved MPTP-induced movement disorder and spatial memory deficits.

60 APPLIED LIFE SCIENCES↗
NASA NTRS1985

Advanced detectors and signal processing for bubble memories

The feasibility of combining silicon and magnetic bubble technologies is demonstrated. Results of bubble film annealing indicate that a low temperature silicon on garnet technology is the most likely one to succeed commercially. Annealing ambients are also shown to have a major effect on the magnetic properties of bubble films. Functional MOSFETs were fabricated on bubble films coated with thick (approximately 1 micron) SiO2 layers. The two main problems with these silicon on garnet MOSFETs are low electron mobilities and large gate leakage currents. Results indicate that the laser recrystallized silicon and gate oxide (SiO2) layers are contaminated. The data suggest that part of the contaminating ions originate in the sputtered oxide spacer layer and part originates in the bubble film itself. A diffusion barrier, such as silicon nitride, placed between the bubble film and the silicon layer should eliminate the contamination induced problem.

Kryder, M. H.↗
DOE OSTI2025

Optimizing the optimizer for physics-informed neural networks and Kolmogorov-Arnold networks

Physics-Informed Neural Networks (PINNs) have revolutionized the computation of PDE solutions by integrating partial differential equations (PDEs) into the neural network’s training process as soft constraints, becoming an important component of the scientific machine learning (SciML) ecosystem. More recently, physics-informed Kolmogorv-Arnold networks (PIKANs) have also shown to be effective and comparable in accuracy with PINNs. In their current implementation, both PINNs and PIKANs are mainly optimized using first-order methods like Adam, as well as quasi-Newton methods such as BFGS and its low-memory variant, L-BFGS. However, these optimizers often struggle with highly nonlinear and non-convex loss landscapes, leading to challenges such as slow convergence, local minima entrapment, and (non)degenerate saddle points. In this study, we investigate the performance of Self- Scaled BFGS (SSBFGS), Self-Scaled Broyden (SSBroyden) methods and other advanced quasi-Newton schemes, including BFGS and L-BFGS with different line search strategies. These methods dynamically rescale updates based on historical gradient information, thus enhancing training efficiency and accuracy. We systematically compare these optimizers – using both PINNs and PIKANs – on key challenging PDEs, including the Burgers, Allen-Cahn, Kuramoto-Sivashinsky, Ginzburg-Landau, and Stokes equations. Additionally, we evaluate the performance of SSBFGS and SSBroyden for Deep Operator Network (DeepONet) architectures, demonstrating their effectiveness for data-driven operator learning. Our findings provide state-of-the-art results with orders-of-magnitude accuracy improvements without the use of adaptive weights or any other enhancements typically employed in PINNs. More broadly, our work reveal insights into the effectiveness of quasi-Newton optimization strategies in significantly improving the convergence and accurate generalization of PINNs and PIKANs.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2023

Ruggedness Evaluation of ASTM International Standard Test Methods for Shape Memory Materials: E3098 Standard Test Method for Mechanical Uniaxial Pre-strain and Thermal Free Recovery of Shape Memory Alloys

This paper evaluates the ruggedness testing of the recently released ASTM International E3098 Standard Test Method for Mechanical Uniaxial Pre-strain and Thermal Free Recovery of Shape Memory Alloys. The ruggedness experiment was designed with eight runs in two replicates, consisting of seven factors: strain rate (ė), heating and cooling rates (Ṫ heat and Ṫ cool , respectively), upper and lower cycle temperatures (UCT and LCT, respectively), hold time (t hold ), and minimum load (F min ) imparted on the samples. The results indicate that the strain rate factor had no effect on any result variable, whereas the minimum load factor had a large effect on several result variables including the austenite start and martensite finish temperatures as well as most of the strain measurements from the thermal cycle. The lower cycle temperature factor had a large effect on the stress at maximum strain during pre-straining (e i ), denoted by S i , at ~7.2 MPa, as well as the A f , at 5.47 °C. Closely following the LCT effect, Ṫ heat was found to have a large effect on the austenite start temperature (A s ), at a value of ~2.72 °C. The testing methodology, analysis techniques, and resulting conclusions on the ruggedness of the test method are presented.

Shape memory alloys↗
DOE OSTI2021

Evaluating sealability of blended smart polymer and fiber additive for geothermal drilling with the effect of fracture opening size

Geothermal formations often contain extensive fracture networks. These fracture networks contribute to the significant loss of drilling fluids during geothermal drilling. Multiple loss circulation materials (LCM) such as fiber, granules, and pills have been proposed to tackle this problem but with only limited success. Recent advances in materials science have led to the development of thermoset shape memory polymers (SMP) to address the lost circulation problem. In this paper, we evaluate a thermoset SMP performance in sealing near wellbore fractures of different sizes in geothermal wells. The SMP performance was assessed using granite disks and cylindrical granite cores having fracture sizes of 1000 μm and 3000 μm. A static filtration test was performed using cedar fiber, CaCO 3 , and SMP. Results showed cedar fiber performed better than the CaCO 3 ., reducing fluid loss by 89% and improving sealing pressure by 200 psi. A novel dynamic testing unit that allows for high-temperature testing under flowing conditions was used in this study. The analysis showed that 3% by weight SMP and fiber blends could bridge and plug the 1000 μm fracture. For a larger fracture of 3000 μm width, there was a need to increase the weight concentration of the SMP to 6% to plug the fracture opening effectively. We showed the influence of key parameters such as the type of LCM, concentration, and particle size distribution in optimizing the performance of drilling fluid loss treatment.

02 PETROLEUM↗
DOE OSTI2018

New Insights into Shape Memory Alloy Bimorph Actuators Formed by Electron Beam Evaporation

In order to create shape memory alloy (SMA) bimorph microactuators with high-precision features, a novel fabrication process combined with electron beam (E-beam) evaporation, lift-off resist and isotropic XeF2 dry etching method was developed. To examine the effect of E-beam deposition and annealing process on nitinol (NiTi) characteristics, the NiTi thin film samples with different deposition rate and overflow conditions during annealing process were investigated. With the characterizations using scanning electron microscope and x-ray diffraction, the results indicated that low E-beam deposition rate and argon employed annealing process could benefit the formation of NiTi crystalline structure. In addition, SMA bimorph microactuators with high-precision features as small as 5 microns were successfully fabricated. Furthermore, the thermomechanical performance was experimentally verified and compared with finite element analysis simulation results.

36 MATERIALS SCIENCE↗
DOE OSTI2014

Thickness independent reduced forming voltage in oxygen engineered HfO{sub 2} based resistive switching memories

The conducting filament forming voltage of stoichiometric hafnium oxide based resistive switching layers increases linearly with layer thickness. Using strongly reduced oxygen deficient hafnium oxide thin films grown on polycrystalline TiN/Si(001) substrates, the thickness dependence of the forming voltage is strongly suppressed. Instead, an almost constant forming voltage of about 3 V is observed up to 200 nm layer thickness. This effect suggests that filament formation and switching occurs for all samples in an oxidized HfO{sub 2} surface layer of a few nanometer thickness while the highly oxygen deficient thin film itself merely serves as a oxygen vacancy reservoir.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2018

Complexity reduction effects on transient, atomic plasmas

Plasma simulations involving collisional-radiative processes present many computational challenges because of the wide range in scales spanned by the governing set of plasma physics equations. These numerical challenges are primarily caused by limitations in resources such as memory and computational time. Collisional-radiative models deliver a part of the challenge because of the large number of levels and transitions that are associated with atomic plasmas. Multiple complexity reduction techniques have been introduced in the past to overcome this obstacle, including the quasi-steady-state solution, uniform grouping, and the recently-devised Boltzmann grouping. These reduction schemes were tested against full numerical simulations of a simplified, time-dependent argon model, with and without an external Planckian radiation field. Lastly, results will show the advantages and disadvantages of using complexity reduction techniques when employed in various plasma regimes.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2018

Caloric effects in ferroic materials

The fundamentals and applications of ferroic materials$-$ferromagnetic, ferroelectric, and ferroelastic$-$are common subjects discussed in just about every graduate course related to functional materials. Looking beyond today’s traditional uses, such as in permanent magnets, capacitors, and shape-memory alloys, there are worthwhile and interesting questions common to the caloric properties of these ferroic materials. Can ferroic materials be used in a cooling cycle? Why are these materials susceptible to external fields? Which combination of properties is required to make some of them suitable for efficient cooling and heat pumping? We address these questions in this introduction to ferroic cooling, which comprises magnetocaloric, electrocaloric, elastocaloric and barocaloric approaches and combinations thereof (i.e., multicalorics). These are addressed in greater detail in the articles in this issue.

36 MATERIALS SCIENCE↗
DOE OSTI2014

Current-driven domain wall motion enhanced by the microwave field

The magnetic domain wall (DW) motion driven by a spin-polarized current opens a new concept for memory and logic devices. However, the critical current density required to overcome the intrinsic and/or extrinsic pinning of DW remains too large for practical applications. Here, we show, by using micromagnetic simulations and analytical approaches, that the application of a microwave field offers an effective solution to this problem. When a transverse microwave field is applied, the adiabatic spin-transfer torque (STT) alone can sustain a steady-state DW motion without the sign of Walker breakdown, meaning that the intrinsic pinning disappears. The extrinsic pinning can also be effectively reduced. Moreover, the DW velocity is increased greatly for the microwave-assisted DW motion. This provides a new way to manipulate the DW motion at low current densities.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2022

Estimation of the time for steam generator trip due to cyber intrusions

The time required to trip a pressurized water reactor (PWR) by inserting malicious signals into its steam generator (SG) control system has been studied using the Generic PWR (GPWR) Simulator. A semi-analytical model is developed to approximately reproduce the simulator response and understand the dynamics of the control unit. A series of two proportional-integral controllers determines control action according to preset constants, the readings from the feedwater level sensor, and those from feedwater and steam flowrate transmitters. It is observed that the most important factor that determines whether a trip will occur is how much additional water is added to or withheld from the SG over time compared to normal operating conditions. In order to determine the effects of control action on the SG, changes in mass inventory are considered. This approach models the SG water level as a function of mass inventory and has a backward temporal memory. A Python interface is developed for the GPWR framework to automatically simulate different spoofing scenarios and post-process the related data. We observe that the trip times predominantly depend on flow mismatch and/or level errors. Controller parameters, including the integral time and gain constants, either speed up or slow down the rate of progression to a trip setpoint but do not cause a trip by themselves. The reactor can trip on a high-level signal when the reading crosses above 78%, increased from its reference level of 57%, or a low-level reading when it is below 25%. The present results show roughly how long the operators would have to respond to an attack, given a specific set of spoofing signals within the issue space analyzed. Furthermore, we have generated a simple surface by fitting a combination of exponential functions to the data obtained from the GPWR Simulator. In general, trips on a low level have been observed to occur faster than those on a high level.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗