Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Gaussian Process”

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 1,621 records · Page 90

Two Mesoporous Domains Are Better Than One for Catalytic Deconstruction of Polyolefins

Catalytic hydrogenolysis of polyolefins into valuable liquid, oil, or wax-like hydrocarbon chains for second-life applications is typically accompanied by the hydrogen-wasting co-formation of low value volatiles, notably methane, that increase greenhouse gas emissions. Catalytic sites confined at the bottom of mesoporous wells, under conditions in which the pore exerts the greatest influence over the mechanism, are capable of producing less gases than unconfined sites. A new architecture was designed to emphasize this pore effect, with the active platinum nanoparticles embedded between linear, hexagonal mesoporous silica and gyroidal cubic MCM-48 silica (mSiO 2 /Pt/MCM-48). This catalyst deconstructs polyolefins selectively into ~C 20 –C 40 paraffins and cleaves C–C bonds at a rate (TOF = 4.2 ± 0.3 s –1 ) exceeding that of materials lacking these combined features while generating negligible volatile side products including methane. The time-independent product distribution is consistent with a processive mechanism for polymer deconstruction. In contrast to time- and polymer length-dependent products obtained from non-porous catalysts, mSiO 2 /Pt/MCM-48 yields a C 28 -centered Gaussian distribution of waxy hydrocarbons from polyolefins of varying molecular weight, composition, and physical properties, including low-density polyethylene, isotactic polypropylene, ultrahigh-molecular-weight polyethylene, and mixtures of multiple, post-industrial polyolefins. Here, coarse-grained simulation reveals that the porous-core architecture enables the paraffins to diffuse away from the active platinum site, preventing secondary reactions that produce gases.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

The physical mechanism on the threshold voltage temperature stability improvement for GaN HEMTs with pre-fluorination argon treatment

In this paper, a normally-off AlGaN/GaN MIS-HEMT with improved threshold voltage (V{sub TH}) thermal stability is reported with investigations on its physical mechanism. The normally-off operation of the device is achieved from novel short argon plasma treatment (APT) prior to the fluorine plasma treatment (FPT) on Al{sub 2}O{sub 3} gate dielectrics. For the MIS-HEMT with FPT only, its V{sub TH} drops from 4.2 V at room temperature to 0.5 V at 200 °C. Alternatively, for the device with APT-then-FPT process, its V{sub TH} can retain at 2.5 V at 200 °C due to the increased amount of deep-level traps that do not emit electrons at 200 °C. This thermally stable V{sub TH} makes this device suitable for high power applications. The depth profile of the F atoms in Al{sub 2}O{sub 3}, measured by the secondary ion mass spectroscopy, reveals a significant increase in the F concentration when APT is conducted prior to FPT. The X-ray photoelectron spectroscopy (XPS) analysis on the plasma-treated Al{sub 2}O{sub 3} surfaces observes higher composition of Al-F bonds if APT was applied before FPT. The enhanced breaking of Al-O bonds due to Ar bombardment assisted in the increased incorporation of F radicals at the surface during the subsequent FPT process. The Schrödinger equation of Al{sub 2}O{sub x}F{sub y} cells, with the same Al-F compositions as obtained from XPS, was solved by Gaussian 09 molecular simulations to extract electron state distribution as a function of energy. The simulation results show creation of the deeper trap states in the Al{sub 2}O{sub 3} bandgap when APT is used before FPT. Finally, the trap distribution extracted from the simulations is verified by the gate-stress experimental characterization to confirm the physical mechanism described.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Simulation and Modeling of Time-Resolved X-Ray Detector for the Saturn Accelerator

In this work, we present the technology-aided computer design (TCAD) device simulation and modeling of a silicon p-i-n diode for detecting time-dependent X-ray radiation. We show that the simulated forward and reverse breakdown current–voltage characteristics agree well with the measured data under nonradiation environment by only calibrating carrier lifetimes for the forward bias case and avalanche model critical fields for the reverse bias condition. Using the calibrated parameters and other nominal material properties, we simulated the radiation responses of the p-i-n diode and compared with experimental data when the diode was exposed to X-ray radiation at Sandia’s Saturn facility and the Idaho State University (ISU) TriMeV facility. For Saturn’s Gaussian dose-rate pulses, we show three findings from TCAD simulations. First, the simulated photocurrents are in excellent agreement with the measured data for two dose-rate pulses with peak values of 1.16×10 10 and 1.88×10 10 rad(Si)/s. Second, the simulation results of high dose-rate pulses predict increased delayed photocurrents with longer time tails in the diode electrical responses due to excess carrier generation. Third, simulated peak values of diode radiation responses versus peak dose rates at different bias conditions provide useful guidance to determine the dose-rate range that the p-i-n diode can reliably detect in experiment. For TriMeV’s non-Gaussian dose-rate pulse, our simulated diode response is in decent agreement with the measured data without further calibration. We also studied the effects of device geometry, recombination process, and dose-rate enhancement via TCAD simulations to understand the higher measured response in the time after the peak dose-rate radiation for the p-i-n diode exposed to TriMeV irradiation.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

CAFE AU LAIT: Compute-Aware Federated Augmented Low-Rank AI Training

Federated finetuning is crucial for unlocking the knowledge embedded in pretrained Large Language Models (LLMs) when data are geographically distributed across clients. Unlike finetuning with data from a single institution, federated finetuning allows collaboration across multiple institutions, enabling the utilization of diverse and decentralized datasets while preserving data privacy. Given the high computing costs of LLM training and the emphasis on energy efficiency in Federated Learning (FL), Low-Rank Adaptation (LoRA) has emerged as a widely adopted algorithm due to its significantly reduced number of trainable parameters. However, this assumes that all data silos have the necessary computing resources to compute local updates of LLMs. Nevertheless, in practice, the computing resources across clients are highly heterogeneous: while some may have access to hundreds of GPUs, others might have limited or no GPU access. Recently, federated finetuning using synthetic data has been proposed, allowing clients to participate in a collaborative training run without training LLMs locally. However, our experimental results reveal a performance gap between models trained using synthetic data and those trained using local updates. Motivated by the observed heterogeneity in computing resources and the performance gap, we propose a novel two-stage algorithm that leverages the storage and computing capabilities of a strong server. In the first stage, under the coordination of the strong server, clients with limited computing resources collaborate to generate synthetic data, which is transferred to and stored on the strong server. In the second stage, the strong server uses this synthetic data on behalf of the resource-constrained clients to perform federated LoRA finetuning alongside clients with sufficient computing resources. This approach ensures that all clients can participate in the finetuning process. Experimental results demonstrate that incorporating local updates from even a small fraction of clients improves performance compared to using synthetic data for all clients. Furthermore, we incorporate the Gaussian mechanism in both stages to guarantee client-level differential privacy.

Wang, Jiayi [ORNL]↗

Calculation of precision satellite orbits with nonsingular elements /VOP formulation/

Review of some results obtained in an effort to develop efficient, high-precision trajectory computation processes for artificial satellites by optimum selection of the form of the equations of motion of the satellite and the numerical integration method. In particular, the matching of a Gaussian variation-of-parameter (VOP) formulation is considered which is expressed in terms of equinoctial orbital elements and partially decouples the motion of the orbital frame from motion within the orbital frame. The performance of the resulting orbit generators is then compared with the popular classical Cowell/Gauss-Jackson formulation/integrator pair for two distinctly different orbit types - namely, the orbit of the ATS satellite at near-geosynchronous conditions and the near-circular orbit of the GEOS-C satellite at 1000 km.

Velez, C. E.↗

Statistics of Cirrus Horizontal Inhomogeneity in the Southern Great Plains

Variability of cloud properties on scales smaller than that of a GCM grid is potentially important both for realistic parameterizations of microphysical processes and for the prediction of the large-scale radiative effects of clouds, have suggested that a simple model of marine stratocumulus variability, based on the assumption of Gaussian variation statistics of cloud depth, can explain the liquid water path histogram shapes observed in Landsat data. In advance of ARM SGP MMCR ice water path climatologies, we have examined aircraft ice water content statistics for cirrus clouds observed over Coffeyville, Kansas during FIRE 2. We find similar associations of histogram shape and cloud cover for these clouds, and we show that a simple modification of the model for cirrus combined with observed mean cloud depths, their standard deviations, and ambient thermodynamic conditions predicts both the histogram shape and cirrus cloud cover fairly well. This suggests that subgrid variability of cloud properties may be similar for vastly different cloud types, and that a universal parameterization of the effects of subgrid variability in GCMs as a function of only a few parameters may be a realistic goal.

Smith, S. A.↗

[Using Molecular Simulations to Understand Complex Nanoscale Dynamic Phenomena in Polymer Solutions]

The first half of the project concentrated on molecular simulation studies of the translocation of model molecules for single-stranded DNA through a nanosized pore. This has resulted in the publication, Translocation of a polymer chain across a nanopore: A Brownian dynamics simulation study, by Pu Tian and Grant D. Smith, JOURNAL OF CHEMICAL PHYSICS VOLUME 119, NUMBER 21 1 DECEMBER 2003, which is attached to this report. In this work we carried out Brownian dynamics simulation studies of the translocation of single polymer chains across a nanosized pore under the driving of an applied field (chemical potential gradient) designed to mimic an electrostatic field. The translocation process can be either dominated by the entropic barrier resulted from restricted motion of flexible polymer chains or by applied forces (or chemical gradient). We focused on the latter case in our studies. Calculation of radius of gyration of the translocating chain at the two opposite sides of the wall shows that the polymer chains are not in equilibrium during the translocation process. Despite this fact, our results show that the one-dimensional diffusion and the nucleation model provide an excellent description of the dependence of average translocation time on the chemical potential gradients, the polymer chain length and the solvent viscosity. In good agreement with experimental results and theoretical predictions, the translocation time distribution of our simple model shows strong non-Gaussian characteristics. It is observed that even for this simple tube-like pore geometry, more than one peak of translocation time distribution can be generated for proper pore diameter and applied field strengths. Both repulsive Weeks-Chandler-Anderson and attractive Lennard-Jones polymer-nanopore interaction were studied. Attraction facilitates the translocation process by shortening the total translocation time and dramatically improve the capturing of polymer chain. The width of the translocation time distribution was found to decrease with increasing temperature, increasing field strength, and decreasing pore diameter.

Smith, Grant↗

GPR Imaging for Deeply Buried Objects: A Comparative Study Based on Compositing of Scanning Frequencies and a Chirp Excitation Function

Compositing of ground penetrating radar (GPR) scans of differing frequencies have been found to produce cleaner images at depth using the Gaussian mixture model (GMM) feature of the expectation-maximization (EM) algorithm. GPR scans at various heights (“Stand Off”), as well as ground-based scans, have been studied. In this paper, we compare the GPR response from a chirp excitation function-based radar with the response from the EM GMM algorithm compositing process, using the same mix of frequencies. A chirp excitation pulse was found to be effective in delineating the defined buried object, but the resulting image is less sharp than the GMM EM method.

58 GEOSCIENCES↗

Effects of nuclear orientation on fusion and fission process for reactions using {sup 238}U target nucleus

Fission fragment mass distributions in the reaction of {sup 30}Si+{sup 238}U were measured at the energies around the Coulomb barrier. At the above-barrier energies, the mass distribution showed Gaussian shape. At the sub-barrier energies, asymmetric fission mode peaked at A{sub L}/A{sub H}approx =90/178 was observed. The asymmetric fission should be attributed to quasifission from the results of the measured evaporation residue (ER) cross-sections produced by {sup 30}Si+{sup 238}U. The cross-section for {sup 263}Sg at the above-barrier energy agree with the statistical model calculation which assumes that the measured fission cross-sections are equal to the fusion cross-sections, whereas the one for {sup 264}Sg measured at the sub-barrier energy is smaller than the calculation, indicating the presence for quasifission. The fragment mass distributions are compared to those for {sup 36}S+{sup 238}U and {sup 40}Ar+{sup 238}U.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Operator learning for predicting multiscale bubble growth dynamics

We report simulating and predicting multiscale problems that couple multiple physics and dynamics across many orders of spatiotemporal scales is a great challenge that has not been investigated systematically by deep neural networks (DNNs). Herein, we develop a framework based on operator regression, the so-called deep operator network (DeepONet), with the long-term objective to simplify multiscale modeling by avoiding the fragile and time-consuming “hand-shaking” interface algorithms for stitching together heterogeneous descriptions of multiscale phenomena. To this end, as a first step, we investigate if a DeepONet can learn the dynamics of different scale regimes, one at the deterministic macroscale and the other at the stochastic microscale regime with inherent thermal fluctuations. Specifically, we test the effectiveness and accuracy of the DeepONet in predicting multirate bubble growth dynamics, which is described by a Rayleigh–Plesset (R–P) equation at the macroscale and modeled as a stochastic nucleation and cavitation process at the microscale by dissipative particle dynamics (DPD). First, we generate data using the R–P equation for multirate bubble growth dynamics caused by randomly time-varying liquid pressures drawn from Gaussian random fields (GRFs). Our results show that properly trained DeepONets can accurately predict the macroscale bubble growth dynamics and can outperform long short-term memory networks. We also demonstrate that the DeepONet can extrapolate accurately outside the input distribution using only very few new measurements. Subsequently, we train the DeepONet with DPD data corresponding to stochastic bubble growth dynamics. Although the DPD data are noisy and we only collect sparse data points on the trajectories, the trained DeepONet model is able to predict accurately the mean bubble dynamics for time-varying GRF pressures. Taken together, our findings demonstrate that DeepONets can be employed to unify the macroscale and microscale models of the multirate bubble growth problem, hence providing new insight into the role of operator regression via DNNs in tackling realistic multiscale problems and in simplifying modeling with heterogeneous descriptions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

First Step Towards a Non-Adiabatic Description of the Fission Process Based on the Generator Coordinate Method

Among the different theoretical approaches able to describe fission, microscopic ones can help us in the understanding of this process, as they have the advantage of describing the nuclear structure and the dynamics in a consistent manner. The sole input of the calculations is the nucleon-nucleon interaction. Such a microscopic time-dependent and quantum mechanical formalism has already been used, based on the Gaussian Overlap Approximation of the Generator Coordinate Method with the adiabatic approximation, to analyze the collective dynamics of low-energy fission in {sup 238}U. However, at higher energies, a few MeV above the barrier, the adiabatic approximation doesn't seem valid anymore. Indeed, manifestations of proton pair breaking have been observed in {sup 238}U and {sup 239}U for an excitation energy of 2.3 MeV above the barrier. Taking the intrinsic excitations into account during the fission process will enable us to determine the coupling between collective and intrinsic degrees of freedom, in particular from saddle to scission. Guidelines of the new formalism under development are presented and some preliminary results on overlaps between non excited and excited states are discussed.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Nonrad: Computing nonradiative capture coefficients from first principles

Point defects in semiconductor crystals provide a means for carriers to recombine nonradiatively. This recombination process impacts the performance of devices. We present the Nonrad code that implements the first-principles approach of Alkauskas et al. (2014) [8] for the evaluation of nonradiative capture coefficients based on a quantum-mechanical description of the capture process. An approach for evaluating electron-phonon coupling within the projector augmented wave formalism is presented. We also show that the common procedure of replacing Dirac delta functions with Gaussians can introduce errors into the resulting capture rate, and implement an alternative scheme to properly account for vibrational broadening. Finally, we assess the accuracy of using an analytic approximation to the Sommerfeld parameter by comparing with direct numerical evaluation.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Less is More: Bigger Data from Compressive Measurements

Compressive sensing approaches are beginning to take hold in (scanning) transmission electron microscopy (S/TEM) [1,2,3]. Compressive sensing is a mathematical theory about acquiring signals in a compressed form (measurements) and the probability of recovering the original signal by solving an inverse problem [4]. The inverse problem is underdetermined (more unknowns than measurements), so it is not obvious that recovery is possible. Compression is achieved by taking inner products of the signal with measurement weight vectors. Both Gaussian random weights and Bernoulli (0,1) random weights form a large class of measurement vectors for which recovery is possible. The measurements can also be designed through an optimization process. The key insight for electron microscopists is that compressive sensing can be used to increase acquisition speed and reduce dose. Building on work initially developed for optical cameras, this new paradigm will allow electron microscopists to solve more problems in the engineering and life sciences. We will be collecting orders of magnitude more data than previously possible. The reason that we will have more data is because we will have increased temporal/spatial/spectral sampling rates, and we will be able ability to interrogate larger classes of samples that were previously too beam sensitive to survive the experiment. For example consider an in-situ experiment that takes 1 minute. With traditional sensing, we might collect 5 images per second for a total of 300 images. With compressive sensing, each of those 300 images can be expanded into 10 more images, making the collection rate 50 images per second, and the decompressed data a total of 3000 images [3]. But, what are the implications, in terms of data, for this new methodology? Acquisition of compressed data will require downstream reconstruction to be useful. The reconstructed data will be much larger than traditional data, we will need space to store the reconstructions during analysis, and the computational demands for analysis will be higher. Moreover, there will be time costs associated with reconstruction. Deep learning [5] is an approach to address these problems. Deep learning is a hierarchical approach to find useful (for a particular task) representations of data. Each layer of the hierarchy is intended to represent higher levels of abstraction. For example, a deep model of faces might have sinusoids, edges and gradients in the first layer; eyes, noses, and mouths in the second layer, and faces in the third layer. There has been significant effort recently in deep learning algorithms for tasks beyond image classification such as compressive reconstruction [6] and image segmentation [7]. A drawback of deep learning, however, is that training the model requires large datasets and dedicated computational resources (to reduce training time to a few days). A second issue is that deep learning is not user-friendly and the meaning behind the results is usually not interpretable. We have shown it is possible to reduce the data set size while maintaining model quality [8] and developed interpretable models for image classification [9], but the demands are still significant. The key to addressing these problems is to NOT reconstruct the data. Instead, we should design computational sensors that give answers to specific problems. A simple version of this idea is compressive classification [10], where the goal is to classify signal type from a small number of compressed measurements. Classification is a much simpler problem than reconstruction, so 1) much fewer measurements will be necessary, and 2) these measurements will probably not be useful for reconstruction. Other simple examples of computational sensing include determining object volume or the number of objects present in the field of view [11].

Stevens, Andrew↗

Time delay of wave packets during their tunnelling through a quantum diode

A modified saddle-point method is used to investigate the process of propagation of a wave packet through a quantum diode. A scattering matrix is constructed for the structure in question. The case of tunnelling of a packet with a Gaussian envelope through the diode is considered in detail. The time delay and the shape of the wave packet transmitted are calculated. The dependence of the delay time on the characteristics of the input packet and the internal characteristics of the quantum diode is studied. Possible applications of the results obtained are discussed. (laser applications and other topics in quantum electronics)

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

IMFIT: A FAST, FLEXIBLE NEW PROGRAM FOR ASTRONOMICAL IMAGE FITTING

I describe a new, open-source astronomical image-fitting program called IMFIT, specialized for galaxies but potentially useful for other sources, which is fast, flexible, and highly extensible. A key characteristic of the program is an object-oriented design that allows new types of image components (two-dimensional surface-brightness functions) to be easily written and added to the program. Image functions provided with IMFIT include the usual suspects for galaxy decompositions (Sérsic, exponential, Gaussian), along with Core-Sérsic and broken-exponential profiles, elliptical rings, and three components that perform line-of-sight integration through three-dimensional luminosity-density models of disks and rings seen at arbitrary inclinations. Available minimization algorithms include Levenberg-Marquardt, Nelder-Mead simplex, and Differential Evolution, allowing trade-offs between speed and decreased sensitivity to local minima in the fit landscape. Minimization can be done using the standard χ{sup 2} statistic (using either data or model values to estimate per-pixel Gaussian errors, or else user-supplied error images) or Poisson-based maximum-likelihood statistics; the latter approach is particularly appropriate for cases of Poisson data in the low-count regime. I show that fitting low-signal-to-noise ratio galaxy images using χ{sup 2} minimization and individual-pixel Gaussian uncertainties can lead to significant biases in fitted parameter values, which are avoided if a Poisson-based statistic is used; this is true even when Gaussian read noise is present.

79 ASTRONOMY AND ASTROPHYSICS↗

Raytracing the long trace profiler

The optical design of the Long Trace Profiler optical system is explored with a commercial raytrace program, Zemax OpticStudio™ (ZOS)1, with the intent of finding and correcting sources of systematic error. ZOS provides both geometric raytracing tools and physical optics Gaussian beam propagation and diffraction image calculation tools, and two design modes, sequential component (SC) and non-sequential component (NSC) that are optimized for different aspects of the design process. The original LTP-II system employs a singlet lens with a 1250 mm focal length. It is optimized to provide minimum distortion over a surface slope angle range of ±5 mrad. Using the ZOS tools, we are able to simulate ghost rays that produce distortion in the beam spot image and can minimize the distortion by deliberate misalignment of the beamsplitter (BS) components. Unfortunately, the reference beam is compromised because of the component tilts. The most recent LTP500 system design simplifies the optical system and makes the reference beam usable again, even with misalignment of the polarizing beamsplitter (PBS). Two lenses are designed for the LTP500 - a cemented doublet that has been fabricated, and a singlet with one aspheric surface. Both have focal distances of 500 mm with an expanded angular measurement range of ±10 mrad. The aspheric singlet provides superior performance. ZOS allows the import of wavefront measurement data produced by commercial interferometer software. We apply the wavefront error measurement from the cemented doublet to the model to show that the 19nm RMS wavefront error needs to be improved by at least a factor of ten in order to reduce the systematic error to a level that will allow the LTP to approach its design limit of a few tens of nanoradians.

Takacs, Peter Z.↗

UQ Toolkit v 2.0

The Uncertainty Quantification (UQ) Toolkit is a software library for the characterizaton and propagation of uncertainties in computational models. For the characterization of uncertainties, Bayesian inference tools are provided to infer uncertain model parameters, as well as Bayesian compressive sensing methods for discovering sparse representations of high-dimensional input-output response surfaces, and also Karhunen-Loève expansions for representing stochastic processes. Uncertain parameters are treated as random variables and represented with Polynomial Chaos expansions (PCEs). The library implements several spectral basis function types (e.g. Hermite basis functions in terms of Gaussian random variables or Legendre basis functions in terms of uniform random variables) that can be used to represent random variables with PCEs. For propagation of uncertainty, tools are provided to propagate PCEs that describe the input uncertainty through the computational model using either intrusive methods (Galerkin projection of equations onto basis functions) or non-intrusive methods (perform deterministic operation at sampled values of the random values and project the obtained results onto basis functions).

Safta, Cosmin↗