Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “gradient sampling”

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 217 records · Page 12
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Reverse-mode differentiation in arbitrary tensor network format: with application to supervised learning.

This paper describes an efficient reverse-mode differentiation algorithm for contraction operations of tensor networks that may have arbitrary and unconventional network topologies. The approach leverages the tensor contraction tree of Evenbly and Pfeifer (2014), which provides an instruction set for the contraction sequence of a network. We show that this tree can be efficiently leveraged for differentiation of a full tensor network contraction using a recursive scheme that exploits (1) the bilinear property of contraction and (2) the property that trees have single path from root to leaves. While differentiation of tensor-tensor contraction is already possible in most automatic differentiation packages, we show that exploiting these two additional properties in the specific context of contraction sequences can improve efficiency. Following a description of the algorithm and computational complexity analysis, we investigate its utility for gradient-based supervised learning for low-rank function recovery and for fitting real-world unstructured datasets. We demonstrate improved performance over alternating least-squares optimization approaches and the capability to handle heterogeneous and arbitrary tensor network formats. When compared to alternating minimization algorithms, we find that the gradient-based approach requires a smaller oversampling ratio (number of samples compared to number model parameters) for recovery. This increased efficiency extends to fitting unstructured data of varying dimensionality and when employing a variety of tensor network formats. Here, we show improved learning using the hierarchical Tucker method over the tensor-train in high-dimensional settings on a number of benchmark problems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

High Throughput Coefficient Thermal Expansion Testing Utilizing Digital Image Correlation

Dr. Fitzgerald, a postdoc at Sandia National Laboratories, works in a materials of mechanics group characterizing material properties of ductile materials. Her presentation focuses specifically on increasing throughput of coefficient of thermal expansion (CTE) measurements with the use of optical strain measurements, called digital image correlation (DIC). Currently, the coefficient of thermal expansion is found through a time intensive process called dilatometry. There are multiple types of dilatometers. One type, a double push rod mechanical dilatometer, uses and LVDT to measure the expansion of a specimen in one direction. It uses a reference material with known properties to determine the CTE of the specimen in question. Testing about 500 samples using the double push rod mechanical dilatometer would take about 2 years if testing Monday through Friday, because the reference material needs to be at a constant temperature and heating must done slowly to ensure no thermal gradients across the rod. A second type, scissors type dilatometer, pinches a sample using a “scissor-like” appendage that also uses a LVDT to measure thermal expansion as the sample is heated. Finally, laser dilatometry, was created to provide a non-contact means to measure thermal expansion. This process greatly reduces the time required to setup a measurement but is still only able to measure one sample at a time. The time required to test 500 samples gets reduced to 3.5 weeks. Additionally, to measure expansion in different directions, multiple lasers must be used. Dr. Fitzgerald solved this conundrum by using an optical measurement technique called digital image correlation to create strain maps in multiple orientations as well as measuring multiple samples at once. Using this technique, Dr. Fitzgerald can test 500 samples, conservatively, in 2 days.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Synthesis of bulk reactive Ni–Al composites using high pressure torsion

Self-propagating exothermic reactions, for instance in the nickel-aluminum (Ni-Al) system, have been widely studied to create high performance intermetallic compounds or for in-situ welding. Their easy ignition once the phase spacing is reduced below the micron scale, makes top-down methods like high-energy ball milling, ideal to fabricate such reactive nanostructures. A major drawback of ball milling is the need of a sintering step to form bulk pieces of the reactive material. However, this is not possible, as the targeted reactions would already proceed. Therefore, we investigate the ability of high pressure torsion as an alternative process, capable to produce bulk nanocomposites from powder mixtures. Severe straining of powder mixtures with a composition of 50 wt% Ni and 50 wt% Al enables fabrication of self-reactive bulk samples with microstructures similar to those obtained from ball milling or magnetron sputtering. Additionally, samples deformed at ambient temperature are highly reactive and can be ignited significantly below the Al melting point, finally predominantly consisting of Al 3 Ni 2 and Al 3 Ni, independent of the applied strain. Although the reaction proceeds first at the edge of the disk, the strain gradient present in the disks does not prevent reaction of the whole sample.

36 MATERIALS SCIENCE↗
DOE OSTI2023

A Predictive Prescription Framework for Stochastic Unit Commitment Using Boosting Ensemble Learning Algorithms

To take unit commitment (UC) decisions under uncertain load, most existing stochastic optimization (SO) frameworks adopt a generic representation of uncertainty. While load levels that materialize on a particular day are influenced by various covariates (such as the day of the week or temperature), SO frameworks typically disregard such side observations, wasting actionable information that could significantly enhance decision quality. Here, this article proposes a contextual SO (CSO) framework for UC under uncertain load, which can effectively exploit covariate observations in conjunction with a class of machine learning (ML) algorithms to improve the out-of-sample performance of UC decisions. It shows how three ML algorithms, adaptive boosting, gradient boosted trees, and extreme gradient boosting, can be used to this end, constituting the first application of these algorithms in any CSO framework. Using real-world data harvested from the New York ISO grid, we measure the out-of-sample performance of the framework in terms of total operation cost, shed load values, locational marginal prices, and total payments by the loads, against several benchmark methods proposed in the literature. The article has an online companion (Yurdakul et al.), wherein we present additional results and lay out further mathematical formulations used in this work.

42 ENGINEERING↗
DOE OSTI2024

Experimental and Computational Studies of Crystal Nucleation in Composition Gradients (Final Report)

The major goals of this project were to develop and validate a predictive nucleation model that incorporates composition gradients and is applicable to a large range of metallic systems which form both stochiometric and non-stochiometric compounds. Computationally we planned to expand the extent of thermodynamics-based theories and lay the groundwork to improve the general understanding and predictive capabilities of the role that gradients play in phase formation, glass formability and stability, and nucleation and growth events. To address these goals, we used Molecular Dynamics (MD) simulations and both isothermal and isochronal nanocalorimetric experiments on amorphous phases with a controlled composition gradient with the hopes of validating and improving the classical nucleation model and its use in solid solutions. While the computations were successful and identified ways to improve the classical nucleation model, the in situ nanocalorimetry studies proved very challenging due to unexpected difficulties in fabricating effective calorimeters and samples. Thus, we could not experimentally validate our predicted influence of composition gradients on nucleation. Nonetheless, important insights were gained and modifications to the classical nucleation theory were suggested.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Gas-phase metallicity determinations in nearby AGNs with SDSS-IV MaNGA: evidence of metal-poor accretion

ABSTRACT We derive the metallicity (traced by the O/H abundance) of the narrow-line region (NLR) of 108 Seyfert galaxies as well as radial metallicity gradients along their galaxy discs and of these of a matched control sample of no active galaxies. In view of that, observational data from the SDSS-IV MaNGA survey and strong emission-line calibrations taken from the literature were considered. The metallicity obtained for the NLRs was compared to the value derived from the extrapolation of the radial oxygen abundance gradient, obtained from H ii region estimates along the galaxy disc, to the central part of the host galaxies. We find that, for most of the objects ($\sim 80\, {{\ \rm per\ cent}}$), the NLR metallicity is lower than the extrapolated value, with the average difference ($\langle$D$\rangle$) between these estimates ranging from 0.16 to 0.30 dex. We suggest that $\langle$D$\rangle$ is due to the accretion of metal-poor gas to the AGN that feeds the nuclear supermassive black hole (SMBH), which is drawn from a reservoir molecular and/or neutral hydrogen around the SMBH. Additionally, we look for correlations between D and the electron density (Ne), [O iii]λ5007, and H α luminosities, extinction coefficient (AV) of the NLRs, as well as the stellar mass (M*) of the host galaxies. Evidence of an inverse correlation between the D and the parameters Ne, M*, and Av was found.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Absence of Barren Plateaus and Scaling of Gradients in the Energy Optimization of Isometric Tensor Network States

Abstract Vanishing gradients can pose substantial obstacles for high-dimensional optimization problems. Here we consider energy minimization problems for quantum many-body systems with extensive Hamiltonians and finite-range interactions, which can be studied on classical computers or in the form of variational quantum eigensolvers on quantum computers. Barren plateaus correspond to scenarios where the average amplitude of the energy gradient decreases exponentially with increasing system size. This occurs, for example, for quantum neural networks and for brickwall quantum circuits when the depth increases polynomially in the system size. Here we prove that the variational optimization problems for matrix product states, tree tensor networks, and the multiscale entanglement renormalization ansatz are free of barren plateaus. The derived scaling properties for the gradient variance provide an analytical guarantee for the trainability of randomly initialized tensor network states (TNS) and motivate certain initialization schemes. In a suitable representation, unitary tensors that parametrize the TNS are sampled according to the uniform Haar measure. We employ a Riemannian formulation of the gradient based optimizations which simplifies the analytical evaluation.

Barthel, Thomas↗
DOE OSTI2022

Estimating Drizzle Parameters by Aircraft Sampling

In-cloud supersaturation is calculated from aircraft derived measurements of the in-cloud vertical gradient of liquid water content, vertical velocity, the pt and 3 rd cloud droplet radius moments and a parameter related to the heat and mass transport during condensation and evaporation of cloud droplets. The aircraft data where taken during the VOCALS 2008 study in marine stratocumulus clouds off the coast of Chile. Approximately 16000 discrete samples, 2.5 meter in length, were used in the present analysis of in-cloud supersaturation and its variability, both of which are indicative of turbulent processes within the cloud and their significant impact on drizzle formation. Two additional drizzle parameters, the eddy persistence time and the variance of fluctuations in in-cloud droplet growth rate were also estimated from this data set. These results will help to constrain parameterizations of underlying turbulent microphysical processes of clouds and thereby advance the future development of warm cloud and precipitation models for climate simulation.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Scalable Techniques for Stochastic Power Flow Problems (Final Report)

The proposed research focuses on developing scalable algorithms for two-stage security-constrained OPF problems with AC power flow constraints, a class of problems complicated by (i) scale arising from a scenario representation; and (ii) the presence of nonlinearity, nonconvexity, and possibly second-stage discreteness or complementarity. Unfortunately, most existing solvers cannot contend with both challenges simultaneously; accordingly, the proposed research focuses on developing solution techniques that can both scale with the number of scenarios and contend with nonconvexity and second-stage complementarity. We consider three avenues for addressing such problems: (i) Variable sample-size SQP (VS-SQP) methods that combine sparse Quasi-Newton updates with a scalable variance-reduced stochastic gradient scheme for stochastic QP subproblems, allowing for contending with second-stage complementarity via regularization; (ii) Variable sample-size stochastic Interior-point (VS-sIP) schemes that propose a sampling-based regularized (to allow for contending with complementarity) interior-point schemes in which a Schur-complement technique is employed for decomposing the Newton direction computation step; (iii) Variable sample-size tractable ADMM (VS-tADMM) schemes combine variable sample-sizes with carefully designed techniques for resolving each of the nonconvex updates (by leveraging the QCQP structures). We intend to compare the three schemes using performance profiles in terms of solution quality, scalability, etc. and then select one scheme which will then be developed and further refined in Python for purposes of the GO competition.

42 ENGINEERING↗
DOE OSTI2023

Gray matter volume drives the brain age gap in schizophrenia: a SHAP study

Neuroimaging-based brain age is a biomarker that is generated by machine learning (ML) predictions. The brain age gap (BAG) is typically defined as the difference between the predicted brain age and chronological age. Studies have consistently reported a positive BAG in individuals with schizophrenia (SCZ). However, there is little understanding of which specific factors drive the ML-based brain age predictions, leading to limited biological interpretations of the BAG. We gathered data from three publicly available databases - COBRE, MCIC, and UCLA - and an additional dataset (TOPSY) of early-stage schizophrenia (82.5% untreated first-episode sample) and calculated brain age with pre-trained gradient-boosted trees. Then, we applied SHapley Additive Explanations (SHAP) to identify which brain features influence brain age predictions. We investigated the interaction between the SHAP score for each feature and group as a function of the BAG. These analyses identified total gray matter volume (group × SHAP interaction term β = 1.71 [0.53; 3.23]; p corr < 0.03) as the feature that influences the BAG observed in SCZ among the brain features that are most predictive of brain age. Other brain features also presented differences in SHAP values between SCZ and HC, but they were not significantly associated with the BAG. We compared the findings with a non-psychotic depression dataset (CAN-BIND), where the interaction was not significant. This study has important implications for the understanding of brain age prediction models and the BAG in SCZ and, potentially, in other psychiatric disorders.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2022

Nonlinear dimension reduction for surrogate modeling using gradient information

We introduce a method for the nonlinear dimension reduction of a high-dimensional function $u:{\mathbb{R}}^d\rightarrow{\mathbb{R}}$, $d\gg 1$. Our objective is to identify a nonlinear feature map $g:{\mathbb{R}}^d\rightarrow{\mathbb{R}}^m$, with a prescribed intermediate dimension $m\ll d$, so that $u$ can be well approximated by $f\circ g$ for some profile function $f:{\mathbb{R}}^m\rightarrow{\mathbb{R}}$. We propose to build the feature map by aligning the Jacobian $\nabla g$ with the gradient $\nabla u$, and we theoretically analyze the properties of the resulting $g$. Once $g$ is built, we construct $f$ by solving a gradient-enhanced least squares problem. Our practical algorithm uses a sample $\{{\textbf{x}}^{(i)},u({\textbf{x}}^{(i)}),\nabla u({\textbf{x}}^{(i)})\}_{i=1}^N$ and builds both $g$ and $f$ on adaptive downward-closed polynomial spaces, using cross validation to avoid overfitting. We numerically evaluate the performance of our algorithm across different benchmarks, and explore the impact of the intermediate dimension $m$. We show that building a nonlinear feature map $g$ can permit more accurate approximation of $u$ than a linear $g$, for the same input data set.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Deep Neural Network Informed Markov Chain Monte Carlo Methods

In subsurface flow modeling, quantifying the uncertainty of model parameters and the corresponding uncertainly on output quantities is a crucial task for groundwater management. Markov chain Monte Carlo (MCMC) methods can take advantage of observed data to estimate parameters in a Bayesian setting. However, MCMC can be slow to converge and produce highly correlated samples when the dimensions of the parameters is high. Using gradients for the posterior distribution can help samplers explore the parameter space more efficiently, but obtaining gradients can be computationally challenging.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

In Situ Non-Destructive Temporal Measurements of the Rhizosphere Microbiome ‘Hot-Spots’ Using Metaproteomics

Rhizosphere arguably embodies the most diverse microbial ecosystem on the planet, yet it is largely a functional ‘black box’ of belowground plant-microbiome interactions. The rhizosphere is the primary site of entry for subsurface injection of fixed carbon (C) into soil with impacts on local to global scale C biogeochemistry and ultimately Earth’s climate. While spatial organization of rhizosphere is central to its function, small scale and steep microbial and geochemical gradients within this dynamic region make it easily disrupted by sampling. The significant challenge presented by sampling blocks elucidation of discreet functions, drivers, and interactions within rhizosphere ecosystems. Here, we describe a non-destructive sampling method linked to metaproteomic analysis in order to measure temporal shifts in the microbial composition and function of rhizosphere. A robust, non-destructive method of sampling microbial hotspots within rhizosphere provides an unperturbed window into the elusive functional interactome of this system over time and space.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Deep probabilistic direction prediction in 3D with applications to directional dark matter detectors

Abstract We present the first method to probabilistically predict 3D direction in a deep neural network model. The probabilistic predictions are modeled as a heteroscedastic von Mises-Fisher distribution on the sphere S 2 , giving a simple way to quantify aleatoric uncertainty. This approach generalizes the cosine distance loss which is a special case of our loss function when the uncertainty is assumed to be uniform across samples. We develop approximations required to make the likelihood function and gradient calculations stable. The method is applied to the task of predicting the 3D directions of electrons, the most complex signal in a class of experimental particle physics detectors designed to demonstrate the particle nature of dark matter and study solar neutrinos. Using simulated Monte Carlo data, the initial direction of recoiling electrons is inferred from their tortuous trajectories, as captured by the 3D detectors. For 40 keV electrons in a 70% He 30% CO 2 gas mixture at STP, the new approach achieves a mean cosine distance of 0.104 (26 ∘ ) compared to 0.556 (64 ∘ ) achieved by a non-machine learning algorithm. We show that the model is well-calibrated and accuracy can be increased further by removing samples with high predicted uncertainty. This advancement in probabilistic 3D directional learning could increase the sensitivity of directional dark matter detectors.

Computer Science↗
DOE OSTI2022

Microfluidic liquid sheets as large-area targets for high repetition XFELs

The high intensity of X-ray free electron lasers (XFELs) can damage solution-phase samples on every scale, ranging from the molecular or electronic structure of a sample to the macroscopic structure of a liquid microjet. By using a large surface area liquid sheet microjet as a sample target instead of a standard cylindrical microjet, the incident X-ray spot size can be increased such that the incident intensity falls below the damage threshold. This capability is becoming particularly important for high repetition rate XFELs, where destroying a target with each pulse would require prohibitively large volumes of sample. We present here a study of microfluidic liquid sheet dimensions as a function of liquid flow rate. Sheet lengths, widths and thickness gradients are shown for three styles of nozzles fabricated from isotropically etched glass. In-vacuum operation and sample recirculation using these nozzles is demonstrated. The effects of intense XFEL pulses on the structure of a liquid sheet are also briefly examined.

47 OTHER INSTRUMENTATION↗
DOE OSTI2026

Score-based deterministic density sampling

We propose a deterministic sampling framework using Score-Based Transport Modeling for sampling an unnormalized target density π given only its score ∇ log π. Our method approximates the Wasserstein gradient flow on KL($f_t$∥π) by learning the time-varying score ∇ log $f_t$ on the fly using score matching. While having the same marginal distribution as Langevin dynamics, our method produces smooth deterministic trajectories, resulting in monotone noise-free convergence. We prove that our method dissipates relative entropy at the same rate as the exact gradient flow, provided sufficient training. Numerical experiments validate our theoretical findings: our method converges at the optimal rate, has smooth trajectories, and is often more sample efficient than its stochastic counterpart. Experiments on high-dimensional image data show that our method produces high-quality generations in as few as 15 steps and exhibits natural exploratory behavior. The memory and runtime scale linearly in the sample size.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Predicting the effects of thermally-induced density gradients on the hydrodynamic behavior of PBX 9502

High explosives are often exposed to thermal and mechanical stimuli within normal and abnormal environments. These stimuli can introduce density gradients in the explosive charge but the effect of such gradients is not well-understood. The goal of this work is to develop a comprehensive methodology to model the evolution of density gradients under various thermal conditions and then to predict shock initiation and detonation behavior of the gradated sample. PBX 9502 was chosen for these scoping studies because it has a large coefficient of thermal expansion and because the available shock initiation data indicates that initial temperature strongly affects the shock-to-detonation transition (SDT). The multiphysics code COMSOL was first used to model induced density gradients in a slab of PBX 9502 with asymmetric heating. Then, the gradated sample was constructed in the Pagosa hydrocode to simulate steel flyer impact scenarios, with SDT and detonation propagation following a calibrated Scaled Uniform Reactive Front (SURF) burn model. Furthermore, the gradated material exhibited very different initiation and corner turning behavior compared to homogeneous materials, and the orientation of the gradation was also found to have some effect. The methodology used here can estimate thermomechanical response of the material for non-uniform thermal environments and compute the resulting detonation properties of the PBX.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2020

Evaluating the Ratio of Electron and Hole Mobilities from a Single Bulk Sample Using Photo-Seebeck Effect

When a semiconductor is under photoexcitation, the voltage response to a temperature gradient is the photo-Seebeck effect. In this work, we study this effect, focusing on the contribution from transport of photo-excited carriers. We demonstrate that by combining photo-Seebeck with photoconductivity measurements, one can determine the ratio between electron and hole mobilities, and hence both of them when one is known. This is found for the case of defect-free samples, where no detail on the absorbance, carrier lifetime or recombination is necessary. Our method reported here does not require chemical doping, which could introduce defects and is often not feasible. It applies to both thin film and bulk samples. Experiment wise, photo-Seebeck effect is relatively easy to implement, or added to existing systems. In a broader context, for semiconductors with significant influence from defects, our result suggests that the photo-Seebeck behavior can still be understood. In this case another photo-transport property is necessary, in order to identify the mobilities of carriers and information regarding the defects. This framework integrates the information from photoexcitation and thermal gradients to provide a general method to determine fundamental electronic properties of materials.

36 MATERIALS SCIENCE↗