Classification of features in a scene typically requires conversion of the incoming photonic field int the electronic domain. Recently, an alternative approach has emerged whereby passive structured materials can perform classification tasks by directly using free-space propagation and diffraction of light. In this manuscript, we present a theoretical and computational study of such systems and establish the basic features that govern their performance. We show that system architecture, material structure, and input light field are intertwined and need to be co-designed to maximize classification accuracy. Our simulations show that a single layer metasurface can achieve classification accuracy better than conventional linear classifiers, with an order of magnitude fewer diffractive features than previously reported. For a wavelength λ, single layer metasurfaces of size 100λ x 100λ with aperture density λ-2 achieve ~96% testing accuracy on the MNIST dataset, for an optimized distance ~100λ to the output plane. This is enabled by an intrinsic nonlinearity in photodetection, despite the use of linear optical metamaterials. Furthermore, we find that once the system is optimized, the number of diffractive features is the main determinant of classification performance. The slow asymptotic scaling with the number of apertures suggests a reason why such systems may benefit from multiple layer designs. Finally, we show a trade-off between the number of apertures and fabrication noise.
Broadly applicable solutions to multimodal and multisensory fusion problems across domains remain a challenge because effective solutions often require substantive domain knowledge and engineering. The chief questions that arise for data fusion are in when to share information from different data sources, and how to accomplish the integration of information. The solutions explored in this work remain agnostic to input representation and terminal decision fusion approaches by sharing information through the learning objective as a compound objective function. The objective function this work uses assumes a one-to-one learning paradigm within a one-to-many domain which allows the assumption that consistency can be enforced across the one-to-many dimension. The domains and tasks we explore in this work include multi-sensor fusion for seismic event location and multimodal hyperspectral target discrimination. We find that our domain- informed consistency objectives are challenging to implement in stable and successful learning because of intersections between inherent data complexity and practical parameter optimization. While multimodal hyperspectral target discrimination was not enhanced across a range of different experiments by the fusion strategies put forward in this work, seismic event location benefited substantially, but only for label-limited scenarios.
Aimone, James B.; Date, Prasanna; Fonseca-Guerra, Gabriel A.; Hamilton, Kathleen E.; Henke, Kyle; Kay, Bill; Kenyon, Garrett T.; Kulkarni, Shruti R.; Parsa, Maryam; Schuman, Catherine D.; Severa, William M.; Smith, John D.
Though neuromorphic computers have typically targeted applications in machine learning and neuroscience (‘cognitive’ applications), they have many computational characteristics that are attractive for a wide variety of computational problems. In this work, we review the current state-of-the-art for non-cognitive applications on neuromorphic computers, including simple computational kernels for composition, graph algorithms, constrained optimization, and signal processing. We discuss the advantages of using neuromorphic computers for these different applications, as well as the challenges that still remain. The ultimate goal of this work is to bring awareness to this class of problems for neuromorphic systems to the broader community, particularly to encourage further work in this area and to make sure that these applications are considered in the design of future neuromorphic systems.
This report details work that was completed to address the Fiscal Year 2022 Advanced Science and Technology (AS&T) Laboratory Directed Research and Development (LDRD) call for “AI-enhanced Co-Design of Next Generation Microelectronics.” This project required concurrent contributions from the fields of 1) materials science, 2) devices and circuits, 3) physics of computing, and 4) algorithms and system architectures. During this project, we developed AI-enhanced circuit design methods that relied on reinforcement learning and evolutionary algorithms. The AI-enhanced design methods were tested on neuromorphic circuit design problems that have real-world applications related to Sandia’s mission needs. The developed methods enable the design of circuits, including circuits that are built from emerging devices, and they were also extended to enable novel device discovery. We expect that these AI-enhanced design methods will accelerate progress towards developing next-generation, high-performance neuromorphic computing systems.
Using a newly developed coupling of the ElectroMagnetic Plasma In Realistic Environments (EMPIRE) code with the Integrated Tiger Series (ITS) code, radiation environment calculations have been performed. The effort was completed as part of the Saturn Recapitalization (Recap) program that represents activities to upgrade and modernize the Saturn accelerator facility. The radiation environment calculations performed provide baseline results with current or planned hardware in the facility. As facility design changes are proposed and implemented as part of Saturn Recap, calculations of the radiation environment will be performed to understand how the changes impact the output of the Saturn accelerator.
Simulating molecules and atomic systems at quantum accuracy is a grand challenge for science in the 21st century. Quantum-accurate simulations would enable the design of new medicines and the discovery of new materials. The defining problem in this challenge is that quantum calculations on large molecules, like proteins or DNA, are fundamentally impossible with current algorithms. In this work, we explore a range of different methods that aim to make large, quantum-accurate simulations possible. We show that using advanced classical models, we can accurately simulate ion channels, an important biomolecular system. We show how advanced classical models can be implemented in an exascale-ready software package. Lastly, we show how machine learning can learn the laws of quantum mechanics from data and enable quantum electronic structure calculations on thousands of atoms, a feat that is impossible for current algorithms. Altogether, this work shows that combining advances in physics models, computing, and machine learning, we are moving closer to the reality of accurately simulating our molecular world.
This document is intended to be utilized with the Equipment Test Environment being developed to provide a standard process by which the ETE can be validated. The ETE is developed with the intent of establishing cyber intrusion, data collection and through automation provide objective goals that provide repeatability. This testing process is being developed to interface with the Technical Area V physical protection system. The document will overview the testing structure, interfaces, device and network logging and data capture. Additionally, it will cover the testing procedure, criteria and constraints necessary to properly capture data and logs and record them for experimental data capture and analysis.
The composition and phase fraction of the intergranular phase of 94ND10 ceramic is determined and fabricated ex situ. The fraction of each phase is 85.96 vol% Al2O3 bulk phase, 9.46 vol% Mg-rich intergranular phase, 4.36 vol% Ca/Si-rich intergranular phase, and 0.22 vol% voids. The Ca/Si-rich phase consists of 0.628 at% Mg, 12.59 at% Si, 10.24 at% Ca, 17.23 at% Al, and balance O. The Mgrich phase consists of 14.17 at% Mg, 0.066 at% Si, 0.047 at% Ca, 28.69 at% Al, and balance O. XRD of the ex situ intergranular material made by mixed oxides consisting of the above phase and element fractions yielded 92 vol% MgAl2O4 phase and 8 vol% CaAl2Si2O8 phase. The formation of MgAl2O4 phase is consistent with prior XRD of 94ND10, while the CaAl2Si2O8 phase may exist in 94ND10 but at a concentration not readily detected with XRD. The MgAl2O4 and CaAl2Si2O8 phases determined from XRD are expected to have the elemental compositions for the Mg-rich and Ca/Si-rich phases above by cation substitutions (e.g., some Mg substituted for by Ca in the Mg-rich phase) and impurity phases not detectable with XRD.
Time-resolved X-ray thermometry is an enabling technology for measuring temperature and phase change of components. However, current diagnostic methods are limited in their ability due to the invasive nature of probes or the requirement of coatings and optical access to the component. Our proposed developments overcome these challenges by utilizing X-rays to directly measure the objects temperature. Variable-Temperature X-ray Diffraction (VT-XRD) was performed over a wide range of temperatures and diffraction angles and was performed on several materials to analyze the patterns of the bulk materials for sensitivity. "High-speed" VT-XRD was then performed for a single material over a small range of diffraction angles to see how fast the experiments could be performed, whilst still maintaining peaks sufficiently large enough for analysis.
Buczkowski, Nicole E.; Foss, Mikil D.; Parks, Michael L.; Radu, Petronela
The paper presents a collection of results on continuous dependence for solutions to nonlocal problems under perturbations of data and system parameters. The integral operators appearing in the systems capture interactions via heterogeneous kernels that exhibit different types of weak singularities, space dependence, even regions of zero-interaction. The stability results showcase explicit bounds involving the measure of the domain and of the interaction collar size, nonlocal Poincaré constant, and other parameters. In the nonlinear setting, the bounds quantify in different Lp norms the sensitivity of solutions under different nonlinearity profiles. The results are validated by numerical simulations showcasing discontinuous solutions, varying horizons of interactions, and symmetric and heterogeneous kernels.
Multiple physical and chemical forms of a given radionuclide may be released in the event of a nuclear accident. Given that variable forms of an isotope may elicit changes in how that isotope moves through the environment and ultimately impacts human receptors, it is pertinent to understand how nuclear accident consequence models, such as MACCS, account for variable forms. This report documents a review of MACCS modeling capabilities for variability in radionuclide chemical and physical forms. This review centers on the current state-of-practice for dosimetry and deposition modeling of varying radionuclide forms to understand how consistent existing MACCS capabilities are with state of practice. This analysis is also used to inform potential MACCS model upgrades. MACCS conceptual models along with dosimetry and deposition related practices are discussed. Recommendations and suggestions for model improvements are posited.
Tracer gases, whether they are chemical or isotopic in nature, are useful tools in examining the flow and transport of gaseous or volatile species in the underground. One application is using detection of short-lived argon and xenon radionuclides to monitor for underground nuclear explosions. However, even chemically inert species, such as the noble gases, have bene observed to exhibit non-conservative behavior when flowing through porous media containing certain materials, such as zeolites, due to gas adsorption processes. This report details the model developed, implemented, and tested in the open source and massively parallel subsurface flow and transport simulator PFLOTRAN for future use in modeling the transport of adsorbing tracer gases.
The purpose of our report is to discuss the notion of entropy and its relationship with statistics. Our goal is to provide a manner in which you can think about entropy, its central role within information theory and relationship with statistics. We review various relationships between information theory and statistics—nearly all are well-known but unfortunately are often not recognized. Entropy quantities the "average amount of surprise" in a random variable and lies at the heart of information theory, which studies the transmission, processing, extraction, and utilization of information. For us, data is information. What is the distinction between information theory and statistics? Information theorists work with probability distributions. Instead, statisticians work with samples. In so many words, information theory using samples is the practice of statistics.
On the path towards climate-neutral future mobility, the usage of synthetic fuels derived from renewable power sources, so-called e-fuels, will be necessary. Oxygenated e-fuels, which contain oxygen in their chemical structure, not only have the potential to realize a climate-neutral powertrain, but also to burn more cleanly in terms of soot formation. Polyoxymethylene dimethyl ethers (PODE or OMEs) are a frequently discussed representative of such combustibles. However, to operate compression ignition engines with these fuels achieving maximum efficiency and minimum emissions, the physical-chemical behavior of OMEs needs to be understood and quantified. Especially the detailed characterization of physical and chemical properties of the spray is of utmost importance for the optimization of the injection and the mixture formation process. The presented work aimed to develop a comprehensive CFD model to specify the differences between OMEs and dodecane, which served as a reference diesel-like fuel, with regards to spray atomization, mixing and auto-ignition for single- and multi-injection patterns. The simulation results were validated against experimental data from a high-temperature and high-pressure combustion vessel. The sprays’ liquid and vapor phase penetration were measured with Mie-scattering and schlieren-imaging as well as diffuse back illumination and Rayleigh-scattering for both fuels. To characterize the ignition process and the flame propagation, measurements of the OH* chemiluminescence of the flame were carried out. Significant differences in the ignition behavior between OMEs and dodecane could be identified in both experiments and CFD simulations. Liquid penetration as well as flame lift-off length are shown to be consistently longer for OMEs. Zones of high reaction activity differ substantially for the two fuels: Along the spray center axis for OMEs and at the shear boundary layers of fuel and ambient air for dodecane. Additionally, the transient behavior of high temperature reactions for OME is predicted to be much faster.
Imaging using THz waves has been a promising option for penetrative measurements in environments that are opaque to visible wavelengths. However, available THz imaging systems have been limited to relatively low frame rates and cannot be applied to study fast dynamics. This work explores the use of upconversion imaging techniques based on nonlinear optics to enable wavelength-flexible high frame rate THz imaging. UpConversion Imaging (UCI) uses nonlinear conversion techniques to shift the THz wavelengths carrying a target image to shorter visible or near-IR wavelengths that can be detected by available high-speed cameras. This report describes the analysis methodology used to design a prototype high-rate THz UCI system and gives a detailed explanations of the design choices that were made. The design uses a high-rate pulse-burst laser system to pump both THz generation and THz upconversion detection, allowing for scaling to acquisition rates in excess of 10 kHz. The design of the prototype system described in this report has been completed and all necessary materials have been procured. Assembly and characterization testing is on-going at the submission of this report. This report proposes future directions for work on high-rate THz UCI and potential applications of future systems.
Reno, Matthew J.; Blakely, Logan; Trevizan, Rodrigo D.; Pena, Bethany; Lave, Matthew S.; Azzolini, Joseph A.; Yusuf, Jubair Y.; Jones, Christian B.; Furlani Bastos, Alvaro F.; Chalamala, Rohit; Korkali, Mert; Sun, Chih-Che; Donadee, Jonathan; Stewart, Emma M.; Donde, Vaibhav; Peppanen, Jouni; Hernandez, Miguel; Deboever, Jeremiah; Rocha, Celso; Rylander, Matthew; Siratarnsophon, Piyapath; Grijalva, Santiago; Talkington, Samuel; Mason, Karl; Vejdan, Sadegh; Khan, Ahmad U.; Mbeleg, Jordan S.; Ashok, Kavya; Divan, Deepak; Li, Feng; Therrien, Francis; Jacques, Patrick; Rao, Vittal; Francis, Cody; Zaragoza, Nicholas; Nordy, David; Glass, Jim; Holman, Derek; Mannon, Tim; Pinney, David
This report summarizes the work performed under a project funded by U.S. DOE Solar Energy Technologies Office (SETO), including some updates from the previous report SAND2022-0215, to use grid edge measurements to calibrate distribution system models for improved planning and grid integration of solar PV. Several physics-based data-driven algorithms are developed to identify inaccuracies in models and to bring increased visibility into distribution system planning. This includes phase identification, secondary system topology and parameter estimation, meter-to-transformer pairing, medium-voltage reconfiguration detection, determination of regulator and capacitor settings, PV system detection, PV parameter and setting estimation, PV dynamic models, and improved load modeling. Each of the algorithms is tested using simulation data and demonstrated on real feeders with our utility partners. The final algorithms demonstrate the potential for future planning and operations of the electric power grid to be more automated and data-driven, with more granularity, higher accuracy, and more comprehensive visibility into the system.
This report summarizes the needs, challenges, and opportunities associated with carbon-free energy and energy storage for manufacturing and industrial decarbonization. Energy needs and challenges for different manufacturing and industrial sectors (e.g., cement/steel production, chemicals, materials synthesis) are identified. Key issues for industry include the need for large, continuous on-site capacity (tens to hundreds of megawatts), compatibility with existing infrastructure, cost, and safety. Energy storage technologies that can potentially address these needs, which include electrochemical, thermal, and chemical energy storage, are presented along with key challenges, gaps, and integration issues. Analysis tools to value energy storage technologies in the context of manufacturing and industrial decarbonizations are also presented. Material is drawn from the Energy Storage for Manufacturing and Industrial Decarbonization (Energy StorM) Workshop, held February 8 - 9, 2022. The objective was to identify research opportunities and needs for the U.S. Department of Energy as part of its Energy Storage Grand Challenge program.
This project created and demonstrated a framework for the efficient and accurate prediction of complex systems with only a limited amount of highly trusted data. These next generation computational multi-fidelity tools fuse multiple information sources of varying cost and accuracy to reduce the computational and experimental resources needed for designing and assessing complex multi-physics/scale/component systems. These tools have already been used to substantially improve the computational efficiency of simulation aided modeling activities from assessing thermal battery performance to predicting material deformation. This report summarizes the work carried out during a two year LDRD project. Specifically we present our technical accomplishments; project outputs such as publications, presentations and professional leadership activities; and the project’s legacy.
This document contains the final report for the midyear LDRD titled "Extension of Interferometric Synthetic Aperture Radar to Multiple Phase-Centers." This report presents an overview of several methods for approaching the two-target in layover problem that exists in interferometric synthetic aperture radar systems. Simulation results for one of the methods are presented. In addition, a new direct approach is introduced.
Widespread integration of social media into daily life has fundamentally changed the way society communicates, and, as a result, how individuals develop attitudes, personal philosophies, and worldviews. The excess spread of disinformation and misinformation due to this increased connectedness and streamlined communication has been extensively studied, simulated, and modeled. Less studied is the interaction of many pieces of misinformation, and the resulting formation of attitudes. We develop a framework for the simulation of attitude formation based on exposure to multiple cognitions. We allow a set of cognitions with some implicit relational topology to spread on a social network, which is defined with separate layers to specify online and offline relationships. An individual’s opinion on each cognition is determined by a process inspired by the Ising model for ferromagnetism. We conduct experimentation using this framework to test the effect of topology, connectedness, and social media adoption on the ultimate prevalence of and exposure to certain attitudes.
The goal of this work was to pioneer a novel, low-overhead protocol for simultaneously assaying cell-surface markers and intracellular gene expression in a single mammalian cell. The purpose of developing such a method is to be able to understand the mechanisms by which pathogens engage with individual mammalian cells, depending on their cell surface proteins, and how both host and pathogen gene expression changes are reflective of these mechanisms. The knowledge gained from such analyses of single cells will ultimately lead to more robust pathogen detection and countermeasures. Our method was aimed at streamlining both the upstream cell sample preparation using microfluidic methods, as well as the actual library making protocol. Specifically, we wanted to implement a random hexamer-based reverse transcription of all RNA within a single cell (as opposed to oligo dT-based which would only capture polyadenylated transcripts), and then use a CRISPR-based method called scDash to deplete ribosomal DNAs (since ribosomal RNAs make up the majority of the RNA in a mammalian cell). After significant troubleshooting, we demonstrate that we are able to prepare cDNA from RNA using the random hexamer primer, and perform the rDNA depletion. We also show that we can visualize individually stained cells, setting up the pipeline for connecting surface markers to RNA-sequencing profiles. Finally, we test a number of devices for various parts of the pipeline, including bead generation, optical barcoding and cell dispensing, and demonstrate that while some of these have potential, more work is needed to optimize this part of the pipeline.
Representation of soil organic carbon (SOC) dynamics in Earth system models (ESMs) is a key source of uncertainty in predicting carbon climate feedbacks. The magnitude of this uncertainty can be reduced by accurate representation of environmental controllers of SOC stocks in ESMs. In this study, we used data of environmental factors, field SOC observations, ESM projections and machine learning approaches to identify dominant environmental controllers of SOC stocks and derive functional relationships between environmental factors and SOC stocks. Our derived functional relationships predicted SOC stocks with similar accuracy as the machine learning approach. We used the derived relationships to benchmark the coupled model intercomparison project phase six ESM representation of SOC stocks. We found divergent environmental control representation in ESMs in comparison to field observations. Representation of SOC in ESMs can be improved by including additional environmental factors and representing their functional relationships with SOC consistent with observations.
This project explores the idea of performing kinetic numerical simulations in the Z inner magnetically insulated transmission line (inner MITL) by reduced physics models such as a guiding center drift kinetic approximation for particles and electrostatic and magnetostatic approximation for the fields. The basic problem explored herein is the generation, formation, and evolution of vortices by electron space charge limited (SCL) emission. The results indicate that for relevant to Z values of peak current and pulse length, these approximations are excellent, while also providing tens to hundreds of times reduction in the computational load. The benefits could be enormous: Implementation of these reduced physics models in present particle-in-cell (PIC) codes could enable them to be routinely used for experimental design while still capturing essential non-thermal (kinetic) physics.
The National Solar Thermal Test Facility (NSTTF) at Sandia National Laboratories New Mexico (SNL/NM) developed this Life Cycle Management Plan (LCMP) to document its process for executing, monitoring, controlling and closing-out Phase 3 of the Gen 3 Particle Pilot Plant (G3P3). This plan serves as a resource for stakeholders who wish to be knowledgeable of project objectives and how they will be accomplished.
Sandia National Laboratories (SNL) is designing and developing an Artificial Intelligence (AI)-enabled smart digital assistant (SDA), Inspecta (International Nuclear Safeguards Personal Examination and Containment Tracking Assistant). The goal is to provide inspectors an in-field digital assistant that can perform tasks identified as tedious, challenging, or prone to human error. During 2021, we defined the requirements for Inspecta based on reviews of International Atomic Energy Agency (IAEA) publications and interviews with former IAEA inspectors. We then mapped the requirements to current commercial or open-source technical capabilities to provide a development path for an initial Inspecta prototype while highlighting potential research and development tasks. We selected a highimpact inspection task that could be performed by an early Inspecta prototype and are developing the initial architecture, including hardware platform. This paper describes the methodology for selecting an initial task scenario, the first set of Inspecta skills needed to assist with that task scenario and finally the design and development of Inspecta’s architecture and platform.
Exterior solar glaze was added to a 3 foot x 3 foot x 3 foot aluminum solar collector that had six triangular dimpled fins for enhanced heat transfer. The interior vertical wall on the south side was also dimpled. The solar glaze was added to compare its solar collection performance with unglazed solar collector experiments conducted at Sandia in 2021. The east, west, front, and top sides of the solar collector were encased with solar glaze glass. Because the solar incident heat on the north and bottom sides was minimal, they were insulated to retain the heat that was collected by the other four sides. The advantages of the solar glaze include the entrapment of more solar heat, as well as insulation from the wind. The disadvantages are that it increases the cost of the solar collector and has fragile structural properties when compared to the aluminum walls. Nevertheless, prior to conducting experiments with the glazed solar collector, it was not clear if the benefits outweighed the disadvantages. These issues are addressed herein, with the conclusion that the additional amount of heat collected by the glaze justifies the additional cost. The solar collector glaze design, experimental data, and costs and benefits are documented in this report.
Quantifying the sensitivity - how a quantity of interest (QoI) varies with respect to a parameter – and response – the representation of a QoI as a function of a parameter - of a computer model of a parametric dynamical system is an important and challenging problem. Traditional methods fail in this context since sensitive dependence on initial conditions implies that the sensitivity and response of a QoI may be ill-conditioned or not well-defined. If a chaotic model has an ergodic attractor, then ergodic averages of QoIs are well-defined quantities and their sensitivity can be used to characterize model sensitivity. The response theorem gives sufficient conditions such that the local forward sensitivity – the derivative with respect to a given parameter - of an ergodic average of a QoI is well-defined. We describe a method based on ergodic and response theory for computing the sensitivity and response of a given QoI with respect to a given parameter in a chaotic model with an ergodic and hyperbolic attractor. This method does not require computation of ensembles of the model with perturbed parameter values. The method is demonstrated and some of the computations are validated on the Lorenz 63 and Lorenz 96 models.
The focus of this project is to accelerate and transform the workflow of multiscale materials modeling by developing an integrated toolchain seamlessly combining DFT, SNAP, LAMMPS, (shown in Figure 1-1) and a machine-learning (ML) model that will more efficiently extract information from a smaller set of first-principles calculations. Our ML model enables us to accelerate first-principles data generation by interpolating existing high fidelity data, and extend the simulation scale by extrapolating high fidelity data (102 atoms) to the mesoscale (104 atoms). It encodes the underlying physics of atomic interactions on the microscopic scale by adapting a variety of ML techniques such as deep neural networks (DNNs), and graph neural networks (GNNs). We developed a new surrogate model for density functional theory using deep neural networks. The developed ML surrogate is demonstrated in a workflow to generate accurate band energies, total energies, and density of the 298K and 933K Aluminum systems. Furthermore, the models can be used to predict the quantities of interest for systems with more number of atoms than the training data set. We have demonstrated that the ML model can be used to compute the quantities of interest for systems with 100,000 Al atoms. When compared with 2000 Al system the new surrogate model is as accurate as DFT, but three orders of magnitude faster. We also explored optimal experimental design techniques to choose the training data and novel Graph Neural Networks to train on smaller data sets. These are promising methods that need to be explored in the future.