SEARCH · Engineering Papers
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.
Deep ReLU networks have exponential expressive potential but behave identically to their shallow counterparts under random initialization. This paper introduces a novel training paradigm that constrains weights into mathematically optimized patterns. This enables the resulting network to make exponential use of its depth.
Bayesian Optimization (BO) is a powerful tool for optimizing complex non-linear systems. However, its performance degrades in high-dimensional problems with tightly coupled parameters and highly asymmetric objective landscapes, where rewards are sparse. In such needle-in-a-haystack scenarios, even advanced methods like trust-region BO (TurBO) often lead to unsatisfactory results. We propose a domain knowledge guided Bayesian Optimization approach, which leverages physical insight to fundamentally simplify the search problem by transforming coordinates to decouple input features and align the active subspaces with the primary search axes. We demonstrate this approach's efficacy on a challenging 12-dimensional, 6-crystal Split-and-Delay optical system, where conventional approaches, including standard BO, TuRBO and multi-objective BO, consistently led to unsatisfactory results. When combined with an reverse annealing exploration strategy, this approach reliably converges to the global optimum. The coordinate transformation itself is the key to this success, significantly accelerating the search by aligning input co-ordinate axes with the problem's active subspaces. As increasingly complex scientific instruments, from large telescopes to new spectrometers at X-ray Free Electron Lasers are deployed, the demand for robust high-dimensional optimization grows. Our results demonstrate a generalizable paradigm: leveraging physical insight to transform high-dimensional, coupled optimization problems into simpler representations can enable rapid and robust automated tuning for consistent high performance while still retaining current optimization algorithms.
The driving goal is to investigate the feasibility of a message-based data transport for SPDK. All of SPDK’s existing data transport protocols are “connection-based”, requiring a target to hold state for each connected client / initiator. At extreme scale, juggling these connections from a large number of initiators creates a performance bottleneck. Under a message-based paradigm, we can bypass this behavior, as all initiators share a global communication channel, held open for the entire run. In theory, this approach should minimize connection management overhead at extreme scale.
We present an ℎ𝑟-adaptivity framework for morphing a given mesh to fit a target surface prescribed as the zero isocontour of a discrete function. In this framework, high-order meshing is posed as a variational minimization problem that depends on the mesh quality prescribed via the target matrix optimization paradigm (TMOP) and position of a subset of mesh nodes with respect to the target surface. The proposed formulation ensures that the variational problem is converged and mesh quality degradation near the surface is limited, even when the mesh topology is incompatible with the target surface. Additionally, a mesh subset-based approach and ℎ-refinement is introduced to efficiently increase fitting accuracy while reducing the computational cost of the mesh morphing problem. The ℎ𝑟-adaptivity technique extends to different element types in two- and three-dimensions, and can be used in existing finite element and spectral element frameworks to obtain high-order body-fitted meshes. Various numerical experiments demonstrate the robustness and accuracy of the fitting approach for problems of practical interest such as Lagrangian hydrodynamics and topology optimization.
In engineering examples, one often encounters the need to sample from unnormalized distributions with complex shapes that may also be implicitly defined through a physical or numerical simulation model, making it computationally expensive to evaluate the associated density function. For such cases, MCMC has proven to be an invaluable tool. Random-walk Metropolis Methods (also known as Metropolis-Hastings (MH)), in particular, are highly popular for their simplicity, flexibility, and ease of implementation. However, most MH algorithms suffer from significant limitations when attempting to sample from distributions with multiple modes (particularly disconnected ones). Here, in this paper, we present Intrepid MCMC - a novel MH scheme that utilizes a simple coordinate transformation to significantly improve the mode-finding ability and convergence rate to the target distribution of random-walk Markov chains while retaining most of the simplicity of the vanilla MH paradigm. Through multiple examples, we showcase the improvement in the performance of Intrepid MCMC over vanilla MH for a wide variety of target distribution shapes. We also provide an analysis of the mixing behavior of the Intrepid Markov chain, as well as the efficiency of our algorithm for increasing dimensions. A thorough discussion is presented on the practical implementation of the Intrepid MCMC algorithm. Finally, its utility is highlighted through a Bayesian parameter inference problem for a two-degree-of-freedom oscillator under free vibration.
Distributed Resources for the Earth System Grid Federation (ESGF) Advanced Management (DREAM) is a proposed system that will enable data from an infinite number of diverse sources to be organized and accessed from anywhere using any handheld or other computer device. The approach offers a powerful roadmap for the creation and integration of a unified knowledge base of an entire ecosystem, including its many geophysical, geographical, social, political, agricultural, energy, transportation, and cyber aspects. The resulting aggregation of data has the potential to generate an informational universe of unprecedented size that has never before been possible due to the prohibitive costs, managerial complexity, and technical barriers associated with ever-changing exponential-growth data flows. We envision that DREAM will accelerate discovery by enabling climate researchers, among other types of researchers, to manage, analyze, and visualize data from earth-scale measurements and simulations. DREAM’s success will be built on proven components that leverage existing services and resources. A key building block for DREAM will be the ESGF, chaired by Dean N. Williams. Expanding on the existing ESGF, the project will ensure that the access, storage, movement, and analysis of the large quantities of data that are processed and produced by diverse science projects can be dynamically distributed with proper resource management. Much of the Office of Science data is currently generated by multiple stand-alone facilities. DREAM can collect data accumulated from these facilities and incorporate it into a fully integrated network accessible from anywhere in the world. The result is a completely new paradigm shift for data management, analysis, and visualization enabling researchers to: Manage their calculations, data, tools, and research results; Ensure that all data are sharable, reproducible and (re)usable—accompanied by appropriate metadata describing its provenance, syntax, and semantics at creation; Advance application performance by selectively adapting APIs and services in response to scientific requirements and architectural complexities; and Provide scalable interactive resource management—navigate data and metadata at multiple levels, provide architecture-aware data integration, analysis and visualization tools. We will engage closely with DOE, NASA, and NOAA science groups working at the leading edge of computing. These engagements—in domains such as biology, climate, and hydrology—will allow us to advance disciplinary science goals and inform our development of technologies that can accelerate discovery across DOE more broadly. We will advertise and promote our technologies via dedicated workshops, tutorials, and sessions at conferences, stand-alone events with broad inter-disciplinary invitation, and engagements with leadership facilities.
Here, it is argued that science of discontinua represents a major paradigm shift in how scientific research is done. First, a historical perspective is given, followed by key aspects of discontinuum simulations and its potential in aiding research and development. This is followed by a philosophical perspective on equivalence of matter and information as inspired by Aristotle’s quest for souls in living things such as cells. Finally, the universality of the discontinuum approach is explored with its consequences for computing and artificial intelligence.
Deep learning has emerged as a promising paradigm to give access to highly accurate predictions of molecular and material properties. A common short-coming shared by current approaches, however, is that neural networks only give point estimates of their predictions and do not come with predictive uncertainties associated with these estimates. Existing uncertainty quantification efforts have primarily leveraged the standard deviation of predictions across an ensemble of independently trained neural networks. This incurs a large computational overhead in both training and prediction, resulting in order-of-magnitude more expensive predictions. Here, we propose a method to estimate the predictive uncertainty based on a single neural network without the need for an ensemble. This allows us to obtain uncertainty estimates with virtually no additional computational overhead over standard training and inference. We demonstrate that the quality of the uncertainty estimates matches those obtained from deep ensembles. We further examine the uncertainty estimates of our methods and deep ensembles across the configuration space of our test system and compare the uncertainties to the potential energy surface. Finally, we study the efficacy of the method in an active learning setting and find the results to match an ensemble-based strategy at order-of-magnitude reduced computational cost.
Significance The understanding of protein function increases with new experimental and evolutionary datasets. A major challenge is to apply machine learning to these datasets to capture essential features of protein function. Here, we analyze the experimentally determined repression function for tens of thousands of mutants of the LacI protein. This study provides a continuous, noncategorical repression value across a majority of all single mutations and for thousands of higher-order mutations. To develop a top-performing model for the prediction of repression by LacI, we compare several leading variant effect prediction algorithms. A deep representation learning paradigm, first trained across millions of proteins from all known protein families and then fine-tuned using LacI experimental data, offers the highest predictive performance of repression function.
Bayesian inference is a powerful paradigm for quantum state tomography, treating uncertainty in meaningful and informative ways. Yet the numerical challenges associated with sampling from complex probability distributions hampers Bayesian tomography in practical settings. In this article, we introduce an improved, self-contained approach for Bayesian quantum state estimation. Leveraging advances in machine learning and statistics, our formulation relies on highly efficient preconditioned Crank–Nicolson sampling and a pseudo-likelihood. We theoretically analyze the computational cost, and provide explicit examples of inference for both actual and simulated datasets, illustrating improved performance with respect to existing approaches.
A scalable density functional electronic code with Gaussian basis set, called UTEP-NRLMOL, is developed to perform simulations of molecular systems in the presence of the environment with particular attention to the memory requirements. In the electronic structure calculations, the memory and computation time are proportional to the number of atoms. Memory requirements for density functional calculations scale as N*N, where N is the number of atoms. While the recent advances in HPC offer platforms with large numbers of cores, the limited amount of memory available on a given node and poor scalability of the electronic structure codes hinder their efficient usage of these platforms. We have introduced new scaling and parallelization paradigms using MPI-3 shared-memory functionality combined with usage of sparse algebra and storage of matrices in sparse format. This extends the range of applicability of the UTEP-NRLMOL code to large systems over 10,000 atoms, or using up to 67,000 basis functions, and making use of HPC architectures using over 6,000 processors utilizing all available cores. We have also interfaced code with effective fragment potential and polarizable continuum model libraries. The code was used in simulations of several applications which are published in reputed scientific journals.
A stabilized continuous Galerkin (CG) method for magnetohydrodynamics (MHD) is presented herein. Ideal, compressible inviscid MHD equations are discretized in space on unstructured meshes using piecewise linear or bilinear finite element bases to get a semi-discrete scheme. Stabilization is then introduced to the semi-discrete method in a strategy that follows the algebraic flux correction paradigm. This involves adding some artificial diffusion to the high order, semi-discrete method and mass lumping in the time derivative term. The result is a low order method that provides local extremum diminishing properties for hyperbolic systems. The difference between the low order method and the high order method is scaled element-wise using a limiter and added to the low order scheme. The limiter is solution dependent and computed via an iterative linearity preserving nodal variation limiting strategy. The stabilization also involves an optional consistent background high order dissipation that reduces phase errors. The resulting stabilized scheme is a semi-discrete method that can be applied to inviscid shock MHD problems and may be even extended to resistive and viscous MHD problems. To satisfy the divergence free constraint of the MHD equations, we add parabolic divergence cleaning to the system. Various time integration methods can be used to discretize the scheme in time. We demonstrate the robustness of the scheme by solving several shock MHD problems.
Membership inference attacks (MIAs), which enable adversaries to determine whether specific data points were part of a model's training dataset, have emerged as an important framework to understand, assess, and quantify the potential information leakage associated with machine learning systems. Designing effective MIAs is a challenging task that usually requires extensive manual exploration of model behaviors to identify potential vulnerabilities. In this paper, we introduce AutoMIA -- a novel framework that leverages large language model (LLM) agents to automate the design and implementation of new MIA signal computations. By utilizing LLM agents, we can systematically explore a vast space of potential attack strategies, enabling the discovery of novel strategies. Our experiments demonstrate AutoMIA can successfully discover new MIAs that are specifically tailored to user-configured target model and dataset, resulting in improvements of up to 0.18 in absolute AUC over existing MIAs. This work provides the first demonstration that LLM agents can serve as an effective and scalable paradigm for designing and implementing MIAs with SOTA performance, opening up new avenues for future exploration.
The growing adoption of distributed energy resources (DERs) like battery energy storage systems and roof top solar/PV and the rapid penetration of electric vehicles (EVs), the electric grid is undergoing a major transformation with elevated stress on legacy grid assets. Despite a lot of expenditure to address these challenges, both in dollars and manpower, utilities have not been able to receive the value that was promised. The gains have been most visible at the transmission and substation level, especially where the main objective was improving operational and economic efficiency for the utility. Improving visibility and control at a few select points enhances the existing and established paradigm of centralized command and control. With changing load patterns, load types and the overall transition to an “active grid”, the centralized control and coordination paradigm gets challenged. To address the challenges, a new architecture and mechanism is needed, one that supports decentralized control and decision making, extracting value streams at the grid edge, particularly as the changes are fueled by transitions occurring in the distribution system. To address this, a communications and data processing platform, “GAMMA” was developed and demonstrated through the project. At the heart of the platform, are distributed, intelligent edge nodes with sensing and compute capabilities, that can record and analyze information locally. They are embedded in sensors and actuators specific to different distribution system applications. Phase 1 of the project focused on developing novel sensor technology that can be used for monitoring utility pole top distribution transformers. The sensors were designed with the objective of being low-cost, communicating with the GAMMA cloud using novel “delay-tolerant” networking using Bluetooth and a secure mobile application. They were non-intrusive in nature so that they can be installed quickly in the field, resulting in overall low cost of deployment and operations. Following the successful completion of Phase 1, the team manufactured 100 units for a field demonstration in Phase 2. The field demonstration was carried out on two real feeder systems with the local utility partner. In total, 100 sensors were installed and operated over a period of 6 months in the state of Georgia. The platform is operational end to end, with the cloud infrastructure deployed on a distributed, serverless environment that can serve multiple data streams, an analytics engine and a portal to securely view the data from multiple assets. The data collected through the GAMMA Mobile Phone app showcased the viability of the novel delay tolerant networking architecture, and the data processing algorithms developed through the course of the project, were successful in extracting important information about the overall network, improving the utility’s visibility and situational awareness in the distribution feeder.
Active- and transfer-learning are applied to microscale dynamics of polymer flows for the multiscale discovery of effective constitutive approximations required in viscoelastic flow simulation. The result is macroscopic rheology directly connected to a microstructural model. Micro and macroscale simulations are adaptively coupled by means of Gaussian process regression (GPR) to run the expensive microscale computations only as necessary. This multiscale method is demonstrated with flows of a polymer solution as a model system. At the microscale level dissipative particle dynamics (DPD) is employed to model the fluid as a suspension of bead-spring micro-structures subjected to steady shear flow. The results yield the non-Newtonian viscosity and the first normal stress difference at strain rates as training data used in a GPR model. DPD parameters are calibrated with respect to experimental data for a real polymer solution. Compliance with these data requires adjustment of the DPD model's cutoff radius, which then becomes a function of the second invariant of the strain rate tensor. The FENE-P model is chosen for the macroscale description using the spectral element method (SEM) to simulate channel flow and flow past a circular cylinder. The DPD results at the lowest possible shear strain rate yield an estimate of the zero-shear rate viscosity, which allows the initiation of the macroscale flow by SEM as a Newtonian fluid. The resulting strain-rate field is surveyed to determine additional shear strain rate sampling points for the DPD system. This new information allows an initial fitting of parameters of the constitutive equation followed by new SEM simulations at the macroscale. Additionally, guided by active-learning GPR to select new sampling points, this process continues until convergence is achieved. The effectiveness of this new simulation paradigm for viscoelastic flows is tested with different macroscale operating conditions. The effective closure learned in the channel simulation is then transferred directly to the flow past a circular cylinder at low Reynolds number, where the results show that only two additional DPD simulations are required to achieve a satisfactory constitutive model. With an increase of the Reynolds number, the active-learning scheme automatically detects the inaccuracy of the learned constitutive model, and initiates additional DPD simulations for the extra data needed to once again close the microscale-macroscale coupled system. This new paradigm of active- and transfer-learning for multiscale modeling is readily applicable to other microscale-macroscale coupled simulations of complex fluids and other materials. Furthermore, the coupling between microscale and macroscale solvers can be seamlessly implemented with our open source multiscale universal interface (MUI) library.
In this paper, we describe the development of a reactive force field for C/O systems under extreme temperatures and pressures, based on the many-body Chebyshev Interaction Model for Efficient Simulation (ChIMES). The resulting model, which targets carbon condensation under thermodynamic conditions of 6500 K and 2.5 g cm –3 , affords a balance between model accuracy, complexity, and training set generation expense. We show that the model recovers much of the accuracy of density functional theory for the prediction of structure, dynamics, and chemistry when applied to dissociative condensed phase systems at 1:1 and 1:2 C:O ratios, as well as molten carbon. Our C/O modeling approach exhibits a 104 increase in efficiency for the same system size (i.e., 128 atoms) and a linear system size scalability over standard quantum molecular dynamics methods, allowing the simulation of significantly larger systems than previously possible. We find that the model captures the condensed-phase reaction-coupled formation of carbon clusters implied by recent experiments, and that this process is susceptible to strong finite size effects. Overall, we find the present ChIMES model to be well suited for studying chemical processes and cluster formation at pressures and temperatures typical of shock waves. We expect that the present C/O modeling paradigm can serve as a template for the development of a broader high pressure–high temperature force-field for condensed phase chemistry in organic materials.
Geological carbon storage (GCS) projects could involve a diverse array of stakeholders or players from public, private, and regulatory sectors, each with different objectives and responsibilities. Given the complexity, scale, and long-term nature of GCS operations, determining whether individual stakeholders can independently optimize their interests — or whether collaborative coalition agreements are needed — remains a central question for effective GCS project planning and management. To access large, high-quality storage resources, future GCS deployment may increasingly occur in geologically connected sites, where shared geological features such as pressure space and reservoir pore capacity can lead to competitive behavior among stakeholders. In this work, we propose a paradigm based on Markov games to quantitatively investigate how different coalition structures affect the goals of stakeholders. We frame this multi-stakeholder multi-site problem as a multi-agent reinforcement learning problem with safety constraints. Our approach enables agents to learn optimal strategies while complying with safety regulations. We present an example where multiple operators are injecting CO 2 into their respective project areas in a geologically connected basin. To address the high computational cost of repeated simulations of high fidelity models, a previously developed surrogate model based on the Embed-to-Control (E2C) framework is employed. Our results demonstrate the effectiveness of the proposed framework in addressing optimal management of CO 2 storage when multiple stakeholders with different objectives and goals are involved.
X-ray scattering experiments using Free Electron Lasers (XFELs) are a powerful tool to determine the molecular structure and function of unknown samples (such as COVID-19 viral proteins). XFEL experiments are a challenge to computing in two ways: i) due to the high cost of running XFELs, a fast turnaround time from data acquisition to data analysis is essential to make informed decisions on experimental protocols; ii) data collection rates are growing exponentially, requiring new scalable algorithms. Here we report our experiences analyzing data from two experiments at the Linac Coherent Light Source (LCLS) during September 2020. Raw data were analyzed on NERSC's Cori XC40 system, using the Superfacility paradigm: our workflow automatically moves raw data between LCLS and NERSC, where it is analyzed using the software package CCTBX. We achieved real time data analysis with a turnaround time from data acquisition to full molecular reconstruction in as little as 10 min -- sufficient time for the experiment's operators to make informed decisions. By hosting the data analysis on Cori, and by automating LCLS-NERSC interoperability, we achieved a data analysis rate which matches the data acquisition rate. Furthermore, completing data analysis with 10 mins is a first for XFEL experiments and an important milestone if we are to keep up with data collection trends.