SEARCH · Engineering Papers
Results for “linear embedding”
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.
Linear-Response Time-Dependent Embedded Mean-Field Theory
We present a time-dependent (TD) linear-response description of excited electronic states within the framework of embedded mean-field theory (EMFT). TD-EMFT allows for subsystems to be described at different mean-field levels of theory, enabling straightforward treatment of excited states and transition properties. We provide benchmark demonstrations of TD-EMFT for both local and nonlocal excitations in organic molecules, as well as applications to chlorophyll a, solvatochromic shifts of a dye in solution, and sulfur K-edge X-ray absorption spectroscopy (XAS). It is found that mixed-basis implementations of TD-EMFT lead to substantial errors in terms of transition properties; however, as previously found for ground-state EMFT, these errors are largely eliminated with the use of Fock-matrix corrections. These results indicate that TD-EMFT is a promising method for the efficient, multilevel description of excited-state electronic structure and dynamics in complex systems.
Mixed-Integer Linear Programming Formulation with Embedded Machine Learning Surrogates for the Design of Chemical Process Families
In previous work, we introduced process family design. The main idea is to design a platform of common elements, and, allowing us to capture additional cost savings, simultaneously design a family of processes, and reducing both engineering and deployment timelines. We formulate this as an optimization problem, specifically a nonlinear generalized disjunctive program (GDP). We have proposed two approaches for reformulating and solving this problem: one based on full-discretization of the design space and one that uses Machine Learning (ML) surrogates to replace the nonlinear process models. Using ML surrogates to predict required system costs and performance indicators allows us to reformulate the nonlinearities in the GDP generate an efficient MILP formulation. In this work, we apply the ML surrogate approach to two case studies. One case study involves designing a family of carbon capture systems to cover a set of different flue gas flow rates and inlet CO 2 concentrations, where we consider the absorber and stripper as common unit module types. The second case study focuses on a water-desalination process, where we design a family of these processes for a variety of salt concentrations and flow rates. In both of these case studies, we demonstrate a scalable optimization approach that enables the design of multiple processes simultaneously, reducing the time-to-market and overall costs by maximizing the cost savings due to both economies of scale and economies of numbers.
An efficient MILP framework for integrating nonlinear process dynamics and control in optimal production scheduling calculations
The emphasis currently placed on enterprise-wide decision making and optimization has led to an increased need for methods of integrating nonlinear process dynamics and control information in scheduling calculations. The inevitable high dimensionality and nonlinearity of first-principles dynamic process models makes incorporating them in scheduling calculations challenging. In this work, we describe a general framework for deriving data-driven surrogate models of the closed-loop process dynamics. Focusing on Hammerstein–Wiener and finite step response (FSR) model forms, we show that these models can be (exactly) linearized and embedded in production scheduling calculations. The resulting scheduling problems are mixed-integer linear programs with a special structure, which we exploit in a novel and efficient solution strategy. A polymerization reactor case study is utilized to demonstrate the merits of this method. Finally, our framework compares favorably to existing approaches that embed dynamics in scheduling calculations, showing considerable reductions in computational effort.
Network Traffic Monitoring Using Poisson Dynamic Linear Models
In this article, we discuss an approach for network forensics using a class of nonstationary Poisson processes with embedded dynamic linear models. As a modeling strategy, the Poisson DLM (PoDLM) provides a very flexible framework for specifying structured effects that may influence the evolution of the underlying Poisson rate parameter, including diurnal and weekly usage patterns. We develop a novel particle learning algorithm for online smoothing and prediction for the PoDLM, and demonstrate the suitability of the approach to real-time deployment settings via a new application to computer network traffic monitoring.
Grand Unification of Quantum Algorithms
Quantum algorithms offer significant speed-ups over their classical counterparts for a variety of problems. The strongest arguments for this advantage are borne by algorithms for quantum search, quantum phase estimation, and Hamiltonian simulation, which appear as subroutines for large families of composite quantum algorithms. A number of these quantum algorithms have recently been tied together by a novel technique known as the quantum singular value transformation (QSVT), which enables one to perform a polynomial transformation of the singular values of a linear operator embedded in a unitary matrix. In the seminal GSLW’19 paper on the QSVT [Gilyén et al., ACM STOC 2019], many algorithms are encompassed, including amplitude amplification, methods for the quantum linear systems problem, and quantum simulation. Here, we provide a pedagogical tutorial through these developments, first illustrating how quantum signal processing may be generalized to the quantum eigenvalue transform, from which the QSVT naturally emerges. Paralleling GSLW’19, we then employ the QSVT to construct intuitive quantum algorithms for search, phase estimation, and Hamiltonian simulation, and also showcase algorithms for the eigenvalue threshold problem and matrix inversion. This overview illustrates how the QSVT is a single framework comprising the three major quantum algorithms, suggesting a grand unification of quantum algorithms.
Characterization of linear plasma synthetic jet actuators in an initially quiescent medium
The plasma synthetic jet actuator (PSJA) is a geometrical variant of the aerodynamic plasma actuator that can be used to produce zero-mass flux jets similar to those created by mechanical devices. This jet can be either three-dimensional using annular electrode arrays (annular PSJA) or nearly two dimensional using two rectangular-strip exposed electrodes and one embedded electrode (linear PSJA). Unsteady pulsing of the PSJA at time scales decoupled to the ac input frequency results in a flow field dominated by counter-rotating vortical structures similar to conventional synthetic jets, and the peak velocity and momentum of the jet is found to be affected by a combination of the pulsing frequency and input power. This paper investigates the fluid dynamic characteristics of linear plasma synthetic jet actuators in an initially quiescent medium. Two-dimensional particle image velocimetry measurements on the actuator are used to validate a previously developed numerical model wherein the plasma behavior is introduced into the Navier-Stokes equations as an electrohydrodynamic force term calculated from Maxwell's equations and solved for the fluid momentum. The numerical model was implemented in an incompressible, unstructured grid code. The results of the simulations are observed to reproduce some aspects of the qualitative and quantitative experimental behavior of the jet for steady and pulsed modes of actuator operation. The self-similarity behavior of plasma synthetic jets are examined and compared to mechanically driven continuous and synthetic jets.
FILAMENTARY CONDENSATIONS IN A YOUNG CLUSTER
New models are presented for star-forming condensations in clusters. In each model, the condensation mass increases linearly with radius on small scales, and more rapidly on large scales, as in 'thermal-nonthermal' models. Spherical condensations with this structure form protostars which match the initial mass function if their infall is subject to equally likely stopping. However, such spherical models do not match the filamentary nature of cluster gas, and they are too extended to form protostars having high mass and short spacing. Two hybrid models are presented, which are spherical on small scales and filamentary on large scales. In and around clusters, cores embedded in linear filaments match the elongation of cluster gas, and the central concentration of low-mass stars. In cluster centers, condensations require a low volume-filling factor to produce massive stars with short spacing. These may have stellate shape, where cores are nodes of filamentary networks, as seen in some simulations of colliding flows and collapsing turbulent clumps. A dense configuration of such stellate condensations may be indistinguishable from a clump forming multiple protostars via filamentary flow paths.
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.
Multidimensional, autoresonant three-wave interactions
The theory of autoresonant three-wave interactions is generalized to more than one space and/or time variation of the background medium. In the most general case, the three waves propagate in a four-dimensional (4D) slowly space-time varying background, with an embedded 3D linear resonance hypersurface, where the linear frequency and wave-vector matching conditions of the three waves are satisfied exactly. The autoresonance in the system is the result of weak nonlinear frequency shifts and nonuniformity in the problem and is manifested by satisfaction of the nonlinear resonance conditions in an extended region of space-time adjacent to the resonance surface despite the variation of the background. The threshold condition for autoresonance is found and further discussed in application to stimulated Raman scattering in a 1D, time-dependent plasma case. Asymptotic description of the autoresonant waves far away from the resonance surface is obtained. The theory is illustrated and tested in 2D numerical simulatio0008.
Sequential Learning of Active Subspaces
This study shows that in recent years, active subspace methods (ASMs) have become a popular means of performing subspace sensitivity analysis on black-box functions. Naively applied, however, ASMs require gradient evaluations of the target function. In the event of noisy, expensive, or stochastic simulators, evaluating gradients via finite differencing may be infeasible. In such cases, often a surrogate model is employed, on which finite differencing is performed. When the surrogate model is a Gaussian process (GP), we show that the ASM estimator is available in closed form, rendering the finite-difference approximation unnecessary. We use our closed-form solution to develop acquisition functions focused on sequential learning tailored to sensitivity analysis on top of ASMs. We also show that the traditional ASM estimator may be viewed as a method of moments estimator for a certain class of GPs. We demonstrate how uncertainty on GP hyperparameters may be propagated to uncertainty on the sensitivity analysis, allowing model-based confidence intervals on the active subspace. Our methodological developments are illustrated on several examples.
SU-G-IeP3-09: Investigating the Interplay of Antiscatter Grids with Modern Detectors and Image Processing in Digital Radiography
Purpose: To use phantom and simulation experiments to relate technique factors, patient size and antiscatter grid use to image quality in portable digital radiography (DR), in light of advancements in detector design and image processing. Methods: Image contrast-to-noise ratio (CNR) on a portable DR system (MobileDaRt Evolution, Shimadzu) was measured by imaging four aluminum inserts of varying thickness, superimposed on a Lucite slab phantom using a pediatric abdominal protocol. Three thicknesses of Lucite were used: 6.1cm, 12cm, and 18.2cm, with both 55 and 65 kVp beams. The mAs was set so that detector entrance exposure (DEE) was matched between kVp values. Each technique and phantom was used with and without an antiscatter grid (focused linear grid embedded in aluminum with an 8:1 ratio). The CNR-improvement-factor was then used to determine the thickness- and technique-dependent appropriateness of grid use. Finally, the same experiment was performed via Monte Carlo simulation, integrating incident energy fluence at each detector pixel, so that effects of detector design and image processing could be isolated from physical factors upstream of the detector. Results: The physical phantom experiment demonstrated a clear improvement for the lower tube voltage (55kVp), along with substantial CNR benefits with grid use for 12–18cm phantoms. Neither trend was evident with Monte Carlo, suggesting that suboptimal quantum-detection-efficiency and automated grid-removal could explain trends in kVp and grid use, respectively. Conclusion: Physical experiments demonstrate marked improvement in CNR when using a grid for phantoms of 12 and 18cm Lucite thickness (above ∼10cm soft-tissue equivalent). This benefit is likely due to image processing, as this result was not seen with Monte Carlo. The impact of image processing on image resolution should also be investigated, and the CNR benefit of low kVp and grid use should be weighed against the increased exposure time necessary to achieve adequate DEE.
Threading–Unthreading Transition of Linear-Ring Polymer Blends in Extensional Flow
Adding small amounts of ring polymers to a matrix of their linear counterparts is known to increase the zero-shear-rate viscosity because of linear-ring threading. Uniaxial extensional rheology measurements show that, unlike its pure linear and ring constituents, the blend exhibits an overshoot in the stress growth coefficient. By combining these measurements with ex-situ small-angle neutron scattering and nonequilibrium molecular dynamics simulations, this overshoot is shown here to be driven by a transient threading–unthreading transition of rings embedded within the linear entanglement network. Prior to unthreading, embedded rings deform affinely with the linear entanglement network and produce a measurably stronger elongation of the linear chains in the blend compared to the pure linear melt. Thus, rings uniquely alter the mechanisms of transient elongation in linear polymers.
Modeling the AC Power Flow Equations with Optimally Compact Neural Networks: Application to Unit Commitment
Nonlinear power flow constraints render a variety of power system optimization problems computationally intractable. Emerging research shows, however, that the nonlinear AC power flow equations can be successfully modeled using neural networks. These neural networks can be exactly transformed into mixed integer linear programs and embedded inside challenging optimization problems, thus replacing nonlinearities that are intractable for many applications with tractable piecewise linear approximations. Such approaches, though, suffer from an explosion of the number of binary variables needed to represent the neural network. Accordingly, this paper develops a technique for training an "optimally compact'' neural network, i.e., one that can represent the power flow equations with a sufficiently high degree of accuracy while still maintaining a tractable number of binary variables. We demonstrate the use of this neural network as an approximator of the nonlinear power flow equations by embedding it in the AC unit commitment problem, transforming the problem from a mixed integer nonlinear program into a more manageable mixed integer linear program. We use the 14-, 57-, and 89-bus networks as test cases and compare the AC-feasibility of commitment decisions resulting from the neural network, DC, and linearized power flow approximations. Our results show that the neural network model outperforms both the DC and linearized power flow approximations when embedded in the unit commitment problem. The neural network formulation most often selects a feasible unit commitment schedule, and furthermore, it only s
Development of U-Frame Bending System for Studying the Vibration Integrity of Spent Nuclear Fuel
A bending fatigue system developed to evaluate the response of spent nuclear fuel rods to vibration loads is presented. Design and analysis, fabrication, modification, calibration, and instrumentation are described. The system is composed of a U-frame testing setup for imposing bending loads on the spent fuel rod test specimen and a method for measuring the curvature of the rod during bending. The U-frame setup consists of two rigid arms, linking members, and linkages to a universal testing machine. The test specimen s curvature of bending is obtained through a three-point deflection measurement method consisting of three LVDTs mounted to the side connecting plates of the U-frame to capture the deformation of the test specimen. The system has some unique features: 1) The test specimen is installed by simple insertion using linear bearings incorporated with rigid sleeves. 2) Reverse cyclic bending tests can be carried out effectively and efficiently by push and pull at the loading point of the setup. Any test machine with a linear motion function can be used to drive the setup. 3) The embedded and preloaded linear roller bearings eliminate the backlash that exists in the conventional reverse bend tests. 4) The number of linkages between the U-frame and the universal machine is minimized. Namely, there are only two linkages needed at the two loading points of a U-frame setup, whereas a conventional four/three-point bend test frame requires four linkages. 5) The curvature measurement is immune to the effects arising from compliant layers and the rigid body motion of the machine. The compliant layers are used at the holding areas of the specimen to prevent contact damage. The tests using surrogate specimens composed of SS cladding/tube revealed several important phenomena that may cast light on the expected response of a spent fuel rod: 1) Cyclic quasi-static load (10 N/s under force control) in compressive mode (with respect to that at the loading point of the U-frame) produced increased irreversible or plastic curvature and also increased flexural rigidity of the surrogate rod. 2) Dynamic cyclic load (at least 1 Hz) in compressive mode resulted in increased flexural rigidity of the surrogate rod prior to SS cladding fracture. 3) Pellets and epoxy bonding exhibited various effects on the response of surrogate rods during the loading process as validated from static tests. 4) Dynamic cyclic load (2 Hz) in reverse mode demonstrated a substantial cyclic softening before the fracture of the surrogate rod. The degree of decrease in flexural rigidity was consistent in both measurement and on-line monitoring. The developed U-frame system is thus verified and demonstrated to be ready for further pursuit in hot-cell tests.
GMRES with embedded ensemble propagation for the efficient solution of parametric linear systems in uncertainty quantification of computational models
In a previous work, embedded ensemble propagation was proposed to improve the efficiency of sampling-based uncertainty quantification methods of computational models on emerging computational architectures. It consists of simultaneously evaluating the model for a subset of samples together, instead of evaluating them individually. A first approach introduced to solve parametric linear systems with ensemble propagation is ensemble reduction. In Krylov methods for example, this reduction consists in coupling the samples together using an inner product that sums the sample contributions. Ensemble reduction has the advantages of being able to use optimized implementations of BLAS functions and having a stopping criterion which involves only one scalar. However, the reduction potentially decreases the rate of convergence due to the gathering of the spectra of the samples. In this paper, we investigate a second approach: ensemble propagation without ensemble reduction in the case of GMRES. This second approach solves each sample simultaneously but independently to improve the convergence compared to ensemble reduction. This raises two new issues which are solved in this paper: the fact that optimized implementations of BLAS functions cannot be used anymore and that ensemble divergence, whereby individual samples within an ensemble must follow different code execution paths, can occur. We tackle those issues by implementing a high-performing ensemble GEMV and by using masks. The proposed ensemble GEMV leads to a similar cost per GMRES iteration for both approaches, i.e. with and without reduction. For illustration, we study the performances of the new linear solver in the context of a mesh tying problem. Furthermore, this example demonstrates improved ensemble propagation speed-up without reduction.
Orthogonality of embedded wave functions for different states in frozen-density embedding theory
Other than lowest-energy stationary embedded wave functions obtained in Frozen-Density Embedding Theory (FDET) [T. A. Wesolowski, Phys. Rev. A 77, 012504 (2008)] can be associated with electronic excited states but they can be mutually non-orthogonal. Although this does not violate any physical principles — embedded wave functions are only auxiliary objects used to obtain stationary densities — working with orthogonal functions has many practical advantages. In the present work, we show numerically that excitation energies obtained using conventional FDET calculations (allowing for non-orthogonality) can be obtained using embedded wave functions which are strictly orthogonal. The used method preserves the mathematical structure of FDET and self-consistency between energy, embedded wave function, and the embedding potential (they are connected through the Euler-Lagrange equations). The orthogonality is built-in through the linearization in the embedded density of the relevant components of the total energy functional. Moreover, we show formally that the differences between the expectation values of the embedded Hamiltonian are equal to the excitation energies, which is the exact result within linearized FDET. Linearized FDET is shown to be a robust approximation for a large class of reference densities.