Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Physics regularization”

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 37 records · Page 2

A physics-informed and hierarchically regularized data-driven model for predicting fluid flow through porous media

This paper presents a new deep learning data-driven model for predicting structure dependent pore-fluid velocity fields in rock. The model is based on a Convolutional Auto-Encoder (CAE) artificial neural network capable of learning from image data generated by direct numerical simulations of fluid flow through pore-structures, such as by Lattice Boltzmann or molecular dynamics methods. The main novelty of the model in comparison to previous CAE-based data-driven approaches consists of three parts. The first is a methodology for decomposing the full-domain of the porous media into sub-regions, or “sub-domains”, in order to reduce the overall size of the CAE, batch process the sub-domains in parallel, and enable the CAE to learn local and generalizable nonlinear mappings of pore-fluid velocities. The second consists of embedding the finite difference solutions of the incompressible Navier-Stokes and continuity equations into convolutional layers prior to the CAE in order to provide the CAE with knowledge of fluid dynamics physics (PhyFlow). The third main novelty is that the training of the CAE is regularized with a hierarchical loss function that encourages the learning of fluid flow patterns (in a way similar to ranked modes in principal component analysis), ranking from most to least important. This is shown to increase the stability in learning, reduce over-fitting, and promote interpretability of the CAE neural network layers (HierCAE). The comprehensive new data-driven model, which we call the PhyFlow-HierCAE model, is shown to exhibit improved accuracy and generalizability of flow field predictions over conventional CAE models, attributable to the embedded physical knowledge and the hierarchical regularization, as well as realize orders of magnitude speed-ups in computation times as a surrogate for the direct numerical simulations. Examples of training and forward predictions on unseen pore-structures are provided and evaluated for data from Lattice Boltzmann and molecular dynamics simulations of pore-fluid flow. The model is shown to be a fast and accurate emulator (or “surrogate”) for predicting effective permeability of unseen pore-structures based on learning from relatively small direct numerical simulation datasets.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Regularizing cosmological singularities by varying physical constants

Varying physical constant cosmologies were claimed to solve standard cosmological problems such as the horizon, the flatness and the Λ-problem. In this paper, we suggest yet another possible application of these theories: solving the singularity problem. By specifying some examples we show that various cosmological singularities may be regularized provided the physical constants evolve in time in an appropriate way.

79 ASTRONOMY AND ASTROPHYSICS↗

Differentiable but exact formulation of density-functional theory

The universal density functional F of density-functional theory is a complicated and ill-behaved function of the density—in particular, F is not differentiable, making many formal manipulations more complicated. While F has been well characterized in terms of convex analysis as forming a conjugate pair (E, F) with the ground-state energy E via the Hohenberg–Kohn and Lieb variation principles, F is nondifferentiable and subdifferentiable only on a small (but dense) subset of its domain. In this article, we apply a tool from convex analysis, Moreau–Yosida regularization, to construct, for any ε > 0, pairs of conjugate functionals ({sup ε}E, {sup ε}F) that converge to (E, F) pointwise everywhere as ε → 0{sup +}, and such that {sup ε}F is (Fréchet) differentiable. For technical reasons, we limit our attention to molecular electronic systems in a finite but large box. It is noteworthy that no information is lost in the Moreau–Yosida regularization: the physical ground-state energy E(v) is exactly recoverable from the regularized ground-state energy {sup ε}E(v) in a simple way. All concepts and results pertaining to the original (E, F) pair have direct counterparts in results for ({sup ε}E, {sup ε}F). The Moreau–Yosida regularization therefore allows for an exact, differentiable formulation of density-functional theory. In particular, taking advantage of the differentiability of {sup ε}F, a rigorous formulation of Kohn–Sham theory is presented that does not suffer from the noninteracting representability problem in standard Kohn–Sham theory.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Extreme sparsification of physics-augmented neural networks for interpretable model discovery in mechanics

Data-driven constitutive modeling with neural networks has received increased interest in recent years due to its ability to easily incorporate physical and mechanistic constraints and to overcome the challenging and time-consuming task of formulating phenomenological constitutive laws that can accurately capture the observed material response. However, even though neural network-based constitutive laws have been shown to generalize proficiently, the generated representations are not easily interpretable due to their high number of trainable parameters. Sparse regression approaches exist that allow for obtaining interpretable expressions, but the user is tasked with creating a library of model forms which by construction limits their expressiveness to the functional forms provided in the libraries. Here, in this work, we propose to train regularized physics-augmented neural network-based constitutive models utilizing a smoothed version of $L^0$-regularization. This aims to maintain the trustworthiness inherited by the physical constraints, but also enables interpretability which has not been possible thus far on any type of machine learning-based constitutive model where model forms were not assumed a priori but were actually discovered. During the training process, the network simultaneously fits the training data and penalizes the number of active parameters, while also ensuring constitutive constraints such as thermodynamic consistency. We show that the method can reliably obtain interpretable and trustworthy constitutive models for compressible and incompressible hyperelasticity, yield functions, and hardening models for elastoplasticity, using synthetic and experimental data. This work aims to set a new paradigm for interpretable machine learning models in the broad area of solid mechanics where low and limited data is available along with prior knowledge of physical constraints that the learned maps need to obey. This paradigm can potentially be extended to a broader spectrum of scientific exploration.

Data-driven constitutive models↗

Physics-Driven Convolutional Autoencoder Approach for CFD Data Compressions: Preprint

With the growing size and complexity of turbulent flow models, data compression approaches are of the utmost importance to analyze, visualize, or restart the simulations. Recently, in-situ autoencoder-based compression approaches have been proposed and shown to be effective at producing reduced representations of turbulent flow data. However, these approaches focus solely on training the model using point-wise sample reconstruction losses that do not take advantage of the physical properties of turbulent flows. In this paper, we show that training autoencoders with additional physics-informed regularizations, e.g., enforcing incompressibility and preserving enstrophy, improves the compression model in three ways: (i) the compressed data better conform to known physics for homogeneous isotropic turbulence without negatively impacting point-wise reconstruction quality, (ii) inspection of the gradients of the trained model uncovers changes to the learned compression mapping that can facilitate the use of explainability techniques, and (iii) as a performance byproduct, training losses are shown to converge up to 12x faster than the baseline model.

auto-encoders↗

Eco-PiNN: A Physics-informed Neural Network for Eco-toll Estimation

The eco-toll estimation problem quantifies the expected environmental cost (e.g., energy consumption, exhaust emissions) for a vehicle to travel along a path. This problem is important for societal applications such as eco-routing, which aims to find paths with the lowest exhaust emissions or energy need. The challenges of this problem are threefold: (1) the dependence of a vehicle's eco-toll on its physical parameters; (2) the lack of access to data with eco-toll information; and (3) the influence of contextual information (i.e. the connections of adjacent segments in the path) on the eco-toll of road segments. Prior work on eco-toll estimation has mostly relied on pure data-driven approaches and has high estimation errors given the limited training data. To address these limitations, we propose a novel Eco-toll estimation Physics-informed Neural Network framework (Eco-PiNN) using three novel ideas, namely, (1) a physics-informed decoder that integrates the physical laws governing vehicle dynamics into the network, (2) an attention-based contextual information encoder, and (3) a physics-informed regularization to reduce overfitting. Experiments on real-world heavy-duty truck data show that the proposed method can greatly improve the accuracy of eco-toll estimation compared with state-of-the-art methods.

97 MATHEMATICS AND COMPUTING↗

A novel numerical treatment of the near-wall regions in the k – ω class of RANS models

In this study, we discuss a novel approach to modeling the near-wall region in the class of k – ω to models. The proposed methodology obviates the need for ad hoc boundary conditions of to omega on the wall as typically required in the k – ω model. The primary motivation of this work is to provide a formulation equivalent to the standard k – ω to and k – ω to SST models, but which at the same time overcomes their limitations in the context of their implementation in high-order methods. This is achieved by subtracting the asymptotically known singular behavior of omega at walls. Imposing a grid-dependent value of omega at walls in high-order codes is not straightforward as is demonstrated below and it causes instability as well as accuracy issues. The mathematical formulation of the two novel approaches, termed as the "regularized k – ω model" and the "regularized k – ω OJ SST model", is discussed in detail. A consistency and verification study for these two approaches is performed by proving that the regularized models recover the results of the standard models in various canonical problems, such as turbulent channel and pipe flows, and a systematic investigation of convergence, using both p- (polynomial order) as well as h- (grid) refinement is reported. Furthermore, comparisons highlighting the performance of the proposed methods in more complex configurations such as flow over a backward facing step and the turbulent mixing of fluid streams of different temperatures in a T-junction are also presented.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Emergent gravity from vanishing energy-momentum tensor

A constraint of vanishing energy-momentum tensor is motivated by a variety of perspectives on quantum gravity. We demonstrate in a concrete example how this constraint leads to a metric-independent theory in which quantum gravity emerges as a nonperturbative artifact of regularization-scale physics. We analyze a scalar theory similar to the Dirac-Born-Infeld (DBI) theory with vanishing gauge fields, with the DBI Lagrangian modulated by a scalar potential. In the limit of a large number of scalars, we explicitly demonstrate the existence of a composite massless spin-2 graviton in the spectrum that couples to matter as in Einstein gravity. As a result, we comment on the cosmological constant problem and the generalization to theories with fermions and gauge fields.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Islands in non-minimal dilaton gravity: exploring effective theories for black hole evaporation

We start from (3 + 1)-dimensional Einstein gravity with minimally coupled massless scalar matter, through spherical dimensional reduction, the matter theory is non-minimally coupled with the dilaton in (1 + 1)-dimensions. Despite its simplicity, constructing a self-consistent one-loop effective theory for this model remains a challenge, partially due to a Weyl-invariant ambiguity in the effective action. With a universal splitting property for the one-loop action, the ambiguity can be identified with the state-dependent part of the covariant quantum stress tensor. By introducing on-shell equivalent auxiliary fields to construct minimal candidates of Weyl-invariant terms, we derive a one-parameter family of one-loop actions with unique, regular, and physical stress tensors corresponding to the Boulware, Hartle-Hawking and Unruh states. We further study the back-reacted geometry and the corresponding quantum extremal islands that were inaccessible without a consistent one-loop theory. Along the way, we elaborate on the implications of our construction for the non-minimal dilaton gravity model.

2D Gravity↗

Machine-learning predictions of the shale wells’ performance

The ultra-low permeability nature of shale reservoirs leads to an extended linear flow and necessitates horizontal wells with multi-stage engineered fractures to efficiently extract hydrocarbons resources. These artificially-generated and naturally-occurring fractures form complex networks that create complex flow regimes which control oil production. These fractures are neither identical nor equally-spaced, which leads to a production profile with a masked onset of the boundary-dominated flow. The combination of the extended linear flow with the indeterminate onset of the boundary-dominated flow challenges the current deterministic analytic approaches to forecast the estimated ultimate recovery (EUR). In this work, we propose a novel machine-learning approach which overcomes these challenges and provides reliable EUR estimates based on field-wide analyses. We implement a novel unsupervised machine learning (ML) methodology, which allows for automatic identification of the optimal number of features (signals) present in the data based on non-negative matrix/tensor factorization coupled with k-means clustering incorporating regularization and physics constraints. In the presented analyses, the input data to the ML algorithm is the available (public) production history from the field collected at existing unconventional reservoirs. We validate our approach through hindcasting of the production data, where we achieved an excellent agreement. In addition, our approach is able to identify the poorly-performing wells, which could benefit from early refracing. Our approach provides fast and accurate estimations of the well performance without presumptions about the state of the well or the flow regime.

03 NATURAL GAS↗

MO-DE-BRA-02: From Teaching to Learning: Systems-Based-Practice and Practice-Based-Learning Innovations in Medical Physics Education Programs

Purpose: The increasing complexity in the field of radiation medicine and concomitant rise in patient safety concerns call for enhanced systems-level training for future medical physicists and thus commensurate innovations in existing educational program curricula. In this work we report on the introduction of three learning opportunities to augment medical physics educational programs towards building systems-based practice and practice-based learning competencies. Methods: All initiatives were introduced for senior -level graduate students and physics residents in an institution with a newly established medical-physics graduate program and therapeutic-physics residency program. The first, centered on incident learning, was based on a spreadsheet tool that incorporated the reporting structure of the Radiation Oncology-incident Learning System (ROILS), included 120 narratives of published incidents and enabled inter-rater variability calculations. The second, centered on best-practices, was a zero-credit seminar course, where students summarized select presentations from the AAPM virtual library on a weekly basis and moderated class discussions using a point/counterpoint approach. Presentation styles were critiqued. The third; centered on learning-by-teaching, required physics residents to regularly explain fundamental concepts in radiological physics from standard textbooks to board certified physics faculty members. Results: Use of the incident-learning system spreadsheet provided a platform to recast known accidents into the framework of ROILS, thereby increasing awareness of factors contributing to unsafe practice and appreciation for inter-rater variability. The seminar course enhanced awareness of best practices, the effectiveness of presentation styles and encouraged critical thinking. The learn-by-teaching rotation allowed residents to stay abreast of and deepen their knowledge of relevant subjects. Conclusion: The incorporation of systems-driven initiatives broadens comprehension of the wider systems context of medical physics, enhances awareness of resources for innovation, communication and sustained learning while maintaining a metric-driven focus on patient safety within the formative phase of student careers. The initiatives were well-received, feasible, and utilized available or shared-resources translatable across educational programs.

60 APPLIED LIFE SCIENCES↗

Constructing Neural Network Based Models for Simulating Dynamical Systems

Dynamical systems see widespread use in natural sciences like physics, biology, and chemistry, as well as engineering disciplines such as circuit analysis, computational fluid dynamics, and control. For simple systems, the differential equations governing the dynamics can be derived by applying fundamental physical laws. However, for more complex systems, this approach becomes exceedingly difficult. Data-driven modeling is an alternative paradigm that seeks to learn an approximation of the dynamics of a system using observations of the true system. In recent years, there has been an increased interest in applying data-driven modeling techniques to solve a wide range of problems in physics and engineering. Here this article provides a survey of the different ways to construct models of dynamical systems using neural networks. In addition to the basic overview, we review the related literature and outline the most significant challenges from numerical simulations that this modeling paradigm must overcome. Based on the reviewed literature and identified challenges, we provide a discussion on promising research areas.

97 MATHEMATICS AND COMPUTING↗

Developing and Managing Data Acquisition Software Using Spack

The Data Acquisition systems of particle physics experiments regularly push the boundaries of high-throughput computing, demanding low-latency collection of data from thousands of devices, collating data into time-sliced events, processing these events and making trigger decisions, and writing the selected data streams to disk. To accomplish these tasks, the DAQ Engineering and Operations department at Fermilab leverages multiple software libraries and builds reusable DAQ frameworks on top. These libraries must be delivered in well-defined bundles and are thoroughly tested for compatibility and functionality before being deployed to live detectors. We have several techniques used to ensure that a consistent set of dependencies can be delivered and re-created at need. We must also support active development of DAQ software components, ideally in an environment as close as possible to that of the detectors. This development often occurs across multiple packages which have to be built in concert and features tested in a consistent and reproducible manner. I will present our scheme for accomplishing these goals using Spack environments, bundle packages, and Github Actions-based CI.

Flumerfelt, Eric [Fermilab]↗

Profiles of Gamma-Ray Bursts and Their Component Pulses

One physically informative regularity of their otherwise heterogeneous ensemble, is that many Gamma-Ray Bursts consist of well defined pulses. To objectively quantify the temporal structure of BATSE bursts, we have developed an automatic modeling procedure that separates overlapping pulses and determines the energy-dependence of the pulse-shape parameters. No binning of photon arrival times is needed, so when applied to time-tagged events (TTE) the procedure captures variability information down to the shortest time scales present in the raw data. Maximizing the Bayesian likelihood function Pr(data/model) yields estimates of the model parameters, including the number of pulses present, and allows intercomparison of models of different forms. As with any nonlinear optimization, good initial guesses are crucial to avoid convergence to undesirable local minima. We find excellent initial pulse decompositions by wavelet-denoising a cumulative distribution of the raw photon arrival data; differentiation then gives a time profile mostly free of the systematic effects of degraded resolution (as in ordinary Fourier smoothing) and binning. We present statistical information on pulse rise-time, decay-time, peakedness, and amplitudes, plus their energy dependences - both within a single burst and for a large ensemble of bursts.

Scargle, Jeff D.↗

Gibbs Measures Over Locally Tree-Like Graphs and Percolative Entropy Over Infinite Regular Trees

Consider a statistical physical model on the d-regular infinite tree T{sub d} described by a set of interactions Φ. Let {G_n} be a sequence of finite graphs with vertex sets V{sub n} that locally converge to T{sub d}. From Φ one can construct a sequence of corresponding models on the graphs G{sub n}. Let {μ_n} be the resulting Gibbs measures. Here we assume that {μ_n} converges to some limiting Gibbs measure μ on T{sub d} in the local weak{sup ∗} sense, and study the consequences of this convergence for the specific entropies |V{sub n}|{sup −1}H(μ{sub n}). We show that the limit supremum of |V{sub n}|{sup −1}H(μ{sub n}) is bounded above by the percolative entropyH{sub perc}(μ), a function of μ itself, and that |V{sub n}|{sup −1}H(μ{sub n}) actually converges to H{sub perc}(μ) in case Φ exhibits strong spatial mixing on T{sub d}. When it is known to exist, the limit of |V{sub n}|{sup −1}H(μ{sub n}) is most commonly shown to be given by the Bethe ansatz. Percolative entropy gives a different formula, and we do not know how to connect it to the Bethe ansatz directly. We discuss a few examples of well-known models for which the latter result holds in the high temperature regime.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Magnetic field mapping of inaccessible regions using physics-informed neural networks

A difficult problem concerns the determination of magnetic field components within an experimentally inaccessible region when direct field measurements are not feasible. In this paper, we propose a new method of accessing magnetic field components using non-disruptive magnetic field measurements on a surface enclosing the experimental region. Magnetic field components in the experimental region are predicted by solving a set of partial differential equations (Ampere’s law and Gauss’ law for magnetism) numerically with the aid of physics-informed neural networks (PINNs). Prediction errors due to noisy magnetic field measurements and small number of magnetic field measurements are regularized by the physics information term in the loss function. We benchmark our model by comparing it with an older method. The new method we present will be of broad interest to experiments requiring precise determination of magnetic field components, such as searches for the neutron electric dipole moment.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

An adaptive and stability-promoting layerwise training approach for sparse deep neural network architecture

This work presents a two-stage adaptive framework for progressively developing deep neural network (DNN) architectures that generalize well for a given training data set. In the first stage, a layerwise training approach is adopted where a new layer is added each time and trained independently by freezing parameters in the previous layers. We impose desirable structures on the DNN by employing manifold regularization, sparsity regularization, and physics-informed terms. We introduce a ε – δ – stability-promoting concept as a desirable property for a learning algorithm and show that employing manifold regularization yields a ε – δ stability-promoting algorithm. Further, we also derive the necessary conditions for the trainability of a newly added layer and investigate the training saturation problem. In the second stage of the algorithm (post-processing), a sequence of shallow networks is employed to extract information from the residual produced in the first stage, thereby improving the prediction accuracy. Numerical investigations on prototype regression and classification problems demonstrate that the proposed approach can outperform fully connected DNNs of the same size. Moreover, by equipping the physics-informed neural network (PINN) with the proposed adaptive architecture strategy to solve partial differential equations, we numerically show that adaptive PINNs not only are superior to standard PINNs but also produce interpretable hidden layers with provable stability. As a result, we also apply our architecture design strategy to solve inverse problems governed by elliptic partial differential equations.

42 ENGINEERING↗