Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “causal models”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 217 records · Page 12

Respecting causality for training physics-informed neural networks

While the popularity of physics-informed neural networks (PINNs) is steadily rising, to this date PINNs have not been successful in simulating dynamical systems whose solution exhibits multi-scale, chaotic or turbulent behavior. In this work we attribute this shortcoming to the inability of existing PINNs formulations to respect the spatio-temporal causal structure that is inherent to the evolution of physical systems. We argue that this is a fundamental limitation and a key source of error that can ultimately steer PINN models to converge towards erroneous solutions. We address this pathology by proposing a simple re-formulation of PINNs loss functions that can explicitly account for physical causality during model training. We demonstrate that this simple modification alone is enough to introduce significant accuracy improvements, as well as a practical quantitative mechanism for assessing the convergence of a PINNs model. Here we provide state-of-the-art numerical results across a series of benchmarks for which existing PINNs formulations fail, including the chaotic Lorenz system, the Kuramoto–Sivashinsky equation in the chaotic regime, and the Navier–Stokes equations. To the best of our knowledge, this is the first time that PINNs have been successful in simulating such systems, introducing new opportunities for their applicability to problems of industrial complexity.

97 MATHEMATICS AND COMPUTING↗

Energy-momentum diffusion from spacetime discreteness

We study potentially observable consequences of spatiotemporal discreteness for the motion of massive and massless particles. First we describe some simple models for the motion of a massive point particle in a fixed causal set background. If the causal set is faithfully embeddable in Minkoswki spacetime, the models give rise to particle motion in the continuum spacetime. At large scales, the microscopic swerves induced by the underlying atomicity manifest themselves as a Lorentz invariant diffusion in energy-momentum governed by a single phenomenological parameter, and we derive in full the corresponding diffusion equation. Inspired by the simplicity of the result, we then derive the most general Lorentz invariant diffusion equation for a massless particle, which turns out to contain two phenomenological parameters describing, respectively, diffusion and drift in the particle's energy. The particles do not leave the light cone however: their worldlines continue to be null geodesics. Finally, we deduce bounds on the drift and diffusion constants for photons from the blackbody nature of the spectrum of the cosmic microwave background radiation.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Diagnosis and Reconfiguration using Bayesian Networks: An Electrical Power System Case Study

Automated diagnosis and reconfiguration are important computational techniques that aim to minimize human intervention in autonomous systems. In this paper, we develop novel techniques and models in the context of diagnosis and reconfiguration reasoning using causal Bayesian networks (BNs). We take as starting point a successful diagnostic approach, using a static BN developed for a real-world electrical power system. We discuss in this paper the extension of this diagnostic approach along two dimensions, namely: (i) from a static BN to a dynamic BN; and (ii) from a diagnostic task to a reconfiguration task. More specifically, we discuss the auto-generation of a dynamic Bayesian network from a static Bayesian network. In addition, we discuss subtle, but important, differences between Bayesian networks when used for diagnosis versus reconfiguration. We discuss a novel reconfiguration agent, which models a system causally, including effects of actions through time, using a dynamic Bayesian network. Though the techniques we discuss are general, we demonstrate them in the context of electrical power systems (EPSs) for aircraft and spacecraft. EPSs are vital subsystems on-board aircraft and spacecraft, and many incidents and accidents of these vehicles have been attributed to EPS failures. We discuss a case study that provides initial but promising results for our approach in the setting of electrical power systems.

Knox, W. Bradley↗

Deep Koopman operators for causal discovery

Causal discovery aims to identify cause-effect mechanisms for better scientific understanding, explainable decision-making, and more accurate modeling. Standard statistical frameworks, such as Granger causality, lack the ability to quantify causal relationships in nonlinear dynamics due to the presence of complex feedback mechanisms, timescale mixing, and nonstationarity. Thus, applying these methods to study causal dynamics in real-world systems, such as the Earth, is a major challenge. Addressing this shortcoming, we leverage deep learning and a Koopman operator-theoretic formalism to present a class of causal discovery algorithms. Kausal uses deep Koopman operator methods to approximate nonlinear dynamics in a linearized vector space in which traditional causal inference methods such as Granger causality can be more easily applied. Our idealized experiments demonstrate Kausal’s superior ability in discovering and characterizing causal signals compared to existing deep learning and non-deep learning state-of-the-art approaches. Finally, the successful identification of major El Niño and La Niña events in observations showcases Kausal’s skill to handle real-world applications.

54 ENVIRONMENTAL SCIENCES↗

Concussion, microvascular injury, and early tauopathy in young athletes after impact head injury and an impact concussion mouse model

The mechanisms underpinning concussion, traumatic brain injury, and chronic traumatic encephalopathy, and the relationships between these disorders, are poorly understood. We examined post-mortem brains from teenage athletes in the acute-subacute period after mild closed-head impact injury and found astrocytosis, myelinated axonopathy, microvascular injury, perivascular neuroinflammation, and phosphorylated tau protein pathology. To investigate causal mechanisms, we developed a mouse model of lateral closed-head impact injury that uses momentum transfer to induce traumatic head acceleration. Unanaesthetized mice subjected to unilateral impact exhibited abrupt onset, transient course, and rapid resolution of a concussion-like syndrome characterized by altered arousal, contralateral hemiparesis, truncal ataxia, locomotor and balance impairments, and neurobehavioural deficits. Experimental impact injury was associated with axonopathy, blood–brain barrier disruption, astrocytosis, microgliosis (with activation of triggering receptor expressed on myeloid cells, TREM2), monocyte infiltration, and phosphorylated tauopathy in cerebral cortex ipsilateral and subjacent to impact. Phosphorylated tauopathy was detected in ipsilateral axons by 24 h, bilateral axons and soma by 2 weeks, and distant cortex bilaterally at 5.5 months post-injury. Impact pathologies co-localized with serum albumin extravasation in the brain that was diagnostically detectable in living mice by dynamic contrast-enhanced MRI. These pathologies were also accompanied by early, persistent, and bilateral impairment in axonal conduction velocity in the hippocampus and defective long-term potentiation of synaptic neurotransmission in the medial prefrontal cortex, brain regions distant from acute brain injury. Surprisingly, acute neurobehavioural deficits at the time of injury did not correlate with blood–brain barrier disruption, microgliosis, neuroinflammation, phosphorylated tauopathy, or electrophysiological dysfunction. Furthermore, concussion-like deficits were observed after impact injury, but not after blast exposure under experimental conditions matched for head kinematics. Computational modelling showed that impact injury generated focal point loading on the head and seven-fold greater peak shear stress in the brain compared to blast exposure. Moreover, intracerebral shear stress peaked before onset of gross head motion. By comparison, blast induced distributed force loading on the head and diffuse, lower magnitude shear stress in the brain. We conclude that force loading mechanics at the time of injury shape acute neurobehavioural responses, structural brain damage, and neuropathological sequelae triggered by neurotrauma. These results indicate that closed-head impact injuries, independent of concussive signs, can induce traumatic brain injury as well as early pathologies and functional sequelae associated with chronic traumatic encephalopathy. These results also shed light on the origins of concussion and relationship to traumatic brain injury and its aftermath.

59 BASIC BIOLOGICAL SCIENCES↗

Domain knowledge-informed, process-mapping AI graph for designing Fe-based alloys

<span style="font-family: Calibri, sans-serif; font-size: 12pt;">Continuous improvement in efficiency of a power plant relies on designing materials for use at increasingly higher temperature and/or pressure, for 100,000s hours of operation. Due to complexity, non-linearity and high-dimensionality of the problem, traditional Machine Learning (ML) approaches require unreasonably large datasets for the data-driven model development. Science-based material and process engineering complements hard data with, sometimes soft and intuitive, empirical domain knowledge. Artificial Intelligence (AI) was used in this study to incorporate such knowledge into computational graph architecture (process-mimicking artificial neuron design, causal layer and graph structures, ensemble modeling of latent states) and learning procedures (variable transformation, fuzzy physics pre-training and freezing of deep layers, virtual microstructure representation, and adversarial multi-objective optimization). The first alloys design pathways suggested by the AI tool (pyroMind) passed a preliminary engineering review on soundness and transparency.</span>

Romanov, Vyacheslav↗

Significant Findings: Tracking the SeaWiFS Record with a Coupled Physical/Biogeochemical/Radiative Model of the Global Oceans

The Sea-Viewing Wide Field-of-view Sensor (SeaWiFS) has observed 2.5 years of routine global chlorophyll observations from space. The mission was launched into a record El Nino event, which eventually gave way to one of the most intensive and longest-lasting La Nina events ever recorded. The SeaWiFS chlorophyll record captured the response of ocean phytoplankton to these significant events in the tropical Indo-Pacific basins, but also indicated significant interannual variability unrelated to the El Nino/La Nina events. This included large variability in the North Atlantic and Pacific basins, in the North Central and equatorial Atlantic, and milder patterns in the North Central Pacific. This SeaWiFS record was tracked with a coupled physical/biogeochemical/radiative model of the global oceans using near-real-time forcing data such as wind stresses, sea surface temperatures, and sea ice. This provided an opportunity to offer physically and biogeochemically meaningful explanations of the variability observed in the SeaWiFS data set, since the causal mechanisms and interrelationships of the model are completely understood. The coupled model was able to represent the seasonal distributions of chlorophyll during the SeaWiFS era, and was capable of differentiating among the widely different processes and dynamics occurring in the global oceans. The model was also reasonably successful in representing the interannual signal, especially when it was large, such as, the El Nino and La Nina events in the tropical Pacific and Indian Oceans. The model provided different phytoplankton group responses for the different events in these regions: diatoms were predominant in the tropical Pacific during the La Nina but other groups were predominant during El Nino. The opposite condition occurred in the tropical Indian Ocean. Both situations were due to the different responses of the basins to El Nino. The interannual variability in the North Atlantic, which was exhibited in SeaWiFS data as a decline in the spring/summer bloom in 1999 relative to 1998, resulted in the model from a more slowly shoaling mixed layer, allowing herbivore populations to keep pace with increasing phytoplankton populations. However, several aspects of the interannual cycle were not well-represented by the model. Explanations ranged from inherent model deficiencies, to monthly averaging of forcing fields, to biases in SeaWiFS atmospheric correction procedures.

Watson, Gregg W.↗

Crime prediction modeling

A study of techniques for the prediction of crime in the City of Los Angeles was conducted. Alternative approaches to crime prediction (causal, quasicausal, associative, extrapolative, and pattern-recognition models) are discussed, as is the environment within which predictions were desired for the immediate application. The decision was made to use time series (extrapolative) models to produce the desired predictions. The characteristics of the data and the procedure used to choose equations for the extrapolations are discussed. The usefulness of different functional forms (constant, quadratic, and exponential forms) and of different parameter estimation techniques (multiple regression and multiple exponential smoothing) are compared, and the quality of the resultant predictions is assessed.

Source record↗

Inflation Basics

The last few years have yielded remarkable discoveries in physics. In particle physics it appears that a fundamental scalar field exists. The Higgs boson is measured to have a mass of about 126 GeV and to have spin zero and positive parity. The Higgs field is the first fundamental scalar to be discovered in physics. The Cosmic Microwave Background, CMB, is known to have a uniform temperature to parts per 10 5 , but has well measured fluctuations, which are thought to evolve gravitationally to provide the seeds of the current structure of the Universe. In addition, the Universe appears to contain, at present, an unknown “dark energy”, which is presently the majority energy density of the Universe, larger than either matter or radiation. This may, indeed, be a fundamental scalar field like the Higgs. “Big Bang” (BB) cosmology is a very successful “standard model” in cosmology. However, it cannot explain the uniformity of the CMB because the CMB consists of many regions not causally connected in the context of the BB model. In addition, the Universe appears to be spatially flat. However, in BB cosmology the present spatial curvature is not stable, so that the initial conditions for BB cosmology would need to be fantastically fine-tuned in order to successfully predict the presently small value of the observed curvature. These issues for BB cosmology have led to the hypothesis of “inflation” which postulates an unknown scalar field, not presumably the Higgs field or the dark energy, which causes an exponential expansion of the Universe at very early times. This attractive hypothesis can account for the problems in BB cosmology of flatness and causal CMB connectivity. In addition, the quantum fluctuations of this postulated field provide a natural explanation of the CMB fluctuations which are the seeds of the structure of galaxies. Researchers are now searching for gravitational waves imprinted on the CMB. These would be a “smoking gun” for inflation since metrical fluctuations, both scalar and tensor, are also produced in inflationary models. Thus, the time appears to be appropriate for a very basic and simple exposition of the inflationary model written from a particle physics perspective. Only the simplest scalar model will be explored because it is easy to understand and contains all the basic elements of the inflationary model.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Overview of the Nephele Perturbed Parameter Ensemble for Aerosol‐Cloud Interactions in E3SMv3

Aerosol-cloud interactions (aci) are the leading source of uncertainty in inferring climate sensitivity from the historical record. Earth system models (ESMs) struggle to represent aci because the processes responsible for these phenomena occur at much finer time and space scales than can be resolved by any ESM. Observational constraints provide key benchmarks to test ESMs, but cannot be used alone to fully understand aci processes except in very specific cases where causality is controlled; some degree of modeling is required to infer aci and estimate radiative forcing. Here, we generate and characterize a perturbed parameter ensemble (PPE) in version 3 of the Energy Exascale ESM (E3SMv3). We perturb 25 parameters that govern aci processes over 250 members and integrate the model over present-day and preindustrial aerosol emissions. We find that the process representation in E3SMv3 is flexible and can generate global-mean effective radiative forcings due to aci (ERFaci) ranging from −3.0 to +0.9 W m −2 . The positive ERFaci values simulated by a portion of the PPE are implausible and result from parameter combinations that produce unrealistic top-of-atmosphere energy fluxes. While global-mean cloud droplet number concentration always increases in response to anthropogenic aerosol, cloud liquid water path can both increase and decrease, suggesting that precipitation suppression is not the only aerosol-cloud adjustment represented by E3SMv3. Analysis of which processes control liquid cloud adjustment in the PPE points toward stratiform precipitation processes and aerosol activation, which is consistent with many previous ESMs, as well as the new two-moment convective cloud microphysics in E3SMv3.

Nugent, Jacqueline M. [Univ. of Wyoming, Laramie, ↗

Time-varying coefficient vector autoregressions model based on dynamic correlation with an application to crude oil and stock markets

This paper proposes a new time-varying coefficient vector autoregressions (VAR) model, in which the coefficient is a linear function of dynamic lagged correlation. The proposed model allows for flexibility in choices of dynamic correlation models (e.g. dynamic conditional correlation generalized autoregressive conditional heteroskedasticity (GARCH) models, Markov-switching GARCH models and multivariate stochastic volatility models), which indicates that it can describe many types of time-varying causal effects. Time-varying causal relations between West Texas Intermediate (WTI) crude oil and the US Standard and Poor’s 500 (S&P 500) stock markets are examined by the proposed model. The empirical results show that their causal relations evolve with time and display complex characters. Both positive and negative causal effects of the WTI on the S&P 500 in the subperiods have been found and confirmed by the traditional VAR models. Similar results have been obtained in the causal effects of S&P 500 on WTI. In addition, the proposed model outperforms the traditional VAR model.

54 ENVIRONMENTAL SCIENCES↗

THE r-PROCESS IN PROTO-NEUTRON-STAR WIND REVISITED

We examine the r-process in the neutrino-driven proto-neutron-star (PNS) wind of core-collapse supernovae in light of the recent findings of massive neutron stars in binaries as well as of an indication of neutron-richness in the PNS ejecta because of the nucleon potential corrections on neutrino opacities. To this end, a spherically symmetric, general relativistic, steady-state wind model is applied for a wide range of PNS masses between 1.2 M{sub Sun} and 2.4 M{sub Sun} with the latter reaching the causality limit. Nucleosynthesis calculations with these PNS models are performed by assuming a time evolution of electron fraction with its minimal value of Y{sub e} = 0.4, which mimics recent hydrodynamical results. The fundamental nucleosynthetic aspect of the PNS wind is found to be the production of Sr, Y, and Zr in quasi-equilibrium and of the elements with A Almost-Equal-To 90-110 by a weak r-process, which can be an explanation for the abundance signatures in r-process-poor Galactic halo stars. PNSs more massive than 2.0 M{sub Sun} can eject heavy r-process elements, however, with substantially smaller amount than what is needed to account for the solar content. PNS winds can be thus the major origin of light trans-iron elements but no more than 10% of those heavier than A {approx} 110, although they may be the sources of the low-level abundances of Sr and Ba found in numerous metal-poor stars if the maximum mass of PNSs exceeds 2.0 M{sub Sun }.

79 ASTRONOMY AND ASTROPHYSICS↗

The emissions of broadband electrostatic noise in the near vicinity of the Shuttle Orbiter

Measurements of the Space Shuttle environment from the STS-3 and Spacelab 2 missions indicate the presence of oblique ion streams and broadband electrostatic noise. A two-dimensional theoretical model is applied to study a possible causal relationship between the ion streams and the broadband noise, especially in terms of the ion acoustic wave and ion-ion wave modes. This model predicts the generation of waves with frequencies ranging from the ion cyclotron frequency up to values greater than the ion plasma frequency, with the maximum growth rate occurring in the 10-kHz range. These results are consistent with the observational data from the STS-3 mission. The model also shows that these two wave modes can coexist only when the wave vectors of the two wave modes are nearly perpendicular. The parametric dependence of the wave instabilities on the plasma parameters and the inclination of the wave propagation vector is also studied.

Hwang, K. S.↗

Exact quantisation of the relativistic Hopfield model

We investigate the quantisation in the Heisenberg representation of a relativistically covariant version of the Hopfield model for dielectric media, which entails the interaction of the quantum electromagnetic field with the matter dipole fields, represented by a mesoscopic polarisation field. A full quantisation of the model is provided in a covariant gauge, with the aim of maintaining explicit relativistic covariance. Breaking of the Lorentz invariance due to the intrinsic presence in the model of a preferred reference frame is also taken into account. Relativistic covariance forces us to deal with the unphysical (scalar and longitudinal) components of the fields, furthermore it introduces, in a more tricky form, the well-known dipole ghost of standard QED in a covariant gauge. In order to correctly dispose of this contribution, we implement a generalised Lautrup trick. Furthermore, causality and the relation of the model with the Wightman axioms are also discussed.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Discovering State-Parameter Mappings in Subsurface Models Using Generative Adversarial Networks

A fundamental problem in geophysical modeling is related to the identification and approximation of causal structures among physical processes. However, resolving the bidirectional mappings between physical parameters and model state variables (i.e., solving the forward and inverse problems) is challenging, especially when parameter dimensionality is high. Deep learning has opened a new door toward knowledge representation and complex pattern identification. In particular, the recently introduced generative adversarial networks (GANs) hold strong promises in learning cross-domain mappings for image translation. This study presents a state-parameter identification GAN (SPID-GAN) for simultaneously learning bidirectional mappings between a high-dimensional parameter space and the corresponding model state space. SPID-GAN is demonstrated using a series of representative problems from subsurface flow modeling. Results show that SPID-GAN achieves satisfactory performance in identifying the bidirectional state-parameter mappings, providing a new deep-learning-based, knowledge representation paradigm for a wide array of complex geophysical problems.

58 GEOSCIENCES↗

The Sufficient Component Cause Model Explaining Low-Dose Radiobiology through an Epidemiologic Lens

One of the major challenges in understanding space radiation-induced carcinogenesis is the uncertainty from translating radiobiological research in cellular and animal models to humans, especially at doses below 100 mSv. Biological studies have shown a host of potential outcomes at low doses in animal and cellular models. The existence of non-targeted effects as bystander and abscopal effects is well-documented from clinical research and basic science1,2,3. While some studies have shown increased effects at low doses, others have shown evidence of hormetic bystander effects, where radiation exposure may be beneficial4,5,6. Despite these varied and diverse findings, epidemiologic studies largely support the linear-no threshold (LNT) assumption used by radiation protection guidance7,8. Translational animal-to-human models have predominantly considered ratio values such as the relative biological effectiveness (RBE) and dose and dose-rate effectiveness factor (DDREF) that rely on the assumption of LNT rather than implementing specific dose-response shapes in the low-dose region, and translational cell-to-human models are uncommon. The sufficient component cause model presents an opportunity to examine biological findings at low doses from an epidemiological lens9. In this model, exposures “sufficient” to cause an outcome of interest are presented in pie charts, such that when all slices of a pie chart are fulfilled, the outcome will occur. Multiple pie charts may exist for a single outcome, illustrating individual differences9. In 1988,Greenland and Poole adapted the sufficient component cause model to incorporate interaction with other exposures (such as genetics or lifestyle factors)10. They show that a range of biological processes are possible in a population, but that an epidemiologic study will only reveal the mean outcome from the population at large10,11. Using this construct, the multiple outcomes presented in radiobiological models to date can be explained in the context of epidemiologic studies. This presentation aims to demonstrate a causal framework that can integrate radiobiological and epidemiological models to date. It is intended as a conversation starter to spur future research.

C M Milder↗

The Potential Outcome Model: Explaining Low-Dose Radiobiology through an Epidemiologic Lens

One of the major challenges in understanding space radiation-induced carcinogenesis is the uncertainty from translating radiobiological research in cellular and animal models to humans, especially at doses below 100 mSv. Biological studies have shown a host of potential outcomes at lowdoses in animal and cellular models. The existence of non-targeted effects as bystander and abscopal effects is well-documented from clinical research and basic science1,2,3. While some studies have shown increased effects at low doses, others have shown evidence of hormetic bystander effects, where radiation exposure may be beneficial4,5,6. Despite these varied and diverse findings, epidemiologic studies largely support the linear-no threshold (LNT) assumption used by radiation protection guidance7,8. Translational animal-to-human models have predominantly considered ratio values such as the relative biological effectiveness (RBE) and dose and dose-rate effectiveness factor (DDREF) that rely on the assumption of LNT rather than implementing specific dose-response shapes in the low-dose region, and translational cell-to-human models are uncommon. The sufficient component cause model presents an opportunity to examine biological findings at low doses from an epidemiological lens9. In this model, exposures “sufficient” to cause an outcome of interest are presented in pie charts, such that when all slices of a pie chart are fulfilled, the outcome will occur. Multiple pie charts may exist for a single outcome, illustrating individual differences9. In 1988,Greenland and Poole adapted the sufficient component cause model to incorporate interaction with other exposures (such as genetics or lifestyle factors)10. They show that a range of biological processes are possible in a population, but that an epidemiologic study will only reveal the mean outcome from the population at large10,11. Using this construct, the multiple outcomes presented in radiobiological models to date can be explained in the context of epidemiologic studies. This presentation aims to demonstrate a causal framework that can integrate radiobiological and epidemiological models to date. It is intended as a conversation starter to spur future research.

C M Milder↗

Neutron star masses and radii from quiescent low-mass x-ray binaries

We perform a systematic analysis of neutron star radius constraints from five quiescent low-mass X-ray binaries and examine how they depend on measurements of their distances and amounts of intervening absorbing material, as well as their assumed atmospheric compositions. We construct and calibrate to published results a semi-analytic model of the neutron star atmosphere which approximates these effects for the predicted masses and radii. Starting from mass and radius probability distributions established from hydrogen-atmosphere spectral fits of quiescent sources, we apply this model to compute alternate sets of probability distributions. We perform Bayesian analyses to estimate neutron star mass-radius curves and equation of state (EOS) parameters that best-fit each set of distributions, assuming the existence of a known low-density neutron star crustal EOS, a simple model for the high-density EOS, causality, and the observation that the neutron star maximum mass exceeds 2 M {sub ☉}. We compute the posterior probabilities for each set of distance measurements and assumptions about absorption and composition. We find that, within the context of our assumptions and our parameterized EOS models, some absorption models are disfavored. We find that neutron stars composed of hadrons are favored relative to those with exotic matter with strong phase transitions. In addition, models in which all five stars have hydrogen atmospheres are found to be weakly disfavored. Our most likely models predict neutron star radii that are consistent with current experimental results concerning the nature of the nucleon-nucleon interaction near the nuclear saturation density.

79 ASTRONOMY AND ASTROPHYSICS↗