Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “MIMIC 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 523 records · Page 29

Keratin Promotes Differentiation of Keratinocytes Seeded on Collagen/Keratin Hydrogels

Keratinocytes undergo a complex process of differentiation to form the stratified stratum corneum layer of the skin. In most biomimetic skin models, a 3D hydrogel fabricated out of collagen type I is used to mimic human skin. However, native skin also contains keratin, which makes up 90% of the epidermis and is produced by the keratinocytes present. We hypothesized that the addition of keratin (KTN) in our collagen hydrogel may aid in the process of keratinocyte differentiation compared to a pure collagen hydrogel. Keratinocytes were seeded on top of a 100% collagen or 50/50 C/KTN hydrogel cultured in either calcium-free (Ca-free) or calcium+ (Ca+) media. Our study demonstrates that the addition of keratin and calcium in the media increased lysosomal activity by measuring the glucocerebrosidase (GBA) activity and lysosomal distribution length, an indication of greater keratinocyte differentiation. We also found that the presence of KTN in the hydrogel also increased the expression of involucrin, a differentiation marker, compared to a pure collagen hydrogel. We demonstrate that a combination (i.e., containing both collagen and kerateine or “C/KTN”) hydrogel was able to increase keratinocyte differentiation compared to a pure collagen hydrogel, and the addition of calcium further increased the differentiation of keratinocytes. This multi-protein hydrogel shows promise in future models or treatments to increase keratinocyte differentiation into the stratum corneum.

Zuniga, Kameel (ORCID:0000000196674007)↗

Harnessing large language models’ zero-shot and few-shot learning capabilities for regulatory research

Abstract Large language models (LLMs) are sophisticated AI-driven models trained on vast sources of natural language data. They are adept at generating responses that closely mimic human conversational patterns. One of the most notable examples is OpenAI's ChatGPT, which has been extensively used across diverse sectors. Despite their flexibility, a significant challenge arises as most users must transmit their data to the servers of companies operating these models. Utilizing ChatGPT or similar models online may inadvertently expose sensitive information to the risk of data breaches. Therefore, implementing LLMs that are open source and smaller in scale within a secure local network becomes a crucial step for organizations where ensuring data privacy and protection has the highest priority, such as regulatory agencies. As a feasibility evaluation, we implemented a series of open-source LLMs within a regulatory agency’s local network and assessed their performance on specific tasks involving extracting relevant clinical pharmacology information from regulatory drug labels. Our research shows that some models work well in the context of few- or zero-shot learning, achieving performance comparable, or even better than, neural network models that needed thousands of training samples. One of the models was selected to address a real-world issue of finding intrinsic factors that affect drugs' clinical exposure without any training or fine-tuning. In a dataset of over 700 000 sentences, the model showed a 78.5% accuracy rate. Our work pointed to the possibility of implementing open-source LLMs within a secure local network and using these models to perform various natural language processing tasks when large numbers of training examples are unavailable.

Biochemistry & Molecular Biology↗

Machine learning models inaccurately predict current and future high-latitude C balances

The high-latitude carbon (C) cycle is a key feedback to the global climate system, yet because of system complexity and data limitations, there is currently disagreement over whether the region is a source or sink of C. Recent advances in big data analytics and computing power have popularized the use of machine learning (ML) algorithms to upscale site measurements of ecosystem processes, and in some cases forecast the response of these processes to climate change. Due to data limitations, however, ML model predictions of these processes are almost never validated with independent datasets. To better understand and characterize the limitations of these methods, we develop an approach to independently evaluate ML upscaling and forecasting. We mimic data-driven upscaling and forecasting efforts by applying ML algorithms to different subsets of regional process-model simulation gridcells, and then test ML performance using the remaining gridcells. In this study, we simulate C fluxes and environmental data across Alaska using ecosys, a process-rich terrestrial ecosystem model, and then apply boosted regression tree ML algorithms to training data configurations that mirror and expand upon existing AmeriFLUX eddy-covariance data availability. We first show that a ML model trained using ecosys outputs from currently-available Alaska AmeriFLUX sites incorrectly predicts that Alaska is presently a modeled net C source. Increased spatial coverage of the training dataset improves ML predictions, halving the bias when 240 modeled sites are used instead of 15. However, even this more accurate ML model incorrectly predicts Alaska C fluxes under 21st century climate change because of changes in atmospheric CO 2 , litter inputs, and vegetation composition that have impacts on C fluxes which cannot be inferred from the training data. Our results provide key insights to future C flux upscaling efforts and expose the potential for inaccurate ML upscaling and forecasting of high-latitude C cycle dynamics.

54 ENVIRONMENTAL SCIENCES↗

Two interaction surfaces between XPA and RPA organize the preincision complex in nucleotide excision repair

The xeroderma pigmentosum protein A (XPA) and replication protein A (RPA) proteins fulfill essential roles in the assembly of the preincision complex in the nucleotide excision repair (NER) pathway. We have previously characterized the two interaction sites, one between the XPA N-terminal (XPA-N) disordered domain and the RPA32 C-terminal domain (RPA32C), and the other with the XPA DNA binding domain (DBD) and the RPA70AB DBDs. Here, we show that XPA mutations that inhibit the physical interaction in either site reduce NER activity in biochemical and cellular systems. Combining mutations in the two sites leads to an additive inhibition of NER, implying that they fulfill distinct roles. Our data suggest a model in which the interaction between XPA-N and RPA32C is important for the initial association of XPA with NER complexes, while the interaction between XPA DBD and RPA70AB is needed for structural organization of the complex to license the dual incision reaction. Integrative structural models of complexes of XPA and RPA bound to single-stranded/double-stranded DNA (ss/dsDNA) junction substrates that mimic the NER bubble reveal key features of the architecture of XPA and RPA in the preincision complex. Most critical among these is that the shape of the NER bubble is far from colinear as depicted in current models, but rather the two strands of unwound DNA must assume a U-shape with the two ss/dsDNA junctions localized in close proximity. Our data suggest that the interaction between XPA and RPA70 is key for the organization of the NER preincision complex.

59 BASIC BIOLOGICAL SCIENCES↗

Analyses of the Boeing 737-MAX Accidents: Formal Models and Psychological Perspectives

Two fatal accidents involving the B737MAX resulted from the flight crews’ inability to overcome the effects of the Maneuvering Characteristics Augmentation System (MCAS). MCAS was designed to mimic the control column feel pressure and pitching behavior of the B737NG, which was the certification basis for the B737MAX. We briefly describe the potential role of formally modeling different perspectives during system design, and how such modeling can reveal gaps and conflicts between perspectives. We also discuss some of the relevant human factors issues involved in these accidents and how the aircraft’s behavior may have affected the pilots’ psychological states. Implications for automation design are considered.

5737MAX↗

PLATO hare-and-hounds exercise: asteroseismic model fitting of main-sequence solar-like pulsators

ABSTRACT Asteroseismology is a powerful tool to infer fundamental stellar properties. The use of these asteroseismic-inferred properties in a growing number of astrophysical contexts makes it vital to understand their accuracy. Consequently, we performed a hare-and-hounds exercise where the hares simulated data for six artificial main-sequence stars and the hounds inferred their properties based on different inference procedures. To mimic a pipeline such as that planned for the PLATO mission, all hounds used the same model grid. Some stars were simulated using the physics adopted in the grid, others a different one. The maximum relative differences found (in absolute value) between the inferred and true values of the mass, radius, and age were 4.32, 1.33, and 11.25 per cent, respectively. The largest systematic differences in radius and age were found for a star simulated assuming gravitational settling, not accounted for in the model grid, with biases of −0.88 per cent (radius) and 8.66 per cent (age). For the mass, the most significant bias (−3.16 per cent) was found for a star with a helium enrichment ratio outside the grid range. Moreover, an ∼7 per cent dispersion in age was found when adopting different prescriptions for the surface corrections or shifting the classical observations by ±1σ. The choice of the relative weight given to the classical and seismic constraints also impacted significantly the accuracy and precision of the results. Interestingly, only a few frequencies were required to achieve accurate results on the mass and radius. For the age the same was true when at least one l = 2 mode was considered.

Cunha, M. S. (ORCID:0000000182377343)↗

Deep Learning and Likelihood Approaches for Viral Phylogeography Converge on the Same Answers Whether the Inference Model Is Right or Wrong

Abstract Analysis of phylogenetic trees has become an essential tool in epidemiology. Likelihood-based methods fit models to phylogenies to draw inferences about the phylodynamics and history of viral transmission. However, these methods are often computationally expensive, which limits the complexity and realism of phylodynamic models and makes them ill-suited for informing policy decisions in real-time during rapidly developing outbreaks. Likelihood-free methods using deep learning are pushing the boundaries of inference beyond these constraints. In this paper, we extend, compare, and contrast a recently developed deep learning method for likelihood-free inference from trees. We trained multiple deep neural networks using phylogenies from simulated outbreaks that spread among 5 locations and found they achieve close to the same levels of accuracy as Bayesian inference under the true simulation model. We compared robustness to model misspecification of a trained neural network to that of a Bayesian method. We found that both models had comparable performance, converging on similar biases. We also implemented a method of uncertainty quantification called conformalized quantile regression that we demonstrate has similar patterns of sensitivity to model misspecification as Bayesian highest posterior density (HPD) and greatly overlap with HPDs, but have lower precision (more conservative). Finally, we trained and tested a neural network against phylogeographic data from a recent study of the SARS-Cov-2 pandemic in Europe and obtained similar estimates of region-specific epidemiological parameters and the location of the common ancestor in Europe. Along with being as accurate and robust as likelihood-based methods, our trained neural networks are on average over 3 orders of magnitude faster after training. Our results support the notion that neural networks can be trained with simulated data to accurately mimic the good and bad statistical properties of the likelihood functions of generative phylogenetic models.

Evolutionary Biology↗

Reactive chemical transport simulations of geologic carbon sequestration: Methods and applications

Chemical reaction simulations are considerably used to quantitatively assess the long-term geologic carbon sequestration (GCS), such as CO 2 sequestration capacity estimations, leakage pathway analyses, enhanced oil recovery (EOR) efficiency studies, and risk assessments of sealing formations (caprocks), wellbores, and overlying underground water resources. All these require a deep understanding of the CO 2 - associated chemical reactions. To ensure long-term, safe CO 2 sequestration in the intended formations, modeling is the only way to plausibly assess the CO 2 flow, reaction, and transport over thousands of years. This review summarizes the multiple methodologies for describing homogeneous and heterogeneous chemical reaction patterns and multiscale application examples, the recent progress and current status of chemical reaction simulations for GCS, and the impact of such simulations on geological CO2 sequestration performance. Technical gaps and future challenges are also discussed for further study. The trends and challenges of such studies include: (1) the combination of coupled chemical, mechanical, and transport processes with calibrated experiments and associated uncertainty/risk assessments; (2) enhancement of the ability to simulate detailed geophysical and geochemical equations to mimic in situ conditions; and (3) characterization of multiscale subsurface systems with detailed conceptual models and assignment of suitable boundary conditions for field-scale sequestration fields. One major gap remaining is the current lack of accurate (and scale-justified) kinetic and equilibrium chemical reaction parameters under reservoir conditions. Advanced models that couple chemical, mechanical, and transport processes with scale-justified parameters, from lab to field-scale experiments, are required for quantitative assessments of sequestration capacity and the long-term safety of GCS projects.

58 GEOSCIENCES↗

Complex Protostellar Chemistry

Two decades ago, our understanding chemistry in protostars was simple -- matter either fell into the central star or was trapped in planetary-scale objects. Some minor chemical changes might occur as the dust and gas fell inward, but such effects were overwhelmed by the much larger-scale processes that occurred even in bodies as small as asteroids. The chemistry that did occur in the nebula was relatively easy to model because the fall from the cold molecular cloud into the growing star was a one-way trip down a well-known temperature pressure gradient; the only free variable was time. However, just over 10 years ago it was suggested that some material could be processed in the inner nebula, flow outward, and become incorporated into comets. This outward flow was confirmed when the Stardust mission returned crystalline mineral fragments from Comet Wild 2 that must have been processed close to the Sun before they were incorporated into the comet. In this week's Science Express, Ciesla and Sandford demonstrate that even the outermost regions of the solar nebula can be a chemically active environment. Their finding could have consequences for the rest of the nebula. Our understanding of the chemistry in protostellar systems has made enormous progress over the last few decades, fueled by an increased awareness of the complex dynamics of these evolving energetic nebulae. We can no longer consider just the simple local environment to explain the composition of a planet, asteroid, or comet as was done in the past, but must now consider chemical processes that might take place within the nebula as a whole as well as the probability of transport and mixing the products of such reactions throughout the system. just as we now find it impossible to explain the complex chemistry of the terrestrial atmosphere without reference to detailed transport models that interconnect highly dissimilar chemical environments, so chemical models of protostars and of the solar nebula must eventually treat these environments as tightly coupled, interactive systems. The demonstration that the chemistry on the surfaces of outward-flowing, dynamically mixing icy grain surfaces both mimics the chemistry in cold cloud cores and strikes at the central assumption of the photochemical self-shielding model for oxygen isotopes in solar system solids only adds emphasis to this conclusion.

Nuth, Joseph A., III↗

A detailed chemical kinetic modeling and experimental investigation of the low- and high-temperature chemistry of n-butylcyclohexane

Chemical kinetic models of gasoline, jet, and diesel fuels and their mixtures with bioderived fuels are needed to assess fuel property effects on efficiency, emissions, and other performance metrics in internal combustion and gas turbine engines. As these real fuels have too many fuel components to be included in a chemical kinetic model, surrogate fuels containing fewer components are used to represent them. These surrogate fuels mimic the chemical classes or molecular structures contained in the real fuel. One of the important chemical classes in gasoline, jet, and diesel fuels comprises cyclohexanes. Cyclohexanes comprise about 30% or more by weight in diesel fuel. Also, Mueller et al proposed n-butylcyclohexane (nBCH) as a component in a nine-component surrogate palette to represent the ignition properties, distillation curve, density, and molecular structures of a diesel certification fuel. Herein, experimental measurements of the ignition delay times (IDTs) of nBCH in a shock tube and in a rapid compression machine are reported over a wide range of temperature, pressure, and equivalence ratio important for enabling the validation of a chemical kinetic model for nBCH for combustion in diesel engines. The range of conditions are temperatures of 630–1420 K, pressures of 10, 30, and 50 bar, and equivalence ratios of 0.3, 0.5, 1.0, and 2.0 in ‘air’. A detailed chemical model is developed for nBCH to simulate its ignition at both low and high-temperature conditions and at relevant elevated pressures. The experimentally measured IDTs are used to improve and validate the chemical kinetic model.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Development and Implementation of Dynamic Scripts to Execute Cycled GSI/WRF Forecasts

The Weather Research and Forecasting (WRF) numerical weather prediction (NWP) model and Gridpoint Statistical Interpolation (GSI) data assimilation (DA) are the operational systems that make up the North American Mesoscale (NAM) model and the NAM Data Assimilation System (NDAS) analysis used by National Weather Service forecasters. The Developmental Testbed Center (DTC) manages and distributes the code for the WRF and GSI, but it is up to individual researchers to link the systems together and write scripts to run the systems, which can take considerable time for those not familiar with the code. The objective of this project is to develop and disseminate a set of dynamic scripts that mimic the unique cycling configuration of the operational NAM to enable researchers to develop new modeling and data assimilation techniques that can be easily transferred to operations. The current version of the SPoRT GSI/WRF Scripts (v3.0.1) is compatible with WRF v3.3 and GSI v3.0.

Zavodsky, Bradley↗

Development and Implementation of Dynamic Scripts to Execute Cycled GSI/WRF Forecasts

The Weather Research and Forecasting (WRF) numerical weather prediction (NWP) model and Gridpoint Statistical Interpolation (GSI) data assimilation (DA) are the operational systems that make up the North American Mesoscale (NAM) model and the NAM Data Assimilation System (NDAS) analysis used by National Weather Service forecasters. The Developmental Testbed Center (DTC) manages and distributes the code for the WRF and GSI, but it is up to individual researchers to link the systems together and write scripts to run the systems, which can take considerable time for those not familiar with the code. The objective of this project is to develop and disseminate a set of dynamic scripts that mimic the unique cycling configuration of the operational NAM to enable researchers to develop new modeling and data assimilation techniques that can be easily transferred to operations. The current version of the SPoRT GSI/WRF Scripts (v3.0.1) is compatible with WRF v3.3 and GSI v3.0.

Zavodsky, Bradley↗

Modeling and Performance Simulation of the Mass Storage Network Environment

This paper describes the application of modeling and simulation in evaluating and predicting the performance of the mass storage network environment. Network traffic is generated to mimic the realistic pattern of file transfer, electronic mail, and web browsing. The behavior and performance of the mass storage network and a typical client-server Local Area Network (LAN) are investigated by modeling and simulation. Performance characteristics in throughput and delay demonstrate the important role of modeling and simulation in network engineering and capacity planning.

Kim, Chan M.↗

Direct Numerical Simulations of Turbulent Convection: Variable Gravity and Uniform Rotation - Part 1

Turbulent convection may have played a major role in determining the structure and evolution of the primordial solar nebula, but current, incomplete models of convection and turbulence give very different results and remain largely untested in the absence of detailed astronomical observations. Numerical simulations provide an 'experimental' database for comparison with these models, and, to this end, direct numerical simulations of turbulent convection were performed with modifications intended to mimic some of the unique physical features of thin accretion disks, such as the primordial solar nebula: internal heating, a gravitational acceleration that is linearly proportional to the distance from midplane of the nebula, and rapid rotation. Peclet numbers in the simulations are comparable to those in solar nebula models; Rossby numbers in the simulations are an order of magnitude larger than those in solar nebula models because of the unrealistically high Prandtl and low, Reynolds numbers required to resolve all scales of the convective flow. We find that, despite the loss of buoyancy at midplane, turbulent motions easily penetrate the midplane region with little loss of intensity, providing efficient transport of heat and turbulent kinetic energy throughout the interior. A simple mixing length model modified to include rotation is found to give convective heat fluxes for the interior flow in rough agreement with the numerical simulations. We discuss these preliminary results with regard to assumptions about heating distributions and convective heat fluxes made in standard solar nebula models. More definitive comparisons with solar nebula modelling will become possible when more realistic effects of shear, density, stratification, and compression are included.

Cabot, W.↗

The effect of an offset polar cap dipolar magnetic field on the modeling of the vela pulsar's γ-ray light curves

We performed geometric pulsar light curve modeling using static, retarded vacuum, and offset polar cap (PC) dipole B-fields (the latter is characterized by a parameter ϵ), in conjunction with standard two-pole caustic (TPC) and outer gap (OG) emission geometries. The offset-PC dipole B-field mimics deviations from the static dipole (which corresponds to ϵ=0). In addition to constant-emissivity geometric models, we also considered a slot gap (SG) E-field associated with the offset-PC dipole B-field and found that its inclusion leads to qualitatively different light curves. Solving the particle transport equation shows that the particle energy only becomes large enough to yield significant curvature radiation at large altitudes above the stellar surface, given this relatively low E-field. Therefore, particles do not always attain the radiation-reaction limit. Our overall optimal light curve fit is for the retarded vacuum dipole field and OG model, at an inclination angle α=78{sub −1}{sup +1{sup {sup ∘}}} and observer angle ζ=69{sub −1}{sup +2{sup {sup ∘}}}. For this B-field, the TPC model is statistically disfavored compared to the OG model. For the static dipole field, neither model is significantly preferred. We found that smaller values of ϵ are favored for the offset-PC dipole field when assuming constant emissivity, and larger ϵ values favored for variable emissivity, but not significantly so. When multiplying the SG E-field by a factor of 100, we found improved light curve fits, with α and ζ being closer to best fits from independent studies, as well as curvature radiation reaction at lower altitudes.

79 ASTRONOMY AND ASTROPHYSICS↗

Aqua-planet simulations of the formation of the South Atlantic convergence zone

The impact of Amazon Basin convection and cold fronts on the formation and maintenance of the South Atlantic convergence zone (SACZ) is studied using aqua-planet simulations with a general circulation model. In the model, a circular patch of warm sea-surface temperature (SST) is used to mimic the effect of the Amazon Basin on South American monsoon convection. The aqua-planet simulations were designed to study the effect of the strength and latitude of Amazon Basin convection on the formation of the SACZ. The simulations indicate that the strength of the SACZ increases as the Amazon convection intensifies and is moved away from the equator. Of the two controls studied here, the latitude of the Amazon convection exerts the strongest effect on the strength of the SACZ. An analysis of the synoptic-scale variability in the simulations shows the importance of frontal systems in the formation of the aqua-planet SACZ. Composite time series of frontal systems that occurred in the simulations show that a robust SACZ occurs when fronts penetrate into the subtropics and become stationary there as they cross eastward of the longitude of the Amazon Basin. Moisture convergence associated with these frontal systems produces rainfall not along the model SACZ region and along a large portion of the northern model Amazon Basin. Simulations in which the warm SST patch was too weak or too close to the equator did not produce frontal systems that extended into the tropics and became stationary, and did not form a SACZ. In the model, the SACZ forms as Amazon Basin convection strengthens and migrates far enough southward to allow frontal systems to penetrate into the tropics and stall over South America. This result is in agreement with observations that the SACZ tends to form after the onset of the monsoon season in the Amazon Basin.

South American Monsoon↗

Interplay between the charged Higgs and squark-gluino events at the LHC

In some extensions of the standard model with extended Higgs sectors, events from new particle production may pass the selection criteria for Higgs search in different channels at the LHC--14 TeV and mimic Higgs signals. This intriguing possibility is illustrated by PYTHIA based simulations using several representative points in the parameter space of the minimal supersymmetric standard model (MSSM) including a point in the minimal supergravity model consistent with the dark matter (DM) relic density data. Our simulations explore the interplay between the charged Higgs signal and typical squark-gluino events. We argue that the standard selections like the one based on the polarization properties of the {tau}'s from charged Higgs decay, though adequate for handling the SM backgrounds, may not be very efficient in the presence of SUSY backgrounds. We then propose an alternative search strategy based on pure kinematics which sufficiently controls both the SM and the MSSM backgrounds. For charged Higgs masses (H{sup {+-}}) in the deep decoupling regime (600 GeV(less-or-similar sign)m{sub H}{sup {+-}}(less-or-similar sign)800 GeV) this method works well and extends the LHC reach close to 800 GeV for an integrated luminosity of 30 fb{sup -1} with or without the SUSY background. For a lighter charged Higgs a judicious combination of the old selections and some of the cuts proposed by us may disentangle the Higgs signal from the squark-gluino backgrounds quite effectively.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Computationally efficient models for aqueous organic redox flow batteries

The rising usage of intermittent energy has garnered the need for large scale energy storage systems. Redox flow batteries (RFB) based energy storage system shows promising potential. Numerical simulations and machine learning approaches have been widely used to study RFB performance. The development of autonomous material discovery framework and digital twin of energy storage system usually needs to query cell performance through fast response models. In this study, two computationally efficient models are introduced: a physics-based analytical flow battery model (EZBattery), and a machine learning operator model (Deep Operator Network, denoted by DeepONet). Both models can provide cell performance near instantly, and prediction accuracy was systematically examined on an application of evaluating the performances of a 780 cm 2 aqueous organic redox flow battery (AORFB), using potential anolyte candidates in dihydroxyphenazine (DHP)-based family of organic materials. A validated computationally expansive 3-dimensional multi-physics finite element model by COMSOL was used as the ground truth and provided the training data set for the DeepONet. 1280 samples were generated with 10 properties to mimic the different possible anolyte candidates, and the cell performances were evaluated under 10 different combined operating conditions. The accuracy comparisons for the two computationally efficient models show that both models can provide comparable accuracy in predicting cell charging/discharging voltage curves. DeepONet can provide slightly higher overall accuracy than EZBattery with faster calculation speed, but highly relies on the training dataset. EZBattery does not need a training dataset and can provide interpretable physics-based explanations of the results, while being more flexible to adjust to adapt any different cell designs, flow battery architectures, and electrolyte materials.

Analytical model↗