Inferring vortex induced vibrations of flexible cylinders using physics-informed neural networks
Not Available
Kharazmi, Ehsan↗
SEARCH · Engineering Papers
Results for “Network inference”
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.
Assessing Decadal Predictability in an Earth-System Model Using Explainable Neural Networks
We show that explainable neural networks can identify regions of oceanic variability that contribute predictability on decadal timescales in a fully coupled Earth-system model. The neural networks learn to use sea-surface temperature anomalies to predict future continental surface temperature anomalies. We then use a neural-network explainability method called layerwise relevance propagation to infer which oceanic patterns lead to accurate predictions made by the neural networks. In particular, regions within the North Atlantic Ocean and North Pacific Ocean lend the most predictability for surface temperature across continental North America. We apply the proposed methodology to decadal variability, although the concept is generalizable to other timescales of predictability. Furthermore, while our approach focuses on predictable patterns of internal variability within climate models, it should be generalizable to observational data as well. Our study contributes to the growing evidence that explainable neural networks are important tools for advancing geoscientific knowledge.
58 GEOSCIENCES↗
Coronal Heating, Spicules, and Solar-B
Falconer et al. investigated the heating of the quiet corona by measuring the increase of coronal luminosity with the amount of the magnetic flux in the underlying network at solar minimum when there were no active regions on the face of the Sun. The coronal luminosity was measured from Fe IX/X - Fe XII pairs of coronal images from SOHO/EIT, under the assumption that practically all of the coronal luminosity in these very quiet regions came from plasma in the temperature range 0.9 x 10(exp 6) K is less than or equal to T is less than or equal to 1.3 x 10(exp 6) K. The network magnetic flux content was measured from SOHO/MDI magnetograms. It was found that luminosity of the corona in these quiet regions increased roughly in proportion to the square root of the magnetic flux content of the network and roughly in proportion to the length of the perimeter of the network flux clumps. From 1) this result; 2) the observed occurrence of many fine-scale explosive events (e.g., spicules) at the edges of network flux clumps; and 3) a demonstration that it is energetically feasible for the heating of the corona in quiet regions to be driven by explosions of granule-sized sheared-core magnetic bipoles embedded in the edges of the network flux clumps, Falconer et al. infer that in quiet regions that are not influenced by active regions the corona is mainly heated by such magnetic activity in the edges of the network flux clumps. From their observational results together with their feasibility analysis, Falconer et al. predict that 1) At the edges of the network flux clumps there are many transient sheared core bipoles of the size and lifetime of granules and having transverse field strengths greater than approx. 100 G; 2) Approx. 30 of these bipoles are present per supergranule; and 3) Most spicules are produced by explosions of these bipoles. The photospheric vector magnetograms, chromospheric filtergrams, and EUV spectra from Solar-B are expected to have sufficient sensitivity, spatial resolution, and cadence to test these predictions. The Falconer et al. (2003) inferred mixed-polarity magnetic flux at the base of spicules is compatible with the observed magnetic structure of Ha macrospicules recently found by Yamuchi et al. (2003).
Moore, Ron↗
In-Situ and Remote-Sensing Data Fusion Using Machine Learning Techniques to Infer Urban and Fire Related Pollution Plumes
Airmass type characterization is key in understanding the relative contribution of various emission sources to atmospheric composition and air quality and can be useful in bottom-up model validation and emission inventories. However, classification of pollution plumes from space is often not trivial. Sub-orbital campaigns, such as SEAC4RS (Studies of Emissions, Atmospheric Composition, Clouds and Climate Coupling by Regional Surveys) give us a unique opportunity to study atmospheric composition in detail, by using a vast suite of in-situ instruments for the detection of trace gases and aerosols. These measurements allow identification of spatial and temporal atmospheric composition changes due to various pollution plumes resulting from urban, biogenic and smoke emissions. Nevertheless, to transfer the knowledge gathered from such campaigns into a global spatial and temporal context, there is a need to develop workflow that can be applicable to measurements from space. In this work we rely on sub-orbital in-situ and total column remote sensing measurements of various pollution plumes taken aboard the NASA DC-8 during 2013 SEAC4RS campaign, linking them through a neural-network (NN) algorithm to allow inference of pollution plume types by input of columnar aerosol and trace-gas measurements. In particular, we use the 4STAR (Spectrometer for Sky-Scanning, Sun-Tracking Atmospheric Research) airborne measurements of wavelength dependent aerosol optical depth (AOD), particle size proxies, O3, NO2 and water vapor to classify different pollution plumes. Our method relies on assigning a-priori ground-truth labeling to the various plumes, which include urban pollution, different fire types (i.e. forest and agriculture) and fire stage (i.e. fresh and aged) using cluster analysis of aerosol and trace-gases in-situ and auxiliary (e.g. trajectory) data and the training of a NN scheme to fit the best prediction parameters using 4STAR measurements as input. We explore our misclassification rates as related to our ground-truth labels, and with multi-layered pollution plume cases. The next step in our analysis is to optimize parameter selection for a scheme that can be applied to space-borne aerosol and trace-gas observation platforms such as OMI, and future geostationary satellites such as TEMPO and GEO-CAPE.
Neural-network↗
A Full-Stack Exploration of Language-Based Parallelism in Fortran 2023
This poster explores native parallel features in Fortran 2023 through the lens of supporting applications with libraries, compilers, and parallel runtimes. The language revision informally named Fortran 2008 introduced parallelism in the form of Single Program Multiple Data (SPMD) execution with two broad feature sets: (1) loop-level parallelism via do concurrent and (2) a Partitioned Global Address Space (PGAS) comprised of distributed “coarray” data structures. Fortran’s native parallelism has demonstrated high performance [1] and reduced the burden of inserting what sometimes amounts to more directives than code. Several compilers support both feature sets, typically by translating do concurrent into serial do loops annotated by parallel directives and by translating SPMD/PGAS features into direct calls to a communication library. Our research focuses primarily on two questions: (1) can the compiler’s parallel runtime library be developed in the language being compiled (Fortran) and (2) can we define an interface to the runtime that liberates compilers from being hardwired to one runtime and vice versa. We are answering these questions by developing the Parallel Runtime Interface for Fortran (PRIF) [2] and the Co-Array Fortran Framework of Efficient Interfaces to Network Environments (Caffeine) [3]. Caffeine is initially targeting adoption by LLVM Flang, a new open-source Fortran compiler developed by a broad community in industry, academia, and government labs. We are also exploring the use of these features in Inference-Engine, a deep learning library designed to facilitate neural network training and inference for high-performance computing applications written in modern Fortran.
Rasmussen, Katherine↗
A deep learning approach to fast analysis of collective Thomson scattering spectra
Fast analysis of collective Thomson scattering ion acoustic wave features using a deep convolutional neural network model is presented. The network was trained from spectra to predict the plasma parameters, including ion velocities, population fractions, and ion and electron temperatures. A fully kinetic particle-in-cell simulation was used to model a laboratory astrophysics experiment and simulate a diagnostic image of the ion acoustic wave feature. Network predictions were compared with Bayesian inference of the plasma model parameters for both the simulated and experimentally measured images. Both approaches were fairly accurate predicting the simulated image and the network predictions matched a good portion of the Bayesian results for the experimentally measured image. The Bayesian approach is more robust to noise and motivates future work to train deep learning models with realistic noise. The advantage of the deep learning model is making thousands of predictions in a few hundred milliseconds, compared to a few seconds to minutes per prediction for the optimization and Bayesian approaches presented here. The results demonstrate promising capabilities of deep learning models to analyze Thomson data orders of magnitude faster than conventional methods when using the neural network for standalone analysis. If more rigorous analysis is needed, neural network predictions can be used to quickly initialize other optimization methods and increase chances of success. This is especially useful when the dataset becomes very large or highly dimensional and manually refining initial conditions for the entire dataset are no longer tractable.
70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
Examining Rail Transportation Route of Crude Oil in the United States Using Crowdsourced Social Media Data
Safety issues associated with transporting crude oil by rail have been a concern since the boom of the U.S. domestic shale oil production in 2012. During the last decade, over 300 crude-oil-by-rail incidents have occurred in the United States. Some of them have caused adverse consequences including fire and hazardous materials leakage. However, only limited information on crude-on-rail routes and their associated risks is available to the public. To this end, this study proposed an unconventional way to reconstruct crude-on-rail routes using geotagged photos harvested from the Flickr website. The proposed method linked the geotagged photos of crude oil trains posted online with national railway networks to identify potential railway segments that those crude oil trains were traveling on. Here, a shortest path-based method was applied to infer the complete crude-on-rail routes, by utilizing the confirmed railway segments as well as their directional information. Validation of the inferred routes was performed using a public map and official crude oil incident data. The results suggested that the inferred routes based on geotagged photos had high coverage, with approximately 96% of the documented crude oil incidents aligned with the reconstructed crude-on-rail network. The inferred crude oil train routes were found to pass through several metropolitan areas of high population density, who were exposed to potential risk. These findings could improve situational awareness for policy makers and transportation planners. In addition, with the inferred routes, this study has established a good foundation for future crude oil train risk-analyses along the rail route.
42 ENGINEERING↗
Lunar velocity structure and compositional and thermal inferences
Seismic data from Apollo Passive Seismic Network stations are analyzed to determine the velocity structure in the lunar crust and mantle. Since the publication of earlier results additional data has become available from the S-IVB impacts of the Apollo 16 and 17 missions and the LM ascent stage impact of the Apollo 17 mission. Data from these artificial impacts now cover a distance range of 9 to 1700 km. Travel times, amplitudes, and theoretical seismograms are used to derive a velocity structure for the outer 150 km of the moon. Preliminary results are given from the analysis of the complete set of seismograms.
Toksoz, M. N.↗
Improved Characterization of Soil Organic Matter by Integrating FT-ICR MS, Liquid Chromatography Tandem Mass Spectrometry, and Molecular Networking: A Case Study of Root Litter Decay under Drought Conditions
Understanding of how soil organic matter (SOM) chemistry is altered in a changing climate has advanced considerably; however, most SOM components remain unidentified, impeding the ability to characterize a major fraction of organic matter and predict what types of molecules, and from which sources, will persist in soil. Here we present a novel approach to better characterize SOM extracts by integrating information from three types of analyses, and we deploy this method to characterize decaying root-detritus soil microcosms subjected to either drought or normal conditions. To observe broad differences in composition, we employed direct infusion Fourier-transform ion cyclotron resonance mass spectrometry (DI-FT-ICR MS). We complemented this with liquid chromatography tandem mass spectrometry (LC-MS/MS) to identify components by library matching. Since libraries contain only a small fraction of SOM components, we also used fragment spectral cosine similarity scores to relate unknowns and library matches through molecular networks. This integrated approach allowed us to corroborate DI-FT-ICR MS molecular formulas using library matches, which included fungal metabolites and related polyphenolic compounds. We also inferred structures of unknowns from molecular networks and improved LC-MS/MS annotation rates from ~5 to 35% by considering DI-FT-ICR MS molecular formula assignments. Under drought conditions, we found greater relative amounts of lignin-like vs condensed aromatic polyphenol formulas and lower average nominal oxidation state of carbon, suggesting reduced decomposition of SOM and/or microbes under stress. Our integrated approach provides a framework for enhanced annotation of SOM components that is more comprehensive than performing individual data analyses in parallel.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Nonvolatile Electrochemical Random‐Access Memory under Short Circuit
Abstract Electrochemical random‐access memory (ECRAM) is a recently developed and highly promising analog resistive memory element for in‐memory computing. One longstanding challenge of ECRAM is attaining retention time beyond a few hours. This short retention has precluded ECRAM from being considered for inference classification in deep neural networks, which is likely the largest opportunity for in‐memory computing. In this work, an ECRAM cell with orders of magnitude longer retention than previously achieved is developed, and which is anticipated to exceed ten years at 85 °C. This study hypothesizes that the origin of this exceptional retention is phase separation, which enables the formation of multiple effectively equilibrium resistance states. This work highlights the promises and opportunities to use phase separation to yield ECRAM cells with exceptionally long, and potentially permanent, retention times.
Kim, Diana S.↗
SIMILE enables alignment of tandem mass spectra with statistical significance
Abstract Interrelating small molecules according to their aligned fragmentation spectra is central to tandem mass spectrometry-based untargeted metabolomics. Current alignment algorithms do not provide statistical significance and compounds that have multiple delocalized structural differences and therefore often fail to have their fragment ions aligned. Here we align fragmentation spectra with both statistical significance and allowance for multiple chemical differences using Significant Interrelation of MS/MS Ions via Laplacian Embedding (SIMILE). SIMILE yields spectral alignment inferred structural connections in molecular networks that are not found with cosine-based scoring algorithms. In addition, it is now possible to rank spectral alignments based on p-values in the exploration of structural relationships between compounds and enhance the chemical connectivity that can be obtained with molecular networking.
59 BASIC BIOLOGICAL SCIENCES↗
Detecting operons in bacterial genomes via visual representation learning
Contiguous genes in prokaryotes are often arranged into operons. Detecting operons plays a critical role in inferring gene functionality and regulatory networks. Human experts annotate operons by visually inspecting gene neighborhoods across pileups of related genomes. These visual representations capture the inter-genic distance, strand direction, gene size, functional relatedness, and gene neighborhood conservation, which are the most prominent operon features mentioned in the literature. By studying these features, an expert can then decide whether a genomic region is part of an operon. We propose a deep learning based method named Operon Hunter that uses visual representations of genomic fragments to make operon predictions. Using transfer learning and data augmentation techniques facilitates leveraging the powerful neural networks trained on image datasets by re-training them on a more limited dataset of extensively validated operons. Our method outperforms the previously reported state-of-the-art tools, especially when it comes to predicting full operons and their boundaries accurately. Furthermore, our approach makes it possible to visually identify the features influencing the network’s decisions to be subsequently cross-checked by human experts.
59 BASIC BIOLOGICAL SCIENCES↗
Interface learning in fluid dynamics: Statistical inference of closures within micro–macro-coupling models
Many complex multiphysics systems in fluid dynamics involve using solvers with varied levels of approximations in different regions of the computational domain to resolve multiple spatiotemporal scales present in the flow. The accuracy of the solution is governed by how the information is exchanged between these solvers at the interface and several methods have been devised for such coupling problems. In this article, we construct a data-driven model by spatially coupling a microscale lattice Boltzmann method (LBM) solver and macroscale finite difference method (FDM) solver for reaction-diffusion systems. The coupling between the micro-macro solvers has one to many mapping at the interface leading to the interface closure problem, and we propose a statistical inference method based on neural networks to learn this closure relation. The performance of the proposed framework in a bifidelity setting partitioned between the FDM and LBM domain shows its promise for complex systems where analytical relations between micro-macro solvers are not available.
42 ENGINEERING↗
Variational deep learning of equilibrium transition path ensembles
Here, we present a time-dependent variational method to learn the mechanisms of equilibrium reactive processes and efficiently evaluate their rates within a transition path ensemble. This approach builds off of the variational path sampling methodology by approximating the time-dependent commitment probability within a neural network ansatz. The reaction mechanisms inferred through this approach are elucidated by a novel decomposition of the rate in terms of the components of a stochastic path action conditioned on a transition. This decomposition affords an ability to resolve the typical contribution of each reactive mode and their couplings to the rare event. The associated rate evaluation is variational and systematically improvable through the development of a cumulant expansion. We demonstrate this method in both over- and under-damped stochastic equations of motion, in low-dimensional model systems, and in the isomerization of a solvated alanine dipeptide. In all examples, we find that we can obtain quantitatively accurate estimates of the rates of the reactive events with minimal trajectory statistics and gain unique insights into transitions through the analysis of their commitment probability.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Three-dimensional reconstruction of laser-direct-drive inertial confinement fusion hot-spot plasma from x-ray diagnostics on the OMEGA laser facility (invited)
A deep-learning convolutional neural network (CNN) is used to infer, from x-ray images along multiple lines of sight, the low-mode shape of the hot-spot emission of deuterium–tritium (DT) laser-direct-drive cryogenic implosions on OMEGA. The motivation of this approach is to develop a physics-informed 3-D reconstruction technique that can be performed within minutes to facilitate the use of the results to inform changes to the initial target and laser conditions for the subsequent implosion. The CNN is trained on a 3D radiation-hydrodynamic simulation database to relate 2D x-ray images to 3D emissivity at stagnation. The CNN accounts for the lack of an absolute spatial reference and the different bands of photon energies in the x-ray images. While previous works studied the effect of mode-1 asymmetries on implosion performance using nuclear diagnostics, this work focuses on the effect of mode 2 inferred from x-ray diagnostics on implosion performance. A current analysis of 19 DT cryogenic implosions indicates there is an upper limit of ~20% reduction in the neutron yield caused by an ℓ = 2 amplitude for ℓ 2 /ℓ 0 ≤ 0.32. Here, these conclusions are supported by 2D simulations.
70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
Distributed Inference with Sparse and Quantized Communication
Here, we consider the problem of distributed inference where agents in a network observe a stream of private signals generated by an unknown state, and aim to uniquely identify this state from a finite set of hypotheses. We focus on scenarios where communication between agents is costly, and takes place over channels with finite bandwidth. To reduce the frequency of communication, we develop a novel event-triggered distributed learning rule that is based on the principle of diffusing low beliefs on each false hypothesis. Building on this principle, we design a trigger condition under which an agent broadcasts only those components of its belief vector that have adequate innovation, to only those neighbors that require such information. We prove that our rule guarantees convergence to the true state exponentially fast almost surely despite sparse communication, and that it has the potential to significantly reduce information flow from uninformative agents to informative agents. Next, to deal with finite-precision communication channels, we propose a distributed learning rule that leverages the idea of adaptive quantization. We show that by sequentially refining the range of the quantizers, every agent can learn the truth exponentially fast almost surely, while using just 1 bit to encode its belief on each hypothesis. For both our proposed algorithms, we rigorously characterize the trade-offs between communication-efficiency and the learning rate.
42 ENGINEERING↗
Search for low mass dark matter in DarkSide-50: the bayesian network approach
Abstract We present a novel approach for the search of dark matter in the DarkSide-50 experiment, relying on Bayesian Networks. This method incorporates the detector response model into the likelihood function, explicitly maintaining the connection with the quantity of interest. No assumptions about the linearity of the problem or the shape of the probability distribution functions are required, and there is no need to morph signal and background spectra as a function of nuisance parameters. By expressing the problem in terms of Bayesian Networks, we have developed an inference algorithm based on a Markov Chain Monte Carlo to calculate the posterior probability. A clever description of the detector response model in terms of parametric matrices allows us to study the impact of systematic variations of any parameter on the final results. Our approach not only provides the desired information on the parameter of interest, but also potential constraints on the response model. Our results are consistent with recent published analyses and further refine the parameters of the detector response model.
72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
Learning and tuning fuzzy logic controllers through reinforcements
This paper presents a new method for learning and tuning a fuzzy logic controller based on reinforcements from a dynamic system. In particular, our generalized approximate reasoning-based intelligent control (GARIC) architecture (1) learns and tunes a fuzzy logic controller even when only weak reinforcement, such as a binary failure signal, is available; (2) introduces a new conjunction operator in computing the rule strengths of fuzzy control rules; (3) introduces a new localized mean of maximum (LMOM) method in combining the conclusions of several firing control rules; and (4) learns to produce real-valued control actions. Learning is achieved by integrating fuzzy inference into a feedforward neural network, which can then adaptively improve performance by using gradient descent methods. We extend the AHC algorithm of Barto et al. (1983) to include the prior control knowledge of human operators. The GARIC architecture is applied to a cart-pole balancing system and demonstrates significant improvements in terms of the speed of learning and robustness to changes in the dynamic system's parameters over previous schemes for cart-pole balancing.
Berenji, Hamid R.↗