Search NASA⌕ Search

SEARCH · Search NASA

Results for “Complex networks”

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 973 records · Page 54

Small Reservoirs Offer a New Perspective on Flood Reduction in Large Basins

The flood reduction potential of individual reservoirs within a large river network continuum remains poorly understood due to the complex interplay between reservoir characteristics and network properties. Here we investigate whether a collection of relatively small reservoirs can play a significant role in mediating downstream floods and assess how that role may be influenced by reservoir network properties compared to traditionally known reservoir characteristics. Our unique contribution was the simulation of downstream flood inundation maps alongside peak flows for each of the 81 major reservoirs (6 × 104 to 8 × 109 m3) across 15,000 river reach segments, integrated into a process-based hydrologic model covering a 415,000 km2 region in the Texas Gulf Coast, United States. The three key takeaways from our study are as follows. (a) Smaller reservoirs can substantially reduce downstream flooding, suggesting that inclusion of large reservoirs—the traditional approach in flood risk management studies—may present only a partial picture. (b) Flood reduction by smaller reservoirs is more effective upstream, although this phenomenon may be linked with aridity and overall water availability. (c) While reservoir size matters, it is not the primary factor determining its downstream flood reduction potential; the influence of network properties, such as catchment area, the count of upstream reservoirs, the cumulative maximum storage capacity of upstream reservoirs, and Stream Order (i.e., location), is equally and often more important. The broader impact of our findings goes beyond just floods, providing foundational insights for addressing emerging challenges such as aging dams and river connectivity.

Patel, Krutikkumar [University of Texas at Arlingt↗

Deep Generative Models that Solve PDEs: Distributed Computing for Training Large Data-Free Models

Recent progress in scientific machine learning (SciML) has opened up the possibility of training novel neural network architectures that solve complex partial differential equations (PDEs). Several (nearly data free) approaches have been recently reported that successfully solve PDEs, with examples including deep feed forward networks, generative networks, and deep encoder-decoder networks. However, practical adoption of these approaches is limited by the difficulty in training these models, especially to make predictions at large output resolutions (≥1024×1024). Here we report on a software framework for data parallel distributed deep learning that resolves the twin challenges of training these large SciML models - training in reasonable time as well as distributing the storage requirements. Our framework provides several out of the box functionality including (a) loss integrity independent of number of processes, (b) synchronized batch normalization, and (c) distributed higher-order optimization methods. We show excellent scalability of this framework on both cloud as well as HPC clusters, and report on the interplay between bandwidth, network topology and bare metal vs cloud. We deploy this approach to train generative models of sizes hitherto not possible, showing that neural PDE solvers can be viably trained for practical applications. We also demonstrate that distributed higher-order optimization methods are 2-3× faster than stochastic gradient-based methods and provide minimal convergence drift with higher batch-size.

PDEs↗

Technosocial Predictive Analytics for Illicit Nuclear Trafficking

Illicit nuclear trafficking networks are a national security threat. These networks can directly lead to nuclear proliferation, as state or non-state actors attempt to identify and acquire nuclear weapons-related expertise, technologies, components, and materials. The ability to characterize and anticipate the key nodes, transit routes, and exchange mechanisms associated with these networks is essential to influence, disrupt, interdict or destroy the function of the networks and their processes. The complexities inherent to the characterization and anticipation of illicit nuclear trafficking networks requires that a variety of modeling and knowledge technologies be jointly harnessed to construct an effective analytical and decision making workflow in which specific case studies can be built in reasonable time and with realistic effort. In this paper, we explore a solution to this challenge that integrates evidentiary and dynamic modeling with knowledge management and analytical gaming, and demonstrate its application to a geopolitical region at risk.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

Generating Massive Scale-free Networks: Novel Parallel Algorithms using the Preferential Attachment Model

Recently, there has been substantial interest in the study of various random networks as mathematical models of complex systems. As real-life complex systems grow larger, the ability to generate progressively large random networks becomes all the more important. This motivates the need for efficient parallel algorithms for generating such networks. Naïve parallelization of sequential algorithms for generating random networks is inefficient due to inherent dependencies among the edges and the possibility of creating duplicate (parallel) edges. In this article, we present message passing interface-based distributed memory parallel algorithms for generating random scale-free networks using the preferential-attachment model. Our algorithms are experimentally verified to scale very well to a large number of processing elements (PEs), providing near-linear speedups. The algorithms have been exercised with regard to scale and speed to generate scale-free networks with one trillion edges in 6 minutes using 1,000 PEs.

97 MATHEMATICS AND COMPUTING↗

The Third Dimension of Ferroelectric Domain Walls

Abstract Ferroelectric domain walls are quasi‐2D systems that show great promise for the development of nonvolatile memory, memristor technology, and electronic components with ultrasmall feature size. Electric fields, for example, can change the domain wall orientation relative to the spontaneous polarization and switch between resistive and conductive states, controlling the electrical current. Being embedded in a 3D material, however, the domain walls are not perfectly flat and can form networks, which leads to complex physical structures. In this work, the importance of the nanoscale structure for the emergent transport properties is demonstrated, studying electronic conduction in the 3D network of neutral and charged domain walls in ErMnO 3 . By combining tomographic microscopy techniques and finite element modeling, the contribution of domain walls within the bulk is clarified and the significance of curvature effects for the local conduction is shown down to the nanoscale. The findings provide insights into the propagation of electrical currents in domain wall networks, reveal additional degrees of freedom for their control, and provide quantitative guidelines for the design of domain‐wall‐based technology.

36 MATERIALS SCIENCE↗

Causally‐Informed Deep Learning to Improve Climate Models and Projections

Abstract Climate models are essential to understand and project climate change, yet long‐standing biases and uncertainties in their projections remain. This is largely associated with the representation of subgrid‐scale processes, particularly clouds and convection. Deep learning can learn these subgrid‐scale processes from computationally expensive storm‐resolving models while retaining many features at a fraction of computational cost. Yet, climate simulations with embedded neural network parameterizations are still challenging and highly depend on the deep learning solution. This is likely associated with spurious non‐physical correlations learned by the neural networks due to the complexity of the physical dynamical system. Here, we show that the combination of causality with deep learning helps removing spurious correlations and optimizing the neural network algorithm. To resolve this, we apply a causal discovery method to unveil causal drivers in the set of input predictors of atmospheric subgrid‐scale processes of a superparameterized climate model in which deep convection is explicitly resolved. The resulting causally‐informed neural networks are coupled to the climate model, hence, replacing the superparameterization and radiation scheme. We show that the climate simulations with causally‐informed neural network parameterizations retain many convection‐related properties and accurately generate the climate of the original high‐resolution climate model, while retaining similar generalization capabilities to unseen climates compared to the non‐causal approach. The combination of causal discovery and deep learning is a new and promising approach that leads to stable and more trustworthy climate simulations and paves the way toward more physically‐based causal deep learning approaches also in other scientific disciplines.

Meteorology & Atmospheric Sciences↗

Two supramolecular complexes based on polyoxometalates and Co-EDTA units via covalent connection or non-covalent interaction

Two new 3D network organic-inorganic hybrid supramolecular complexes ([Na{sub 6}(CoEDTA){sub 2}(H{sub 2}O){sub 13}]·(H{sub 2}SiW{sub 12}O{sub 40})·xH{sub 2}O)n (1) and [CoH{sub 4}EDTA(H{sub 2}O)]{sub 2}(SiW{sub 12}O{sub 40})·15H{sub 2}O (2) (H{sub 4}EDTA=Ethylenediamine tetraacetic acid) have been successfully synthesized by solution method, and characterized by infrared spectrum (IR), thermogravimetric-differential thermal analysis (TG-DTA), cyclic voltammetry (CV) and single{sup −}crystal X-ray diffraction (XRD). Both of the complexes are the supramolecules, but with different liking mode, they are two representative models of supramolecule. complex (1) is a 3D infinite network supramolecular coordination polymer with a rare multi-metal sturcture of sodium-cobalt-containing, which is mainly linked through coordinate-covalent bonds. While complex (2) is normal supramolecule, which linked by non-covalent interactions, such as H-bonding interaction, electrostatic interaction and van der waals force. Both of complex (1) and (2) exhibit good catalytic activities for catalytic oxidation of methanol, when the initial concentration of methanol is 3.0 g m{sup −3}, flow rate is 10 mL min{sup −1}, and the quality of catalyst is 0.2 g, for complex (1) and complex (2) the maximum elimination rates of methanol are 85% (150 °C) and 92% (120 °C), respectively. - Graphical abstract: Two new organic-inorganic hybrid supramolecular complexes based on Co-EDTA, and Keggin polyanions have been successfully synthesized with different pH value by solution method. They are attributed to two representative models of supramolecule. Complex(1) is an infinite coordination polymer with a rare multi-metal sturcture of sodium-cobalt-containing, which is mainly linked through covalent bonds. Complex (2) is a normal supramolecule, which linked by non-covalent interactions of H-bonding interaction, electrostatic interaction and van der waals force. - Highlights: • Two supramolecules are linked by covalent or non-covalent interactions. • They are attributed to two representative models of supramolecule. • A rare multi-metal infinite supramolecular coordination polymer was formed. • They exhibit good catalytic activities for catalytic oxidation of methanol.

36 MATERIALS SCIENCE↗

Physics-guided neural networks with engineering domain knowledge for hybrid process modeling

As neural networks are more frequently used to solve problems in science and engineering, the methods used to incorporate scientific knowledge into these networks are becoming increasingly complex. Here, this work breaks down these complicated techniques into a set of basic strategies which can easily be applied to diverse situations. Several novel neural networks are built using the categories laid out in this work. These networks are tested on simulated data from a continuous stirred tank reactor (CSTR) model to evaluate the advantages provided by each network. The three points demonstrated in this work are: (1) architectural hybrid models can speed up convergence and reduce the amount of data necessary to train a model; (2) adding a physics-guided loss function can improve model generalization and make models more physically consistent; (3) using physics-guided initialization and transfer learning improves accuracy and speeds up convergence, but can harm generalizability if used incorrectly.

42 ENGINEERING↗

Reconstruction of sediment supply and transport behind the Elwha River dams

The Elwha River watershed in Washington State’s Olympic National Park is predominantly a steep mountainous landscape where dominant geomorphic processes include landslides, debris flows, and gullying. The removal of the two massive hydropower dams along the Elwha River in 2013 marked the largest global dam removal in history. Over the century-long lifespan of the dams, approximately 21 million cubic meters of sediment was impounded behind them. In this study, we test different approaches to reconstructing sediment supply and delivery to the Elwha dams throughout their lifespan. Available observational data that aid in this approach include DEM, channel morphology, meteorology, and streamflow and sediment discharge. We analyze the performance of three numerical modeling constructs — Cases I, II, and III — in predicting historic reservoir sediment yields in the Elwha River watershed. In Case I, we use a sediment rating curve approach and improve upon a published sediment rating curve (Curran et al., 2009) for the Elwha River using a modified, calibrated version of the Wilcock & Crowe (2003) sediment transport equation. In Cases II and III we use a network model that we have developed which connects processes of sediment supply from hillslopes with sediment detachment and transport at stream links of a channel network. The model is driven by streamflow and sediment forcings which can be of varying complexity. In Case II, the network model is driven by stochastic streamflow forcings that require simple input data and parameterization. In Case III, the network model is driven by physically-based model outputs of streamflow from a physically-based distributed model, DHSVM. Using these three cases, we strive to answer the questions: Is using a watershed network model warranted in predicting dam sedimentation? If so, what level of complexity is needed in the driving streamflow and sediment forcings? Is a rating curve approach adequate and robust enough for predicting dam sedimentation, particularly with expected changes in environmental conditions? Cases I, II, and III in this report require further calibration before these questions can fully be answered, however are valuable “proof-of-concept” models. We show that our uncalibrated modeling results along with the insights we have developed provide a promising path in moving forward to answer these questions.

13 HYDRO ENERGY↗

Fast Parallel Tensor Times Same Vector for Hypergraphs

Hypergraphs are a popular paradigm to rep- resent complex real-world networks exhibiting multi-way relationships of varying sizes. Mining centrality in hyper- graphs via symmetric adjacency tensors has only recently become computationally feasible for large and complex datasets. To enable scalable computation of these and related hypergraph analytics, here we focus on the Sparse Symmetric Tensor Times Same Vector (S3TTVC) oper- ation. We introduce the Compound Compressed Sparse Symmetric (CCSS) format, an extension of the compact CSS format for hypergraphs of varying hyperedge sizes and present a shared-memory parallel algorithm to compute S3TTVC. We experimentally show S3TTVC computation using the CCSS format achieves better performance than the naive baseline, and is subsequently more performant for hypergraph H-eigenvector centrality.

Shivakumar, Shruti↗

An intra-node HPC network architecture with nanosecond-scale photonic switches

We propose a single-stage network architecture for intra-node connectivity that makes use of nanosecond-scale photonic switches. Although buffering at the switch points is of vital importance for complex multi-stage networks, this is not the case for smaller-scale single-stage networks where the end nodes are located only one hop apart. By limiting the buffering to the end points, the proposed architecture manages to minimize the required electro-optic and opto-electronic conversions, leading in this way to both low end-to-end latency and better energy efficiency. Combining these advantages with nanosecond-scale switching times can allow for high-throughput operation even for frequent switch reconfigurations. The performance of the proposed architecture is evaluated via discrete-event simulations for a wide range of synthetic-traffic cases. The simulation results show that high-throughput operation of ≥90% can be achieved even for small message sizes, i.e., 32 KB for all-to-all communication and 2 KB for uniform random traffic, at a data rate of 400 Gb/s and a switch reconfiguration time of ≤72 ns. Moreover, if 100-ns reconfiguration times are achievable as opposed to 150-ns, then for the all-to-all traffic case a 16% and 28% reduction in completion time can be achieved for message sizes of 8 KB and 1 KB, respectively. In a forthcoming era of optically interfaced processors and accelerators, nanosecond-scale photonic switches appear as a highly promising solution for keeping up with the intra-node bandwidth scaling due to their high-bandwidth, low-latency and fast-switching capabilities.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Nonideality in Silicone Network Formation via Solvent Swelling and 1 H Double-Quantum NMR

The versatile crosslinking chemistry of poly(dimethylsiloxane) (PDMS) based materials affords a large research space in which polymers with widely varying elastomeric properties may be synthesized. Parameters such as chain length, crosslink density, crosslink functionality, filler content, and chain chemistry can all be modified to produce materials with specific physical and mechanical properties. Commercial polysiloxane based, ‘silicone’ elastomers are generally intractable, which makes the precise characterization of their networks problematic. We report here the application of equilibrium solvent uptake analysis and 1 H double-quantum nuclear magnetic resonance ( 1 H DQ NMR) spectroscopy to determine the network topology of end-linked PDMS networks with non-ideal network topology. Despite their structural complexity, we can quantify both the classical and non-classical contributions to network structure using 1 H DQ NMR which are in reasonable agreement with solvent uptake data. These findings serve as the foundation for future investigations of even more complex commercial silicones using 1 H DQ NMR.

36 MATERIALS SCIENCE↗

Development of a microwave capacitive method for the spectroscopy of the complex permittivity

We describe a vector network analyzer-based method to study the electromagnetic properties of nanoscale dielectrics at microwave frequencies (1 MHz–40 GHz). The complex permittivity spectrum of a given dielectric can be determined by placing it in a capacitor accessed on its both electrodes by coplanar waveguides. However, inherent propagation delays along the signal paths together with frequency-dependent effective surface of the capacitor at microwave frequencies can lead to significant distortion in the measured permittivity, which in turn can give rise to artificial frequency variations of the complex permittivity. We detail a fully analytical rigorous correction sequence with neither recourse to extrinsic loss mechanisms nor to arbitrary parasitic signal paths. We illustrate our method on 3 emblematic dielectrics: ferroelectric morphotropic lead zirconate titanate, its paraelectric pyrochlore counterpart, and strontium titanate. Permittivity spectra taken at various points along the hysteresis loop help shedding light onto the nature of the different dielectric energy loss mechanisms. Thanks to the analytical character of our method, we can discuss routes to extend it to higher frequencies and we can identify unambiguously the sources of potential artifacts.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Adaptive fuzzy leader clustering of complex data sets in pattern recognition

A modular, unsupervised neural network architecture for clustering and classification of complex data sets is presented. The adaptive fuzzy leader clustering (AFLC) architecture is a hybrid neural-fuzzy system that learns on-line in a stable and efficient manner. The initial classification is performed in two stages: a simple competitive stage and a distance metric comparison stage. The cluster prototypes are then incrementally updated by relocating the centroid positions from fuzzy C-means system equations for the centroids and the membership values. The AFLC algorithm is applied to the Anderson Iris data and laser-luminescent fingerprint image data. It is concluded that the AFLC algorithm successfully classifies features extracted from real data, discrete or continuous.

Newton, Scott C.↗

Machine Learning for the Prediction of Local Asteroid Damages

Risk assessment studies of local asteroid hazards traditionally simulate the physics of meteors with engineering models tailored to analyze tens-of-millions of scenarios. However, these simplified approaches still need to solve time-dependent ODEs to model the entry process and the resulting ground damage. With a computational cost of O(0.01 CPU.s) per scenario, simulating these large numbers of potential entry conditions in risk assessment studies can take several days on local computers. To improve computational efficiency, we propose in this paper an orthogonal approach based on machine learning models to predict the size of damaged areas given a list of entry parameters. We train 5 machine learning methods and compare the predictions to the outputs of the PAIR model, first only with primitive entry condition variables, and then with more advanced features. We find that complex models like neural networks are well-suited to estimate blast hazards, while simpler linear models can accurately assess thermal damage. For both types of hazards, the radii of damaged areas can be predicted with around 10% average errors and a coefficient of determination (R2) of 0.99. The CPU time is decreased by a factor O(10 3 ) compared to the PAIR model, which enables the simulation of millions of scenarios in minutes, on a local computer. We then use the same machine learning approaches for a classification task where the models are trained to predict if an asteroid will produce a given level of damage. Results show that complex models like the gradient boosting classifier and the neural network can perform this task with 98% accuracy. Beyond surrogate models, we finally incorporate the machine learning algorithms to the state-of-the-art Shapley sensitivity analysis and present a ranking of the entry parameters based on their contributions to ground damages.

SMD↗

Oestrogen regulates SOX9 bioavailability by rapidly activating ERK1/2 and stabilising microtubules in a human testis-derived cell line

Highlights: • Brief oestrogen treatment causes the cytoplasmic retention of the testis factor SOX9. • The cytoplasmic retention of SOX9 by oestrogen is associated with the rapid stabilisation of microtubules. • ERK1/2 is required for the stabilisation of microtubules by activated oestrogen receptors. • Oestrogen can act on both a non-genomic and genomic level to impact gonad somatic cell fate. Nuclear SOX9 is essential for Sertoli cell differentiation and the development of a testis. Exposure of Sertoli cells to exogenous oestrogen causes cytoplasmic retention of SOX9, inhibiting testis development and promoting ovarian development. The cytoplasmic localisation of SOX9 requires a stabilised microtubule network and a key MAPK complex, ERK1/2, is responsive to oestrogen and known to affect the microtubule network. We hypothesised that oestrogen could stabilise microtubules through the activation of ERK1/2 to promote the cytoplasmic retention of SOX9. Treatment of human testis-derived NT2/D1 cells for 30 min with oestrogen rapidly activated ERK1/2, stabilised the microtubule network and increased cytoplasmic localisation of SOX9. The effects of oestrogen on SOX9 and tubulin were blocked by the ERK1/2 inhibitor U0126, demonstrating that ERK1/2 mediates the stabilisation of microtubules and cytoplasmic retention of SOX9 by oestrogen. Together, these data revealed a previously unknown mechanism for oestrogen in impacting MAPK signalling to block SOX9 bioavailability and the differentiation of Sertoli cells.

60 APPLIED LIFE SCIENCES↗

An overview of the gene regulatory network controlling trichome development in the model plant, Arabidopsis

Trichomes are specialized epidermal cells located on aerial parts of plants and are associated with a wide array of biological processes. Trichomes protect plants from adverse conditions including UV light and herbivore attack and are also an important source of a number of phytochemicals. The simple unicellular trichomes of Arabidopsis serve as an excellent model to study molecular mechanism of cell differentiation and pattern formation in plants. The emerging picture suggests that the developmental process is controlled by a transcriptional network involving three major groups of transcription factors (TFs): the R2R3 MYB, basic helix-loop-helix (bHLH), and WD40 repeat (WDR) protein. These regulatory proteins form a trimeric activator complex that positively regulates trichome development. The single repeat R3 MYBs act as negative regulators of trichome development. They compete with the R2R3 MYBs to bind the bHLH factor and form a repressor complex. In addition to activator–repressor mechanism, a depletion mechanism may operate in parallel during trichome development. In this mechanism, the bHLH factor traps the WDR protein which results in depletion of WDR protein in neighboring cells. Consequently, the cells with high levels of bHLH and WDR proteins are developed into trichomes. A group of C2H2 zinc finger TFs has also been implicated in trichome development. Phytohormones, including gibberellins and jasmonic acid, play significant roles in this developmental process. Recently, microRNAs have been shown to be involved in trichome development. Furthermore, it has been demonstrated that the activities of the key regulatory proteins involved in trichome development are controlled by the 26S/ubiquitin proteasome system (UPS), highlighting the complexity of the regulatory network controlling this developmental process. To complement several excellent recent relevant reviews, this review focuses on the transcriptional network and hormonal interplay controlling trichome development in Arabidopsis.

59 BASIC BIOLOGICAL SCIENCES↗

Generalizing Gillespie’s Direct Method to Enable Network-Free Simulations

Gillespie’s direct method for stochastic simulation of chemical kinetics is a staple of computational systems biology research. However, the algorithm requires explicit enumeration of all reactions and all chemical species that may arise in the system. In many cases, this is not feasible due to the combinatorial explosion of reactions and species in biological networks. Rule-based modeling frameworks provide a way to exactly represent networks containing such combinatorial complexity, and generalizations of Gillespie’s direct method have been developed as simulation engines for rule-based modeling languages. Here, we provide both a high-level description of the algorithms underlying the simulation engines, termed network-free simulation algorithms, and how they have been applied in systems biology research. We also define a generic rule-based modeling framework and describe a number of technical details required for adapting Gillespie’s direct method for network-free simulation. Lastly, we briefly discuss potential avenues for advancing network-free simulation and the role they continue to play in modeling dynamical systems in biology.

59 BASIC BIOLOGICAL SCIENCES↗