Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Natural Language Processing”

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 361 records · Page 20

PDF Entity Annotation Tool (PEAT)

While different text mining approaches – including the use of Artificial Intelligence (AI) and other machine based methods - continue to expand at a rapid pace, the tools used by researchers to create the labeled datasets required for training, modeling, and evaluation remain rudimentary. Labeled datasets contain the target attributes the machine is going to learn; for example, training an algorithm to delineate between images of a car or truck would generally require a set of images with a quantitative description of the underlying features of each vehicle type. Development of labeled textual data that can be used to build natural language machine learning models for scientific literature is not currently integrated into existing manual workflows used by domain experts. Published literature is rich with important information, such as different types of embedded text, plots, and tables that can all be used as inputs to train ML/natural language processing (NLP) models, when extracted and prepared in machine readable formats. Currently, both normalized data extraction of use to domain experts and extraction to support development of ML/NLP models are labor intensive and cumbersome manual processes. Automatic extraction of data and information from formats such as PDFs that are optimized for layout and human readability, not machine readability. The PDF (Portable Document Format) Entity Annotation Tool (PEAT) was developed with the goal of allowing users to annotate publications within their current print format, while also allowing those annotations to be captured in a machine-readable format. One of the main issues with traditional annotation tools is that they require transforming the PDF into plain text to facilitate the annotation process. While doing so lessens the technical challenges of annotating data, the user loses all structure and provenance that was inherent in the underlying PDF. Also, textual data extraction from PDFs can be an error prone process. Challenges include identifying sequential blocks of text and a multitude of document formats (multiple columns, font encodings, etc.). As a result of these challenges, using existing tools for development of NLP/ML models directly from PDFs is difficult because the generated outputs are not interoperable. We created a system that allows annotations to be completed on the original PDF document structure, with no plain text extraction. The result is an application that allows for easier and more accurate annotations. In addition, by including a feature that grants the user the ability to easily create a schema, we have developed a system that can be used to annotate text for different domain-centric schemas of relevance to subject matter experts. Different knowledge domains require distinct schemas and annotation tags to support machine learning.

97 MATHEMATICS AND COMPUTING↗

Status on the Component Models Developed in the Modelica Framework: High-Temperature Steam Electrolysis Plant & Gas Turbine Power Plant

This report has been prepared as part of an effort to design and build a modeling and simulation (M&S) framework to assess the economic viability of a nuclear-renewable hybrid energy system (N-R HES). In order to facilitate dynamic M&S of such an integrated system, research groups in multiple national laboratories have been developing various subsystems as dynamic physics-based components using the Modelica programming language. In fiscal year (FY) 2015, Idaho National Laboratory (INL) performed a dynamic analysis of two region-specific N-R HES configurations, including the gas-to-liquid (natural gas to Fischer-Tropsch synthetic fuel) and brackish water reverse osmosis desalination plants as industrial processes. In FY 2016, INL has developed two additional subsystems in the Modelica framework: a high-temperature steam electrolysis (HTSE) plant and a gas turbine power plant (GTPP). HTSE has been proposed as a high priority industrial process to be integrated with a light water reactor (LWR) in an N-R HES. This integrated energy system would be capable of dynamically apportioning thermal and electrical energy (1) to provide responsive generation to the power grid and (2) to produce alternative industrial products (i.e., hydrogen and oxygen) without generating any greenhouse gases. A dynamic performance analysis of the LWR/HTSE integration case was carried out to evaluate the technical feasibility (load-following capability) and safety of such a system operating under highly variable conditions requiring flexible output. To support the dynamic analysis, the detailed dynamic model and control design of the HTSE process, which employs solid oxide electrolysis cells, have been developed to predict the process behavior over a large range of operating conditions. As first-generation N-R HES technology will be based on LWRs, which provide thermal energy at a relatively low temperature, complementary temperature-boosting technology was suggested for integration with the HTSE process that requires higher temperature input. Simulation results involving several case studies show that the suggested control scheme could maintain the controlled variables (including the steam utilization factor, cathode stream inlet composition, and temperatures of the process streams at various locations) within desired limits under various plant operating conditions. The results also indicate that the proposed HTSE plant could provide operational flexibility to participate in energy management at the utility scale by dynamically optimizing the use of excess plant capacity within an N-R HES. A natural-gas fired GTPP has been proposed as a secondary energy supply to be included in an N-R HES. This auxiliary generator could be used to cover rapid dynamics in grid demand that cannot be met by the remainder of the N-R HES. To evaluate the operability and controllability of the proposed process during transients between load (demand) levels, the dynamic model and control design were developed. Special attention was given to the design of feedback controllers to regulate the power frequency, and exhaust gas and turbine firing temperatures. Several case studies were performed to investigate the system responses to the major disturbance (power load demand) in such a control system. The simulation results show that the performance of the proposed control strategies was satisfactory under each test when the GTPP experienced high rapid variations in the load.

08 HYDROGEN↗

Status Report on the High-Temperature Steam Electrolysis Plant Model Developed in the Modelica Framework (FY17)

This report has been prepared as part of an effort to design and build a modeling and simulation (M&S) framework to assess the economic viability of a nuclear-renewable hybrid energy system (N-R HES). In order to facilitate dynamic M&S of such an integrated system, research groups in multiple national laboratories have been developing various subsystems as dynamic physics-based components using the Modelica programming language. In fiscal year 2015 (FY15), Idaho National Laboratory (INL) performed a dynamic analysis of two region-specific N-R HES configurations, including the gas-to-liquid (natural gas to Fischer-Tropsch synthetic fuel) and brackish water reverse osmosis desalination plants as industrial processes. In FY16, INL developed two additional subsystems in the Modelica framework: (1) a high-temperature steam electrolysis (HTSE) plant as a high priority industrial plant to be integrated with a light water reactor (LWR) within an N-R HES and (2) a gas turbine power plant as a secondary energy supply. In FY17, five new components (i.e., a feedwater pump, a multi-stage compression system, a sweep-gas turbine, flow control valves, and pressure control valves) have been incorporated into the HTSE system proposed in FY16, aiming to better realistically characterize all key components of concern. Special attention has been given to the controller settings based on process models (i.e., direct synthesis method), aiming to improve process dynamics and controllability. A dynamic performance analysis of the improved LWR/HTSE integration case was carried out to evaluate the technical feasibility (load-following capability) and safety of such a system operating under highly variable conditions requiring flexible output. The analysis (evaluated in terms of the step response) clearly shows that the FY17 model resulted in superior output responses with much smaller settling times and less oscillatory behavior in response to disturbances in the electric load than those observed with the FY16 model. Simulation results involving several case studies show that the suggested control scheme could maintain the controlled variables (including the steam utilization factor, cathode stream inlet composition, and temperatures and pressures of the process streams at various locations) within desired limits under various plant operating conditions. The results also indicate that the proposed HTSE plant could provide operational flexibility to participate in energy management at the utility scale by dynamically optimizing the use of excess plant capacity within an N-R HES.

08 HYDROGEN↗

Status Report on the Component Models Developed in the Modelica Framework: Reverse Osmosis Desalination Plant & Thermal Energy Storage

This report has been prepared as part of an effort to design and build a Modeling and Simulation (M&S) framework to assess the economic viability of a Nuclear-Renewable Hybrid Energy System (N-R HES). In order to facilitate dynamic M&S of such an integrated system, research groups in multiple national laboratories and universities have been developing various subsystems as dynamic physics-based components using the Modelica programming language. In Fiscal Years (FYs) 2015, Idaho National Laboratory (INL) performed a dynamic analysis of two region-specific N-R HES configurations, including the gas-to-liquid (natural gas to Fischer-Tropsch synthetic fuel) and brackish water Reverse Osmosis (RO) desalination plants as industrial processes. In FYs 2016–2017, INL developed two additional subsystems in the Modelica framework: (1) a high-temperature steam electrolysis plant as a high priority industrial plant to be integrated with a light water reactor within an N-R HES and (2) a gas turbine power plant as a secondary energy supply. In FY 2018, the RO desalination system model developed in FY 2015 has been updated such that the model is compatible with the most recent version of the ThermoPower library. Special attention has been given to the controller settings based on process models, aiming to improve process dynamics and controllability. A dynamic performance analysis of the updated RO desalination plant was carried out to evaluate the technical feasibility (load-following capability) of such a system operating under highly variable conditions requiring flexible output. Simulation results involving several case studies show that the suggested control scheme could maintain the controlled variables (including the variable electrical load and RO feed pressure) within desired limits under various plant operating conditions. The results also indicate that the proposed RO plant could provide operational flexibility to participate in energy management at the utility scale by dynamically optimizing the use of excess plant capacity within an N-R HES. For a small-scale energy storage system, a sensible Thermal Energy Storage (TES) model has been developed in the Modelica Framework in FY 2018.

25 ENERGY STORAGE↗

Bayesian Optimization of Catalysis with In-Context Learning

Large language models (LLMs) can perform accurate classification with zero or few examples through in-context learning (ICL), allowing the model to observe query-relevant examples at inference time and eliminating the need for additional weight updates to generalize beyond its original training data. We extend this capability to regression with uncertainty estimation using frozen LLMs (e.g., GPT-4o, Gemini), enabling Bayesian optimization (BO) in natural language without explicit model training or feature engineering. We apply this to materials discovery by representing materials as synthesis and testing procedures for use in natural language prompts. This Bayesian, design-first approach prioritizes optimization toward target material properties before detailed characterization, in contrast to conventional experimental workflows that often emphasize characterization of suboptimal materials. On benchmarks like aqueous solubility and oxidative coupling of methane (OCM), BO-ICL matches or outperforms Gaussian processes. In live experiments on the reverse water–gas shift (RWGS) reaction, BO-ICL identifies multimetallic catalysts that approach equilibrium CO yield within 6 and 10 iterations from a pool of 3,700 and 360,000 candidates, respectively. Our method redefines materials representation and accelerates discovery, with broad applications across catalysis, materials science, and AI.

Calibration↗

Implementing High-Performance Geometric Multigrid Solver with Naturally Grained Messages

Structured-grid linear solvers often require manually packing and unpacking of communication data to achieve high performance.Orchestrating this process efficiently is challenging, labor-intensive, and potentially error-prone.In this paper, we explore an alternative approach that communicates the data with naturally grained messagesizes without manual packing and unpacking. This approach is the distributed analogue of shared-memory programming, taking advantage of the global addressspace in PGAS languages to provide substantial programming ease. However, its performance may suffer from the large number of small messages. We investigate theruntime support required in the UPC ++ library for this naturally grained version to close the performance gap between the two approaches and attain comparable performance at scale using the High-Performance Geometric Multgrid (HPGMG-FV) benchmark as a driver.

97 MATHEMATICS AND COMPUTING↗

BuildingQA: A Benchmark for Natural Language Question Answering over Building Knowledge Graphs

Graph-based representations of building metadata using ontologies like Brick are vital for smart building applications, but querying them remains a challenge for practitioners. Knowledge Graph Question Answering (KGQA) systems, meant to retrieve answers from natural language questions, traditionally require large-scale training data, making them ill-suited for the specialized and data-scarce building domain. The advent of Large Language Models (LLMs) offers a paradigm shift, enabling zero-shot natural language querying without building/domain-specific training. Yet, there is no standardized benchmark for building-specific KGQA which can guide and validate research in this area. To address this gap, our work makes three primary contributions. First, we introduce the BuildingQA Benchmark Dataset, constructed through a multi-stage process of collecting practitioner data, augmenting it with LLMs for linguistic diversity, and curating a final set of 188 questions across 4 buildings. Second, we characterize the benchmark's complexity and ambiguity, introducing a novel method to quantify its "lexical gap" and providing a four-stage diagnostic framework for analyzing how systems fail. Third, we benchmark zero-shot LLM-powered KGQA systems to establish baseline performance and analyze their failure modes. Our evaluation reveals that top-performing systems achieve a maximum F1 score of only 0.38. This result does not indicate a failure of these powerful systems, but rather underscores the unique challenges posed by our benchmark. It demonstrates a critical performance gap, showing that current methods successful on general KGs struggle with the specific lexical and structural nuances of the building domain. BuildingQA1 thus provides the benchmark dataset and foundational analysis needed to drive the development of novel, domain-aware methods required to unlock the use of semantic data in buildings.

Mulayim, Ozan Baris↗

The Grammatical Universe and the Laws of Thermodynamics and Quantum Entanglement

The universal nilpotent computational rewrite system (UNCRS) is shown to formalize an irreversible process of evolution in conformity with the First, Second and Third Laws of Thermodynamics, in terms of a single algebraic creation operator (ikE+ip+jm) which delivers the whole quantum mechanical language apparatus, where k, i, j are quaternions units and E, p, m are energy, momentum and rest mass. This nilpotent evolution describes 'a dynamic zero totality universe' in terms of its fermion states (each of which, by Pauli exclusion, is unique and nonzero), where, together with their boson interactions, these define physics at the fundamental level. (The UNCRS implies that the inseparability of objects and fields in the quantum universe is based on the fact that the only valid mathematical representations are all automorphisms of the universe itself, and that this is the mathematical meaning of quantum entanglement. It thus appears that the nilpotent fermion states are in fact what is called the splitting field in Quantum Mechanics of the Galois group which leads to the roots of the corresponding algebraic equation, and concerns in this case the alternating group of even permutations which are themselves automorphisms). In the nilpotent evolutionary process: (i) the Quantum Carnot Engine (QCE) extended model of thermodynamic irreversibility, consisting of a single heat bath of an ensemble of Standard Model elementary particles, retains a small amount of quantum coherence / entanglement, so as to constitute new emergent fermion states of matter, and (ii) the metric (E{sup 2}-p{sup 2}m{sup 2}) = 0 ensures the First Law of the conservation of energy operates at each nilpotent stage, so that (iii) prior to each creation (and implied corresponding annihilation / conserve operation), E and m can be postulated to constitute dark energy and matter respectively. It says that the natural language form of the rewrite grammar of the evolution consists of the well known precepts of the Laws of Thermodynamics, formalized by the UNCRS regress, so as to become (as UNCRS rewrites already published at CASYS), firstly the Quantum Laws of Physics in the form of the generalized Dirac equation and later at higher stages of QCE ensemble complexity, the Laws of Life in the form of Nature's (DNA / RNA genetic) Code and then subsequently those of Intelligence and Consciousness (Nature's Rules).

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

NOUS: A Knowledge Graph Management System

Knowledge graphs represent information as entities and relationships between them. For tasks such as natural language question answering or automated analysis of text, a knowledge graph provides valuable context to establish the specific type of entities being discussed. It allow us to derive better context about newly arriving information and leads to intelligent reasoning capabilities. We address two primary needs: A) Automated construction of knowledge graphs is a technically challenging, expensive process; and B) The ability to synthesize new information by monitoring newly emerging knowledge is a transformational capability that does not exist in state of the art systems.

Source record↗

NOUS: A Knowledge Graph Management System

Knowledge graphs represent information as entities and relationships between them. For tasks such as natural language question answering or automated analysis of text, a knowledge graph provides valuable context to establish the specific type of entities being discussed. It allow us to derive better context about newly arriving information and leads to intelligent reasoning capabilities. We address two primary needs: A) Automated construction of knowledge graphs is a technically challenging, expensive process; and B) The ability to synthesize new information by monitoring newly emerging knowledge is a transformational capability that does not exist in state of the art systems.

Source record↗

Arctic shrub expansion generates regional variation in litter decomposition by altering litter quality and the decomposition environment

Abstract The expansion of deciduous shrubs into the graminoid‐dominated arctic tundra is expected to alter litter decomposition by changing litter quality and local abiotic and biotic conditions. However, it remains unclear how shrub expansion will affect litter decomposition at regional scales, where macroclimate is expected to be the dominant regulator of decomposition. To determine the relative influence of macroclimate and local controls on regional patterns of litter mass loss and nitrogen release, we conducted two hierarchical litter decomposition experiments across spatial scales. We decomposed leaf and root litter from a prominent graminoid ( Eriophorum vaginatum ) and three genera of deciduous shrubs ( Betula , Alnus and Salix ) for 1 year within replicated plots at five sites spanning a 160 km latitudinal gradient in northern Alaska. Using Eriophorum litter as a substrate, we found that macroclimate was the primary regulator of mass loss but had opposing effects on leaf and root litter. As summer temperature increased along the latitudinal gradient (11.9 to 13.9°C), leaf litter mass loss increased by 20% whereas root litter mass loss decreased by 33%. Leaf nitrogen release also increased with summer temperature. Conversely, root nitrogen release was controlled by the vegetation type of the decomposition environment. Using different shrub litters as substrates, we found that litter quality and its interaction with soil microclimate and macroclimate controlled decomposition. Overall, shrub root litters decomposed faster than Eriophorum root litter, losing 53% more mass and 190% more nitrogen across all sites and decomposition environments. For leaf litter, however, patterns varied by litter genus, with Salix losing more mass and Betula and Alnus losing less mass than Eriophorum . Our findings demonstrate that shrub expansion in the Arctic can regulate leaf and root litter decomposition at the regional scale through its effects on local controls, primarily litter quality. Ongoing increases in shrub cover are likely to accelerate the turnover of root litter carbon and nitrogen pools in tundra ecosystems. Therefore, including shrub‐related processes in Earth system models will improve our ability to predict regional‐scale litter decomposition and its effects on carbon and nutrient cycling in a warming Arctic. Read the free Plain Language Summary for this article on the Journal blog.

Vozzo, Justin T. [Department of Natural Resources ↗

Integrating System Dynamics and Bayesian Networks with Application to Counter-IED Scenarios

The practice of choosing a single modeling paradigm for predictive analysis can limit the scope and relevance of predictions and their utility to decision-making processes. Considering multiple modeling methods simultaneously may improve this situation, but a better solution provides a framework for directly integrating different, potentially complementary modeling paradigms to enable more comprehensive modeling and predictions, and thus better-informed decisions. The primary challenges of this kind of model integration are to bridge language and conceptual gaps between modeling paradigms, and to determine whether natural and useful linkages can be made in a formal mathematical manner. To address these challenges in the context of two specific modeling paradigms, we explore mathematical and computational options for linking System Dynamics (SD) and Bayesian network (BN) models and incorporating data into the integrated models. We demonstrate that integrated SD/BN models can naturally be described as either state space equations or Dynamic Bayes Nets, which enables the use of many existing computational methods for simulation and data integration. To demonstrate, we apply our model integration approach to techno-social models of insurgent-led attacks and security force counter-measures centered on improvised explosive devices.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

STARlight: A Monte Carlo simulation program for ultra-peripheral collisions of relativistic ions

© 2016 Elsevier B.V. Ultra-peripheral collisions (UPCs) have been a significant source of study at RHIC and the LHC. In these collisions, the two colliding nuclei interact electromagnetically, via two-photon or photonuclear interactions, but not hadronically; they effectively miss each other. Photonuclear interactions produce vector meson states or more general photonuclear final states, while two-photon interactions can produce lepton or meson pairs, or single mesons. In these interactions, the collision geometry plays a major role. We present a program, STARlight, that calculates the cross-sections for a variety of UPC final states and also creates, via Monte Carlo simulation, events for use in determining detector efficiency. Program summary Program Title: STARlight (v2.2) Program Files doi: http://dx.doi.org/10.17632/xjpf4rxtbj.1 Licensing provisions: GNU GPLv3 Programming Language: C++ External Routines: PYTHIA 8.2 and DPMJET 3.0 are needed for some final states. Nature of problem: The cross-section for ultra-peripheral collisions is obtained by integrating the photon fluxes in transverse impact parameter space, subject to the requirement (which is also impact parameter dependent) that the colliding nuclei do not interact hadronically. The program is a two step process. First, it calculates the cross-sections for the reaction of interest, as a function of W (photon–Pomeron or two-photon center of mass energy), Y (final state rapidity) and pT (final state transverse momentum). Second, STARlight generates Monte Carlo events which can be used to determine cross-sections within specific kinematic constraints or for studies of detector efficiencies. The second step includes the decay of any unstable particles produced in the reaction, with appropriate consideration of particle spins and parity. It outputs these events in ASCII format. Solution method: The program generates a two dimensional look-up table of the production cross-section as a function of final state rapidity and mass. The dimensions of the table are selectable, allowing the user to choose the desired accuracy. For certain final states, a second two-dimensional look-up table, giving the transverse momentum distribution, as a function of rapidity, is also used. With these look-up tables, the program generates final states. Particle decays and the final angular distributions are calculated for each event. Restrictions: The program is focused on ultra-relativistic collisions at Brookhaven's RHIC (Relativistic Heavy Ion Collider) and CERN's LHC (Large Hadron Collider), with final states that are visible in a central detector. At lower energies (i.e., at the CERN SPS), caution should be exercised because STARlight does not account for the longitudinal momentum transfer to the nucleus; this is larger at low beam energies. References: http://starlight.hepforge.org and references in this article.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Reaction mechanism and reaction coordinates from the viewpoint of energy flow

Reaction coordinates are of central importance for correct understanding of reaction dynamics in complex systems, but their counter-intuitive nature made it a daunting challenge to identify them. Starting from an energetic view of a reaction process as stochastic energy flows biased towards preferred channels, which we deemed the reaction coordinates, we developed a rigorous scheme for decomposing energy changes of a system, both potential and kinetic, into pairwise components. The pairwise energy flows between different coordinates provide a concrete statistical mechanical language for depicting reaction mechanisms. Application of this scheme to the C{sub 7eq} → C{sub 7ax} transition of the alanine dipeptide in vacuum revealed novel and intriguing mechanisms that eluded previous investigations of this well studied prototype system for biomolecular conformational dynamics. Using a cost function developed from the energy decomposition components by proper averaging over the transition path ensemble, we were able to identify signatures of the reaction coordinates of this system without requiring any input from human intuition.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

AI-based Cyber Event OSINT via Twitter Data

Open-Source Intelligence (OSINT) is largely regarded as a necessary component for cybersecurity intelligence gathering to secure network systems. With the advancement of artificial intelligence (AI) and increasing usage of social media, like Twitter, we have a unique opportunity to obtain and aggregate information from social media. In this study, we propose an AI-based scheme capable of automatically pulling information from Twitter, filtering out security-irrelevant tweets, performing natural language analysis to correlate the tweets about each cybersecurity event (e.g., a malware campaign), and validating the information. This scheme has many applications, such as providing a means for security operators to gain insight into ongoing events and helping them prioritize vulnerabilities to deal with. To give examples of the possible uses, we present three case studies demonstrating the event discovery and investigation processes.

Dale, Dakota↗

Open Source Intelligence for Cybersecurity Events via Twitter Data

Open-Source Intelligence (OSINT) is largely regarded as a necessary component for cybersecurity intelligence gathering to secure network systems. With the advancement of artificial intelligence (AI) and increasing usage of social media, like Twitter, we have a unique opportunity to obtain and aggregate information from social media. In this study, we propose an AI-based scheme capable of automatically pulling information from Twitter, filtering out security-irrelevant tweets, performing natural language analysis to correlate the tweets about each cybersecurity event (e.g., a malware campaign), and validating the information. This scheme has many applications, such as providing a means for security operators to gain insight into ongoing events and helping them prioritize vulnerabilities to deal with. To give examples of the possible uses, we present three case studies demonstrating the event discovery and investigation processes. We also examine the potential of OSINT for identifying the network protocols associated with specific events, which can aid in the mitigation procedures by informing operators if the vulnerability is exploitable given their system’s network configurations.

Dale, Dakota↗

The Design and Implementation of a Secure Datastore Based on Ethereum Smart Contract

In this paper, we present a secure datastore based on an Ethereum smart contract. Our research is guided by three research questions. First, we will explore to what extend a smart-contract-based datastore should resemble a traditional database system. Second, we will investigate how to store the data in a smart-contract-based datastore for maximum flexibility while minimizing the gas consumption. Third, we seek answers regarding whether or not a smart-contract-based datastore should incorporate complex processing such as data encryption and data analytic algorithms. The proposed smart-contract-based datastore aims to strike a good balance between several constraints: (1) smart contracts are publicly visible, which may create a confidentiality concern for the data stored in the datastore; (2) unlike traditional database systems, the Ethereum smart contract programming language (i.e., Solidity) offers very limited data structures for data management; (3) all operations that mutate the blockchain state would incur financial costs and the developers for smart contracts must make sure sufficient gas is provisioned for every smart contract call, and ideally, the gas consumption should be minimized. Our investigation shows that although it is essential for a smart-contract-based datastore to offer some basic data query functionality, it is impractical to offer query flexibility that resembles that of a traditional database system. Furthermore, we propose that data should be structured as tag-value pairs, where the tag serves as a non-unique key that describes the nature of the value. We also conclude that complex processing should not be allowed in the smart contract due to the financial burden and security concerns. The tag-based secure datastore designed this way also defines its applicative perimeter, i.e., only applications that align with our strategy would find the proposed datastore a good fit. Those that would rather incur higher financial cost for more data query flexibility and/or less user burden on data pre- and post-processing would find the proposed database too restrictive.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Automatic Code Generation for High-Performance Graph Algorithms

Graph problems are common across fields of scientific computing and social sciences. However, despite their importance, implementing graph algorithms effectively on modern computing systems is a challenging task that requires significant programming effort and generally results in customized implementations. Current computing and memory hierarchies are not architected for irregular computations resulting in challenges for graph algorithms to achieve high performance on those architectures. In this paper, we present GraphX, a novel compiler framework and DSL designed to simplify the development of efficient graph algorithms and achieve high performance on modern computing systems. GraphX consists of a DSL for efficient implementation of graph algorithms, various optimizations, such as support for sparse linear algebra and workspace transformations, optimized graph primitives, including semiring and masking, and a high-performance code generation engine. Using GraphX, users can implement graph algorithms using a semantically-rich language with graph-oriented operators. GraphX uses these semantics to automatically generate efficient code for target architectures, increasing performance and portability across architectures. The composable nature of GraphX makes it possible to extend the set of optimizations and architectures without modifying the source code. We demonstrate GraphX outperforms state-of-the-art graph libraries, such as LAGraph, up to $3.7 speedup in semiring operations, $2.19 speedup in an important sparse computational kernel, and $9.05 speedup in graph processing algorithms.

compiler, graph algorithms, semiring, masking, wor↗