Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Graph 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 433 records · Page 24
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

A Visual Comparison of Silent Error Propagation

High-performance computing (HPC) systems play a critical role in facilitating scientific discoveries. Their scale and complexity (e.g., the number of computational units and software stack) continue to grow as new systems are expected to process increasingly more data and reduce computing time. However, with more processing elements, the probability that these systems will experience a random bit-flip error that corrupts a program's output also increases, which is often recognized as silent data corruption. Analyzing the resiliency of HPC applications in extreme-scale computing to silent data corruption is crucial but difficult. An HPC application often contains a large number of computation units that need to be tested, and error propagation caused by error corruption is complex and difficult to interpret. Here, to accommodate this challenge, we propose an interactive visualization system that helps HPC researchers understand the resiliency of HPC applications and compare their error propagation. Our system models an application's error propagation to study a program's resiliency by constructing and visualizing its fault tolerance boundary. Coordinating with multiple interactive designs, our system enables domain experts to efficiently explore the complicated spatial and temporal correlation between error propagations. At the end, the system integrated a nonmonotonic error propagation analysis with an adjustable graph propagation visualization to help domain experts examine the details of error propagation and answer such questions as why an error is mitigated or amplified by program execution.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Using Likwid and Byfl to Benchmark Hardware Performance

This paper outlines a benchmarking study conducted during my internship at LANL, focusing on CPU (Computer Processing Unit) and program performance assessment. The primary goal was to gather memory access data using three methods across five polybench kernels The data gathered would then be used to compare and contrast to one another and calculate operational intensity for performance comparisons. Benchmarking tools like Byfl and Likwid were employed, with Byfl offering hardware-independent data through LLVM compiler communication and Likwid directly interacting with computer hardware. The study considered various benchmarking factors, including optimization levels, Big O notation ((n)), CPU diversity and specific kernel equations. Big O notation was utilized to simplify code complexity, with detailed breakdwons of operations and memory components for each polybench application. Specific O(n) equations enabled nuanced kernel compariosns, facilitating the identification of performance variations. CPU efficiency assessments were conducted using Likwid tests on two CPUs. The central focus on code optimization aimed at achieving higher speeds and reduced memory usage through streamlined code. Future work propsoes creating a roofline model, synthesizing benchmarking data into a comprehensive data graph to assist in optimizing code and improving hardware performance. The potential impact on the laboratory or national mission was underscored, emphasizing the importance of optimizing applications and hardware to conserve resources and accelerate program execution. The specific relevance to LANL’s operations in math-intensive fields such as Nuclear Fission, Space Exploration, and Nanotechnology highlights the necessity of efficient benchmarking for resource conservation and proram speed. Overall, this study contributes to the understanding of CPU and program performance, providing insights for future optimization efforts in a laboratory setting

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Graph Neural Networks for low-energy event classification & reconstruction in IceCube

IceCube, a cubic-kilometer array of optical sensors built to detect atmospheric and astrophysical neutrinos between 1 GeV and 1 PeV, is deployed 1.45 km to 2.45 km below the surface of the ice sheet at the South Pole. The classification and reconstruction of events from the in-ice detectors play a central role in the analysis of data from IceCube. Reconstructing and classifying events is a challenge due to the irregular detector geometry, inhomogeneous scattering and absorption of light in the ice and, below 100 GeV, the relatively low number of signal photons produced per event. To address this challenge, it is possible to represent IceCube events as point cloud graphs and use a Graph Neural Network (GNN) as the classification and reconstruction method. The GNN is capable of distinguishing neutrino events from cosmic-ray backgrounds, classifying different neutrino event types, and reconstructing the deposited energy, direction and interaction vertex. Based on simulation, we provide a comparison in the 1 GeV–100 GeV energy range to the current state-of-the-art maximum likelihood techniques used in current IceCube analyses, including the effects of known systematic uncertainties. For neutrino event classification, the GNN increases the signal efficiency by 18% at a fixed background rate, compared to current IceCube methods. Alternatively, the GNN offers a reduction of the background (i.e. false positive) rate by over a factor 8 (to below half a percent) at a fixed signal efficiency. For the reconstruction of energy, direction, and interaction vertex, the resolution improves by an average of 13%–20% compared to current maximum likelihood techniques in the energy range of 1 GeV–30 GeV. The GNN, when run on a GPU, is capable of processing IceCube events at a rate nearly double of the median IceCube trigger rate of 2.7 kHz, which opens the possibility of using low energy neutrinos in online searches for transient events.

47 OTHER INSTRUMENTATION↗
DOE OSTI2021

A Theoretical Approach for Reliability Within Information Supply Chains With Cycles and Negations

Complex networks of information processing systems, or information supply chains, present challenges for performance analysis. Here, we establish a mathematical setting, in which a process within an information supply chain can be analyzed in terms of the functionality of the system’s components. Principles of this methodology are rigorously defended and induce a model for determining the reliability for the various products in these networks. Our model does not limit us from having cycles in the network, as long as the cycles do not contain negation. It is shown that our approach to reliability resolves the nonuniqueness caused by cycles in a probabilistic Boolean network. An iterative algorithm is given to find the reliability values of the model, using a process that can be fully automated. This automated method of discerning reliability is beneficial for systems managers. As a systems manager considers systems modification, such as the replacement of owned and maintained hardware systems with cloud computing resources, the need for comparative analysis of system reliability is paramount. The model is extended to handle conditional knowledge about the network, allowing one to make predictions of weaknesses in the system. Finally, to illustrate the model’s flexibility over different forms, it is demonstrated on a system of components and subcomponents.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Trajectory Optimization via Unsupervised Probabilistic Learning On Manifolds

This report investigates the use of unsupervised probabilistic learning techniques for the analysis of hypersonic trajectories. The algorithm first extracts the intrinsic structure in the data via a diffusion map approach. Using the diffusion coordinates on the graph of training samples, the probabilistic framework augments the original data with samples that are statistically consistent with the original set. The augmented samples are then used to construct conditional statistics that are ultimately assembled in a path-planing algorithm. In this framework the controls are determined stage by stage during the flight to adapt to changing mission objectives in real-time. A 3DOF model was employed to generate optimal hypersonic trajectories that comprise the training datasets. The diffusion map algorithm identfied that data resides on manifolds of much lower dimensionality compared to the high-dimensional state space that describes each trajectory. In addition to the path-planing worflow we also propose an algorithm that utilizes the diffusion map coordinates along the manifold to label and possibly remove outlier samples from the training data. This algorithm can be used to both identify edge cases for further analysis as well as to remove them from the training set to create a more robust set of samples to be used for the path-planing process.

42 ENGINEERING↗
DOE OSTI2022

Exploring the Use of Novel Spatial Accelerators in Scientific Applications

Driven by the need to find alternative accelerators which can viably replace GPUs in next-generation Supercomputing systems, this paper proposes a methodology to enable agile application/hardware co-design. The application-first methodology provides the ability to come up with design of accelerators while working with real-world workloads, available accelerators, and system software. The iterative design process targets a set of kernels in a workload for performance estimates that can prune the design space for later phases of detailed architectural evaluations. To this effect, in this paper, a novel data-parallel device model is introduced that simulates the latency of performance-sensitive operations in an accelerator including data transfers and kernel computation using multi-core CPUs. The use of off-the-shelf simulators, such as pre-RTL simulator Aladdin or multiple tools available for exploring the design of deep neural network accelerators (e.g., Timeloop) is demonstrated for evaluation of various accelerator designs using applications with realistic inputs. Examples of multiple device configurations that are instantiable in a system are explored to evaluate the performance benefit of deploying novel accelerators. The proposed device is integrated with a programming model and system software to potentially explore the impacts of high-level programming languages/compilers and low-level effects such as task scheduling on multiple accelerators. We analyze our methodology for a set of applications that represent high-performance computing (HPC) and graph analytics. The applications include a computational chemistry kernel realized using tensor contractions, triangle counting, GraphSAGE and Breadth-first Search. These applications include kernels such as dense matrix-dense matrix multiplication, sparse matrix-spare matrix multiplication, and sparse matrix-dense vector multiplication. Our results indicate potential performance benefits and insights for system design by including accelerators that realize these kernels along-side general purpose accelerators.

AI, codesign, Accelerated Computing, Modeling and ↗
DOE OSTI2024

SmartFuse: Reconfigurable Smart Switches to Accelerate Fused Collectives in HPC Applications

Communication switches have sometimes been augmented to process collectives (e.g., the IBM BlueGene project and the Mellanox SHArP switch). In this work, we find that there is a great acceleration opportunity through the further augmentation of switches to accelerate more complex functions that combine communication with computation. We consider three types of such functions. The first is fully-fused collectives built by fusing multiple existing collectives like Allreduce with Alltoall. The second is semi-fused collectives built by combining a collective with another computation. The third we refer to as higher-order collectives built by combining multiple computations and communications, such as to perform matrix-matrix multiply (PGEMM). In this work, we propose a framework called SmartFuse to accelerate fused collective functions. The core of SmartFuse is a reconfigurable smart switch to support these operations. The semi/fully fused collectives are implemented with a CGRAlike architecture, while higher-order collectives are implemented with a more specialized computational unit that can also schedule communication. Supporting our framework is software to evaluate and translate relevant parts of the input program, compile them into a control data flow graph, and then map this graph to the switch hardware. The proposed framework, once deployed, has the strong potential to accelerate existing HPC applications transparently by encapsulation within an MPI implementation. Experimental results show that this approach improves the performance of the PGEMM kernel, MINIFE, and AMG by, on average, 94%, 15%, and 13%, respectively.

Haghi, Pouya↗
DOE OSTI2025

Role of the chiral anomaly in polarized deeply inelastic scattering. III. Wess-Zumino-Witten contributions and chiral Ward identities for finite quark mass

We extend our prior results on the worldline computation of the axial vector-vector-vector (AVV) triangle anomaly in polarized deeply inelastic scattering (DIS) to the finite mass case by computing in addition the pseudoscalar-vector-vector (PVV) triangle graph. For the well-studied QED case, we show explicitly how the off-forward AVV pole exactly cancels an identical PVV pole. We then demonstrate the dramatic difference in QCD due to the chiral condensate, which qualitatively modifies anomalous Ward identities. As in the massless case, the anomaly pole in QCD is canceled by the dynamics of a primordial isosinglet pseudoscalar η ¯ -meson, whose Wess-Zumino-Witten coupling to the topological charge density shifts the pole to the physical η ′ mass, with the finite quark mass contribution differing by O ( 10 % ) from the Witten-Veneziano formula. We obtain a compact analytic expression for the finite mass corrections to Shore and Veneziano’s result that the proton’s net quark helicity Δ Σ ∝ χ QCD ′ | m = 0 ( 0 ) , the forward slope of the topological susceptibility in the chiral limit, and show they are of the order of a few percentages. Our prior prediction that the polarized DIS structure function g 1 is quenched by sphaleronlike topological transitions at small x is unaffected by quark mass effects. Our results illustrate how worldline computations of anomalous processes, in synergy with lattice computations and nonet chiral perturbation theory, can uncover novel nonperturbative features of QCD at the Electron-Ion Collider. Published by the American Physical Society 2025

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2025

A change language for ontologies and knowledge graphs

Ontologies and knowledge graphs (KGs) are general-purpose computable representations of some domain, such as human anatomy, and are frequently a crucial part of modern information systems. Most of these structures change over time, incorporating new knowledge or information that was previously missing. Managing these changes is a challenge, both in terms of communicating changes to users and providing mechanisms to make it easier for multiple stakeholders to contribute. To fill that need, we have created KGCL, the Knowledge Graph Change Language (https://github.com/INCATools/kgcl), a standard data model for describing changes to KGs and ontologies at a high level, and an accompanying human-readable Controlled Natural Language (CNL). This language serves two purposes: a curator can use it to request desired changes, and it can also be used to describe changes that have already happened, corresponding to the concepts of “apply patch” and “diff” commonly used for managing changes in text documents and computer programs. Another key feature of KGCL is that descriptions are at a high enough level to be useful and understood by a variety of stakeholders—e.g. ontology edits can be specified by commands like “add synonym ‘arm’ to ‘forelimb’” or “move ‘Parkinson disease’ under ‘neurodegenerative disease’.” We have also built a suite of tools for managing ontology changes. These include an automated agent that integrates with and monitors GitHub ontology repositories and applies any requested changes and a new component in the BioPortal ontology resource that allows users to make change requests directly from within the BioPortal user interface. Overall, the KGCL data model, its CNL, and associated tooling allow for easier management and processing of changes associated with the development of ontologies and KGs.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗
DOE OSTI2021

Dark moments for the Standard Model?

If dark matter (DM) interacts with the Standard Model (SM) via the kinetic mixing (KM) portal, it necessitates the existence of portal matter (PM) particles which carry both dark and SM quantum numbers that will appear in vacuum polarization-like loop graphs. In addition to the familiar ~ eϵQ strength, QED-like interaction for the dark photon (DP), in some setups different loop graphs of these PM states can also induce other coupling structures for the SM fermions that may come to dominate in at least some regions of parameter space regions and which can take the form of ‘dark’ moments, e.g., magnetic dipole-type interactions in the IR, associated with a large mass scale, Λ. In this paper, motivated by a simple toy model, we perform a phenomenological investigation of a possible loop-induced dark magnetic dipole moment for SM fermions, in particular, for the electron. We show that at the phenomenological level such a scenario can not only be made compatible with existing experimental constraints for a significant range of correlated values for Λ and the dark U(1) D gauge coupling, g D , but can also lead to quantitatively different signatures once the DP is discovered. In this setup, assuming complex scalar DM to satisfy CMB constraints, parameter space regions where the DP decays invisibly are found to be somewhat preferred if PM mass limits from direct searches at the LHC and our toy model setup are all taken seriously. High precision searches for, or measurements of, the e + e - → γ + DP process at Belle II are shown to provide some of the strongest future constraints on this scenario.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Network analysis of memristive device circuits: dynamics, stability and correlations

Abstract Networks with memristive devices are a potential basis for the next generation of computing devices. They are also an important model system for basic science, from modeling nanoscale conductivity to providing insight into the information-processing of neurons. The resistance in a memristive device depends on the history of the applied bias and thus displays a type of memory. The interplay of this memory with the dynamic properties of the network can give rise to new behavior, offering many fascinating theoretical challenges. But methods to analyze general memristive circuits are not well described in the literature. In this paper we develop a general circuit analysis for networks that combine memristive devices alongside resistors, capacitors and inductors and under various types of control. We derive equations of motion for the memory parameters of these circuits and describe the conditions for which a network should display properties characteristic of a resonator system. For the case of a purely memresistive network, we derive Lyapunov functions, which can be used to study the stability of the network dynamics. Surprisingly, analysis of the Lyapunov functions show that these circuits do not always have a stable equilibrium in the case of nonlinear resistance and window functions. The Lyapunov function allows us to study circuit invariances, wherein different circuits give rise to similar equations of motion, which manifest through a gauge freedom and node permutations. Finally, we identify the relation between the graph Laplacian and the operators governing the dynamics of memristor networks operators, and we use these tools to study the correlations between distant memristive devices through the effective resistance.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

COMPASS-FME Terrestrial Ecosystem Manipulation to Probe the Effects of Storm Treatments (TEMPEST) Experiment Level 1 Sensor Data v1-2

This is the version 1-2 Level 1 (L1) data release for COMPASS-FME environmental sensors located at our Terrestrial Ecosystem Manipulation to Probe the Effects of Storm Treatments (TEMPEST) experimental site. This manipulative, ecosystem-scale TEMPEST experiment addresses the potential for freshwater and estuarine-water disturbance events to alter tree function, species composition, and ecosystem processes in a deciduous coastal forest in MD, USA. The experiment uses a large-unit (2000 m2), un-replicated experimental design, with three 50 m × 40 m plots serving as control, freshwater, and estuarine-water treatments.L1 data are close to raw, but are units-transformed and have out-of-instrument-bounds and out-of-service flags added. Duplicates and missing data are removed but otherwise these data are not filtered, and have not been subject to any additional algorithmic or human QA/QC. Any scientific analyses of L1 data should be performed with care. **This dataset will be updated quarterly with new data for the duration of the project**This dataset includes:- An overall dataset README file that describes the current version, gives citation and contact information, etc.- Site- and year-specific folders, each holding up to 12 CSV (comma separated value) data files for each site and plot in that year.- Metadata files within each site-year folder provide full information on data units, expected ranges, contact information, detailed flood times, as well as a general description of the site.- Environmental sensor types that appear in the data files include weather (ClimaVUE50, CS, RM Young, and LI instruments in the graphs below); soil conditions (TEROS12); soil redox state (Redox); groundwater variables (AquaTROLL200 and AquaTROLL600); open water sondes (Exo); tree sap velocity (Sapflow); and system voltage and state (Datalogger). Data are normally logged every 15 minutes.Please see v1-2 TEMPEST L1 Sensor Package Quick Start.pdf for detailed information on data package structure, temporal coverage, and versioning.The TEMPEST flood events occurred on the following dates. They lasted for ~10 hours each day and delivered ~80,000 gallons to each plot; many data streams are available at 1 or 5 minute frequency during these periods.* Tests: Aug 25 (fresh plot) and Sep 9 (salt plot), 2021* TEMPEST 1: June 22, 2022* TEMPEST 2: June 6-7, 2023* TEMPEST 3: June 11-13, 2024

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2026

Data Summarization and Inference at Scale

This is the final report for the DOE ASCR grant SC-0022260, Data Summarization and Inference at Scale, PI: Alex Pothen, Purdue University. The goal of the project was to solve data-intensive and compute-intensive problems in the physical sciences, engineering, information science, data science, etc. by designing and implementing new algorithms that could work with a subset of the data. The four subgoals were: (a) The solution of problems where the data is too large to be stored in the memory of a computer. In this streaming model of computation, the data arrives as a stream of elements to the computer, each element is processed as it arrives, and a decision is made to discard the data or to store it; only a small subset of the data proportional to the size of the output solution is stored, and when all the data has been streamed, a solution to the problem is computed from the stored subset. (b) The use of machine learning methods to compute solutions to data-intensive problems. The use of GPUs is critical to obtain high performance on machine learning tasks, but their memory sizes are smaller relative to that of CPUs. For large-scale problems, the data is sampled many times, and small samples are used with repetition, for robustness, to compute solutions to inference tasks. This sampling reduces the memory required to solve the problem, but attention is needed to avoid slow convergence to the solutions, and reduced accuracy of inference. We propose submodular optimization, Large Language Models, and physics-informed neural networks to enable GPU computations here. (c) Modeling and visualization of high-dimensional data using interpretable features. Clinical proteomic data sets from immunology for the detection of cancer and other diseases are temporal and high-dimensional, and algorithms for visualizing these data sets using clinically interpretable features are lacking. We propose methods that compute distances based on the optimal transportation problem and graph edit distances to address this problem. We also propose the use of optimal transport-based distances, spatial statistics, and network structure to classify image data sets, We apply these algorithms to electron micrographs of the peripheral nervous system in the digestive tract. (d) The design of data-intensive algorithms on emerging architectures, specifically, noisy, intermediate-scale quantum (NISQ) devices. Quantum computers offer the possibility of exploring large solution spaces due to the principle of superposition, but current quantum computers are limited by few qubits, short coherence times due to noise, poor interconections among the qubits, etc. We propose the use of the divide and conquer paradigm to solve large-scale problems, wherein collections of small subproblems are solved on the quantum devices, and the solutions to the subproblems are integrated into a solution for the original problem on a classical computer.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Using mixed methods to construct and analyze a participatory agent-based model of a complex Zimbabwean agro-pastoral system

Complex social-ecological systems can be difficult to study and manage. Simulation models can facilitate exploration of system behavior under novel conditions, and participatory modeling can involve stakeholders in developing appropriate management processes. Participatory modeling already typically involves qualitative structural validation of models with stakeholders, but with increased data and more sophisticated models, quantitative behavioral validation may be possible as well. In this study, we created a novel agent-basedmodel applied to a specific context: Zimbabwean non-governmental organization the Muonde Trust has been collecting data on their agro-pastoral system for the last 35 years and had concerns about land-use planning and the effectiveness of management interventions in the face of climate change. We collaboratively created an agent-based model of their system using their data archive, qualitatively calibrating it to the observed behavior of the real system without tuning any parameters to match specific quantitative outputs. We then behaviorally validated the model using quantitative community-based data and conducted a sensitivity analysis to determine the relative impact of underlying parameter assumptions, Indigenous management interventions, and different rainfall variation scenarios. We found that our process resulted in a model which was successfully structurally validated and sufficiently realistic to be useful for Muonde researchers as a discussion tool. The model was inconsistently behaviorally validated, however, with some model variables matching field data better than others. We observed increased model system instability due to increasing variability in underlying drivers (rainfall), and also due to management interventions that broke feedbacks between the components of the system. Interventions that smoothed year-to-year variation rather than exaggerating it tended to improve sustainability. The Muonde trust has used the model to successfully advocate to local leaders for changes in land-use planning policy that will increase the sustainability of their system.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Causal discovery from data assisted by large language models

Knowledge-driven discovery of novel materials necessitates the development of causal models for property emergence. While in the classical physical paradigm, the causal relationships are deduced based on physical principles or via experiment, the rapid accumulation of observational data necessitates learning causal relationships between dissimilar aspects of material structure and functionalities based on observations. For this, it is essential to integrate experimental data with prior domain knowledge. Here, we demonstrate this approach by combining high-resolution scanning transmission electron microscopy data with insights derived from large language models (LLMs). By applying ChatGPT to domain-specific literature, such as arXiv papers on ferroelectrics, and combining the obtained information with data-driven causal discovery, we construct adjacency matrices for directed acyclic graphs that map the causal relationships between structural, chemical, and polarization degrees of freedom in Sm-doped BiFeO 3 . This approach enables us to hypothesize how synthesis conditions influence material properties and guides experimental validation. Furthermore, the ultimate objective of this work is to develop a unified framework that integrates LLM-driven literature analysis with data-driven discovery, facilitating the precise engineering of ferroelectric materials by establishing clear connections between synthesis conditions and their resulting material properties.

Causal inference↗
DOE OSTI2025

wa-hls4ml and lui-gnn: A benchmark and GNN-based surrogate model for hls4ml resource and latency estimation

As machine learning (ML) increasingly serves as a tool for addressing real-time challenges in scientific applications, the development of advanced tooling has significantly reduced the time required to iterate on various designs. These advancements have solved major obstacles, but also exposed new challenges. For example, processes that were not previously considered bottlenecks, such as model synthesis, are now becoming limiting factors in the rapid iteration of designs. To reduce these emerging constraints, multiple efforts are being launched toward designing an ML-based surrogate model that estimates resource usage of synthesized accelerator architectures. This model would reduce the design iteration time, especially when designing within a set of given hardware constraints. This approach shows considerable potential, but as it stands, the effort is early and would benefit from coordination and standardization to assist future work as it emerges. We introduce wa-hls4ml, a benchmark for ML accelerator resource and latency estimation, and its corresponding initial dataset of more than 100,000 fully connected neural networks, all synthesized using hls4ml and targeting Xilinx FPGAs. In addition to the resource utilization and latency data provided, the dataset includes generated artifacts and log files for many of the synthesized neural networks, in order to support future research in ML-based code generation. The benchmark evaluates the performance of resource and latency predictors against several common ML model architectures, primarily originating from scientific domains, as exemplar models, as well as the average performance across a subset of the dataset. We measure the performance of a given predictor model through multiple metrics, including $R^2$ score and SMAPE on regression tasks, as well as inference time to further characterize the estimator under test. Additionally, we introduce the latency/utilization inference graph neural network (lui-gnn), a surrogate model that uses a graph neural network to represent input architectures in the form of a directed graph. This graph representation allows for a diverse set of model architectures to all be effectively handled by a surrogate model. We present the architecture and performance of the model, as evaluated by the new proposed benchmark, including SMAPE, $R^2$ score, and inference times, and find that lui-gnn generally predicts latency and utilization for the 75\% quantile within several percent of the synthesized resources on the synthetic test dataset, indicating that this approach of estimating resource and latency via a surrogate models has promise and warrants further research.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Explainable Synthesizability Prediction of Inorganic Crystal Polymorphs Using Large Language Models

Abstract We evaluate the ability of machine learning to predict whether a hypothetical crystal structure can be synthesized and explain those predictions to scientists. Fine‐tuned large language models (LLMs) trained on a human‐readable text description of the target crystal structure perform comparably to previous bespoke convolutional graph neural network methods, but better prediction quality can be achieved by training a positive‐unlabeled learning model on a text‐embedding representation of the structure. An LLM‐based workflow can then be used to generate human‐readable explanations for the types of factors governing synthesizability, extract the underlying physical rules, and assess the veracity of those rules. These explanations can guide chemists in modifying or optimizing non‐synthesizable hypothetical structures to make them more feasible for materials design.

Kim, Seongmin [Department of Chemical and Biologic↗
DOE OSTI2025

Explainable Synthesizability Prediction of Inorganic Crystal Polymorphs Using Large Language Models

Abstract We evaluate the ability of machine learning to predict whether a hypothetical crystal structure can be synthesized and explain those predictions to scientists. Fine‐tuned large language models (LLMs) trained on a human‐readable text description of the target crystal structure perform comparably to previous bespoke convolutional graph neural network methods, but better prediction quality can be achieved by training a positive‐unlabeled learning model on a text‐embedding representation of the structure. An LLM‐based workflow can then be used to generate human‐readable explanations for the types of factors governing synthesizability, extract the underlying physical rules, and assess the veracity of those rules. These explanations can guide chemists in modifying or optimizing non‐synthesizable hypothetical structures to make them more feasible for materials design.

Kim, Seongmin [Department of Chemical and Biologic↗