Search NASA⌕ Search

SEARCH · Search NASA

Results for “Structured”

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 109 records · Page 6
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Structural Phase Separation and Enhanced Superconductivity in La 1.875 Ba 0.125 CuO 4 Under Uniaxial Strain

Strain engineering has attracted significant attention in recent years due to its capability in tuning lattice and electronic structures of quantum materials. Using moderate uniaxial compressive strain, structural phase separation is induced in the low-temperature phase of x = 1/8 La 2-x Ba x CuO 4 (LBCO) single crystals. These structures are low temperature tetragonal (LTT), low temperature less orthorhombic (LTLO), and a plastically deformed nano-domain structure (PDNS), comprised of few-nanometer-sized orthorhombic domains within an amorphous matrix. These three structures exhibit distinct superconducting behaviors. The volume fraction of the LTT structure is suppressed with increasing strain, while its superconducting transition temperature increases and broadens. The LTLO structure exhibits a sharp superconducting transition above 32 K, which increases up to ≈ 36 K at maximum strain. The PDNS phase exhibits a very broad superconducting transition and persists even after removing the strain. This study illustrates the sensitivity of superconductivity to the structure of the LBCO sample near its stripe instability.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2025

Visualizing and analyzing 3D biomolecular structures using Mol* at RCSB.org: Influenza A H5N1 virus proteome case study

The easiest and often most useful way to work with experimentally determined or computationally predicted structures of biomolecules is by viewing their three-dimensional (3D) shapes using a molecular visualization tool. Mol* was collaboratively developed by RCSB Protein Data Bank (RCSB PDB, RCSB.org) and Protein Data Bank in Europe (PDBe, PDBe.org) as an open-source, web-based, 3D visualization software suite for examination and analyses of biostructures. It is capable of displaying atomic coordinates and related experimental data of biomolecular structures together with a variety of annotations, facilitating basic and applied research, training, education, and information dissemination. Across RCSB.org, the RCSB PDB research-focused web portal, Mol* has been implemented to support single-mouse-click atomic-level visualization of biomolecules (e.g., proteins, nucleic acids, carbohydrates) with bound cofactors, small-molecule ligands, ions, water molecules, or other macromolecules. RCSB.org Mol* can seamlessly display 3D structures from various sources, allowing structure interrogation, superimposition, and comparison. Using influenza A H5N1 virus as a topical case study of an important pathogen, we exemplify how Mol* has been embedded within various RCSB.org tools—allowing users to view polymer sequence and structure-based annotations integrated from trusted bioinformatics data resources, assess patterns and trends in groups of structures, and view structures of any size and compositional complexity. In addition to being linked to every experimentally determined biostructure and Computed Structure Model made available at RCSB.org, Standalone Mol* is freely available for visualizing any atomic-level or multi-scale biostructure at rcsb.org/3d-view.

3D biostructure↗
DOE OSTI2025

Functional Relevance of CASP16 Nucleic Acid Predictions as Evaluated by Structure Providers

ABSTRACT Accurate biomolecular structure prediction enables the prediction of mutational effects, the speculation of function based on predicted structural homology, the analysis of ligand binding modes, experimental model building, and many other applications. Such algorithms to predict essential functional and structural features remain out of reach for biomolecular complexes containing nucleic acids. Here, we report a quantitative and qualitative evaluation of nucleic acid structures for the CASP16 blind prediction challenge by 12 of the experimental groups who provided nucleic acid targets. Blind predictions accurately model secondary structure and some aspects of tertiary structure, including reasonable global folds for some complex RNAs; however, predictions often lack accuracy in the regions of highest functional importance. All models have inaccuracies in non‐canonical regions where, for example, the nucleic‐acid backbone bends, deviating from an A‐form helix geometry, or a base forms a non‐standard hydrogen bond (not a Watson‐Crick base pair). These bends and non‐canonical interactions are integral to forming functionally important regions such as RNA enzymatic active sites. Additionally, the modeling of conserved and functional interfaces between nucleic acids and ligands, proteins, or other nucleic acids remains poor. For some targets, the experimental structures may not represent the only structure the biomolecular complex occupies in solution or in its functional life cycle, posing a future challenge for the community.

Biochemistry & Molecular Biology↗
DOE OSTI2024

Thermal conductivity of 3D-printed block-copolymer-inspired structures

This study primarily focuses on examining the impact that geometric structure has on thermal conductivity of multi-phase constructs in different 3D-printed poly(lactic acid), PLA, samples. The investigated structures are inspired by morphologies formed by diblock copolymers: lamellae, hexagonally packed cylinders, and gyroid. This research also investigates how volume percentage and material combination influence the thermal conductivity of these structures. Further, the samples can be tailored to simulate various thermal management structures observed in practical applications, such as thermal interface materials in electronic devices. Thermal conductivity ratio is controlled using air, the least conductive material at 0.026 W/(m K), PLA at 0.136 W/(m K), and thermal paste at 5.11 W/(m K). Different models were tested against thermal conductivity measurements in order to capture the effect of material type (PLA-Air versus PLA-Thermal Paste), volume percentage, structure, and orientation. Simple, effective medium models were good predictions of thermal conductivity in lamellar structures, but it was necessary to develop models for conduction through cylindrical and gyroid structures. Finally, all results were normalized to find a universal model that is independent of structure and material. This approach provides a simple method to predict how to reduce or enhance transport properties and heat management capabilities of 3D printed objects.

36 MATERIALS SCIENCE↗
DOE OSTI2026

Effects of Composition and Oxidation States on the Structures of Chromium-Containing Sodium Silicate Glasses: Molecular Dynamics Simulations using Machine Learning Interatomic Potentials

Chromium represents a significant challenge for the vitrification of high-level nuclear waste into silicate and borosilicate glasses due to its low solubility and variable oxidation states, which can limit the waste loading due to promotion of crystallization or phase separation during processing. In this study, we modeled chromium containing silicate glasses using molecular dynamics simulations with three machine learning interatomic potentials (MLIPs), MACE, CHGNet, and PFP were employed, to gain insights on glass composition and oxidation states on the structures of these glasses. One of the goals is to evaluate their ability of these MLIPs to accurately represent the general structure of silicate glasses and chromium local environments as a function of chromium oxidation states. Density Functional Theory (DFT) based calculations and experimental data such as neutron structure factors were used to validate the structural models. It was found that the foundation models of all three MLIPs are able to reproduce general structural features of the sodium silicate glass structure consistent with experimental and DFT data, but only CHGNet and PFP can accurately capture the oxidation states and local environment of chromium: tetrahedral for Cr6+ and octahedral for Cr3+. Furthermore, we studied the effect of varying Cr3+/ Cr6+ (Cr3+/Crtotal) ratio and total chromium content using PFP. Our results show that Cr6+ enhances network polymerization by reducing non-bridging oxygens through Na? charge compensation required due to the formation of chromate (CrO42-) species, while Cr³? acts as a network modifier that disrupts connectivity. System size effects on the structural characteristics and chromium environments were also tested using the PFP potential. This work highlights the importance of careful validation on the precision, transferability, and potential of MLIPs for modeling glasses containing transition metal elements that can exist in multiple oxidation states. It is also encouraging to see the foundational models are all three MLFFs are able to reproduce the basic sodium silicate glass structures, while suggesting additional training or refining is needed to improve the description of more complex systems containing transition metals.

Puga, Christina L.↗
DOE OSTI2025

GMFOLD: Subgraph matching for high-throughput DNA-aptamer secondary structure classification and machine learning interpretability

Aptamers are oligonucleotide receptors that bind to their targets with high affinity. Here, we consider aptamers comprised of single-stranded DNA that undergo target-binding-induced conformational changes, giving rise to unique secondary and tertiary structures. Given a specific aptamer primary sequence, there are well-established computational tools (notably mfold) to predict the secondary structure via free energy minimization algorithms. While mfold generates secondary structures for individual sequences, there is a need for a high-throughput process whereby thousands of DNA structures can be predicted in real-time for use in an interactive setting, when combined with aptamer selections that generate candidate pools that are too large to be experimentally interrogated. We developed a new Python code for high-throughput aptamer secondary structure determination (GMfold). GMfold uses subgraph matching methods to group aptamer candidates by secondary structure similarities. We also improve an open-source code, SeqFold, to incorporate subgraph matching concepts. We represent each secondary structure as a lowest-energy bipartite subgraph matching of the DNA graph to itself. These new tools enable thousands of DNA sequences to be compared based on their secondary structures, using machine-learning algorithms. This process is advantageous when analyzing sequences that arise from aptamer selections via systematic evolution of ligands by exponential enrichment (SELEX). This work is a building block for future machine-learning-informed DNA-aptamer selection processes to identify aptamers with improved target affinity and selectivity and advance aptamer biosensors and therapeutics.

Aptamer↗
DOE OSTI2024

Insights into coordination and ligand trends of lanthanide complexes from the Cambridge Structural Database

Abstract Understanding lanthanide coordination chemistry can help develop new ligands for more efficient separation of lanthanides for critical materials needs. The Cambridge Structural Database (CSD) contains tens of thousands of single crystal structures of lanthanide complexes that can serve as a training ground for both fundamental chemical insights and future machine learning and generative artificial intelligence models. This work aims to understand the currently available structures of lanthanide complexes in CSD by analyzing the coordination shell, donor types, and ligand types, from the perspective of rare-earth element (REE) separations. We obtain four sets of lanthanide complexes from CSD: Subset 1, all Ln-containing complexes (49472 structures); Subset 2, mononuclear Ln complexes (27858 structures); Subset 3, mononuclear Ln complexes without cyclopentadienyl ligands (Cp) (26156 structures); Subset 4, Ln complexes with at least one 1,10-phenanthroline (phen) or its derivative as a coordinating ligand (2226 structures). The subsequent analysis of lanthanide complexes in these subsets examines the trends in coordination numbers and first shell distances as well as identifies and characterizes the ligands and donor groups. In addition, examples of Ln-complexes with commercially available complexants and phen-based ligands are interrogated in detail. This systematic investigation lays the groundwork for future data-driven ligand designs for REE separations based on the structural insights into the lanthanide coordination chemistry.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Crystal structure prediction with host-guided inpainting generation and foundation potentials

Unconditional crystal structure generation with diffusion models faces challenges in identifying symmetric crystals as the unit cell size increases. Here, we present the crystal host-guided generation (CHGGen) framework to address this challenge through conditional generation using an inpainting method, which optimizes a fraction of atomic positions within a predefined and symmetrized host structure to improve the success rate for symmetric structure generation. By integrating inpainting structure generation with a foundation potential for structure optimization, we demonstrate the method on the ZnS–P 2 S 5 and Li–Si chemical systems, where the inpainting method generates a higher fraction of symmetric structures than unconditional generation. The practical significance of CHGGen extends to enabling the structural modification of crystal structures, particularly for systems with partial occupancy or intercalation chemistry. The inpainting method also allows for seamless integration with other generative models, providing a versatile framework for accelerating materials discovery.

Zhong, Peichen [University of California, Berkeley↗
DOE OSTI2025

Impact of the d 0 transition metal on local structural transformations in disordered rock salt cathodes

Although it is widely accepted that the long-range (average) crystal structure plays a critical role in determining the electrochemical performance of battery materials, the relationship between local structural features and electrochemical performance is rarely studied. Disordered rock salt oxides (DRX), which have become serious contenders for next generation Li-ion electrode materials, provide an ideal platform for exploring correlations between local structure and electrochemical performance as they exhibit a simple face-centered cubic structure and combine long-range disorder and short-range order on the cation sublattice. This work examines the Li 1.1 Mn 0.7 Zr 0.2−x Ti x O 2 series of DRX cathodes and investigates the links between local structure rearrangements and capacity activation. The end-member Li 1.1 Mn 0.7 Zr 0.2 O 2 compound exhibits a low capacity in the as-synthesized state, attributed to unfavorable short-range order that hinders Li-ion transport, yet its capacity increases seven-fold, from 20 to 140 mAh g −1 , after chemical delithiation followed by a 400 °C heat treatment. Capacity activation is associated with the appearance of local spinel-like structural features that depart from the short-range order originally present in the material, without significant change to the bulk composition and average crystal structure. Investigation of a series of Li 1.1 Mn 0.7 Zr 0.2−x Ti x O 2 (x ≤ 0.2) DRX compounds reveals that the correlation length of the spinel-like ordering that emerges during the heat treatment strongly depends on the Zr : Ti ratio. Yet, dramatic capacity activation and electrochemical (pseudo-)plateaus reminiscent of Mn-based spinel cathodes are observed for all compounds irrespective of the size of the ordered domains. To explain this phenomenon, we propose that the DRX phase undergoes a complete transformation to a spinel-like domain structure, which improves bulk Li-ion transport regardless of domain size.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Repetitive proteins that undergo large conformational changes evade structural prediction algorithms

Protein structure prediction algorithms, such as AlphaFold, have accelerated protein design and advanced the understanding of the relationship between amino acid sequence and protein structure. However, these algorithms are limited in their ability to predict the structures of conformationally dynamic, intrinsically disordered, and stimuli-responsive proteins. To evaluate sequence-to-structure predictions of such challenging proteins, we explored a class of conformationally dynamic, repeats-in-toxin (RTX) proteins. RTX proteins adopt intrinsically disordered conformations in the absence of calcium and undergo reversible folding into β-roll structures upon binding to calcium. RTX proteins are characterized by tandem repeats of the sequence GGXGXDXUX, in which X can be any amino acid and U is an aliphatic amino acid. We designed RTX sequence variants with global substitutions of nonconserved amino acids, tandem repeats of consensus sequences GGAGXDTLY, and tandem repeats of scrambled sequences GGAGXDTYL. AlphaFold2 and AlphaFold3 predicted that all of these RTX variants adopt β-roll structures, characteristic of wild-type RTX bound to calcium. However, modeling the predicted structures with molecular dynamics simulations and characterizing the protein variants with circular dichroism spectroscopy, small-angle x-ray scattering, and x-ray crystallography revealed that variants adopt diverse, sequence-dependent structures in the absence and presence of calcium. To better design proteins for applications in biotechnology and sustainability, it is critical to build predictive tools that consider intrinsically disordered protein states and validate these tools with multi-mode, multi-scale experimental data.

Chang, Marina P. [Stanford Univ., CA (United State↗
DOE OSTI2024

Structural and functional insights into the interaction between the bacteriophage T4 DNA processing proteins gp32 and Dda

Abstract Bacteriophage T4 is a classic model system for studying the mechanisms of DNA processing. A key protein in T4 DNA processing is the gp32 single-stranded DNA-binding protein. gp32 has two key functions: it binds cooperatively to single-stranded DNA (ssDNA) to protect it from nucleases and remove regions of secondary structure, and it recruits proteins to initiate DNA processes including replication and repair. Dda is a T4 helicase recruited by gp32, and we purified and crystallized a gp32–Dda–ssDNA complex. The low-resolution structure revealed how the C-terminus of gp32 engages Dda. Analytical ultracentrifugation analyses were consistent with the crystal structure. An optimal Dda binding peptide from the gp32 C-terminus was identified using surface plasmon resonance. The crystal structure of the Dda–peptide complex was consistent with the corresponding interaction in the gp32–Dda–ssDNA structure. A Dda-dependent DNA unwinding assay supported the structural conclusions and confirmed that the bound gp32 sequesters the ssDNA generated by Dda. The structure of the gp32–Dda–ssDNA complex, together with the known structure of the gp32 body, reveals the entire ssDNA binding surface of gp32. gp32–Dda–ssDNA complexes in the crystal are connected by the N-terminal region of one gp32 binding to an adjacent gp32, and this provides key insights into this interaction.

Biochemistry & Molecular Biology↗
DOE OSTI2025

Assessment of thermally driven local structural phase changes in 1⁢𝑇′−MoTe 2

The role of layer disorder is important in establishing the topological phases of MoTe 2 . A rich tapestry of atomic ordering influences the structural phase transitions (SPTs), but there is little understanding of the mechanistic details of the phase transition. An atomistic level study was conducted to investigate the local structure of the 1⁢𝑇′ and 𝑇 𝑑 phases of MoTe 2 by using the pair distribution function (PDF) technique. While the average structure exhibits an SPT and coexistence of phases as a function of temperature, the local structure showed the suppression of SPT. The sample retained its monoclinic structure at all temperatures in short-range order. A sharp PDF peak observed at short distances indicated a strong atom-atom correlation between the Mo and Te atoms within the Mo octahedra. In addition, a large-box modeling of the PDF data indicated a preferential motion of Te atoms towards 𝑐 axis at all temperatures. Structural defects, such as stacking faults, likely result in the coexistence of phases in the average structure and suppress the local SPT of MoTe 2 . These results are stepping stones to understand the long-debated origins of structural, vibrational, and electronic properties of MoTe 2 and similar transition metal dichalcogenides.

2-dimensional systems↗
DOE OSTI2025

Signal sequences target enzymes and structural proteins to bacterial microcompartments and are critical for microcompartment formation

ABSTRACT Spatial organization of pathway enzymes has emerged as a promising tool to address several challenges in metabolic engineering, such as flux imbalances and off-target product formation. Bacterial microcompartments (MCPs) are a spatial organization strategy used natively by many bacteria to encapsulate metabolic pathways that produce toxic, volatile intermediates. Several recent studies have focused on engineering MCPs to encapsulate heterologous pathways of interest, but how this engineering affects MCP assembly and function is poorly understood. In this study, we investigated the role of signal sequences, short domains that target proteins to the MCP core, in the assembly of 1,2-propanediol utilization (Pdu) MCPs. We characterized two novel Pdu signal sequences on the structural proteins PduM and PduB, which constitute the first report of metabolosome signal sequences on structural proteins rather than enzymes. We then explored the role of enzymatic and structural Pdu signal sequences on MCP assembly by deleting their encoding sequences from the genome alone and in combination. Deleting enzymatic signal sequences decreased the MCP formation, but this defect could be recovered in some cases by overexpressing genes encoding the knocked-out signal sequence fused to a heterologous protein. By contrast, deleting structural signal sequences caused similar defects to knocking out the genes encoding the full-length PduM and PduB proteins. Our results contribute to a growing understanding of how MCPs form and function in bacteria and provide strategies to mitigate assembly disruption when encapsulating heterologous pathways in MCPs. IMPORTANCE Spatially organizing biosynthetic pathway enzymes is a promising strategy to increase pathway throughput and yield. Bacterial microcompartments (MCPs) are proteinaceous organelles that many bacteria natively use as a spatial organization strategy to encapsulate niche metabolic pathways, providing significant metabolic benefits. Encapsulating heterologous pathways of interest in MCPs could confer these benefits to industrially relevant pathways. Here, we investigate the role of signal sequences, short domains that target proteins for encapsulation in MCPs, in the assembly of 1,2-propanediol utilization (Pdu) MCPs. We characterize two novel signal sequences on structural proteins, constituting the first Pdu signal sequences found on structural proteins rather than enzymes, and perform knockout studies to compare the impacts of enzymatic and structural signal sequences on MCP assembly. Our results demonstrate that enzymatic and structural signal sequences play critical but distinct roles in Pdu MCP assembly and provide design rules for engineering MCPs while minimizing disruption to MCP assembly.

Johnson, Elizabeth R. (ORCID:0000000179236881)↗
DOE OSTI2025

Structural and functional analyses of SARS-CoV-2 Nsp3 and its specific interactions with the 5’ UTR of the viral genome

ABSTRACT Non-structural protein 3 (Nsp3) is the largest open reading frame encoded in the SARS-CoV-2 genome, essential for the formation of double-membrane vesicles (DMV) wherein viral RNA replication occurs. We conducted an extensive structure-function analysis of Nsp3 and determined the crystal structures of the ubiquitin-like 1 (Ubl1), nucleic acid binding (NAB), β-coronavirus-specific marker (βSM) domains, and a sub-region of the Y domain of this protein. We show that the Ubl1, ADP-ribose phosphatase (ADRP), human SARS Unique (HSUD), NAB, and Y domains of Nsp3 bind the 5’ UTR of the viral genome and that the Ubl1 and Y domains possess affinity for recognition of this region, suggesting high specificity. The Ubl1-Nucleocapsid (N) protein complex binds the 5’ UTR with greater affinity than the individual proteins alone. Our results suggest that multiple domains of Nsp3, particularly Ubl1 and Y, shepherd the 5’ UTR of the viral genome during translocation through the DMV membrane, priming the Ubl1 domain to load the genome onto N protein. IMPORTANCE The largest protein encoded by the SARS-CoV-2 genome is Nsp3. In infected cells, this multi-domain protein forms a pore structure in the virus-induced double-membrane vesicles (DMV). We have incomplete data on Nsp3 molecular structure, and here, we describe crystal structures for multiple domains of Nsp3. It is thought that newly replicated viral RNA transits through the DMV pore; however, we possess incomplete data on which regions of Nsp3 actually interact with RNA. Here, we present data showing that five domains of Nsp3 interact with the 5’ UTR of the SARS-CoV-2 RNA, including the Y domain for which no function has ever been discovered. These data suggest that the pore structure plays an active role in recognizing the terminal end of the genome, transiting and loading the viral RNA onto the cytoplasmic nucleocapsid protein. These data help expand our knowledge of Nsp3 structure and function and the SARS-CoV-2 replication cycle.

Microbiology↗
DOE OSTI2025

Confocal Raman Microscopy for Measuring In Situ Temperature-Dependent Structural Changes in Poly(Ethylene Oxide) Thin Films

Crystallization from the melt is a critical process governing the properties of semi-crystalline polymeric materials. While structural analyses of melting and crystallization transitions in bulk polymers have been widely reported, in contrast, those in thin polymer films on solid supports have been underexplored. Herein, in situ Raman microscopy and self-modeling curve resolution (SMCR) analysis are applied to investigate the temperature-dependent structural changes in poly(ethylene oxide) (PEO) films during melting and crystallization phase transitions. By resolving complex overlapping sets of spectra, SMCR analysis reveals that the thermal transitions of 50 µm thick PEO films comprise two structural phases: an ordered crystalline phase and a disordered amorphous phase. The ordered structure of the crystalline PEO film entirely disappears as the polymer is heated; conversely, the disordered structure of the amorphous PEO film reverts to the ordered structure as the polymer is cooled. Broadening of the Raman bands was observed in PEO films above the melting temperature (67 °C), while sharpening of bands was observed below the crystallization temperature (45 °C). The temperatures at which these spectral changes occurred were in good agreement with differential scanning calorimetry (DSC) measurements, especially during the melting transition. The results illustrate that in situ Raman microscopy coupled with SMCR analysis is a powerful approach for unraveling complex structural changes in thin polymer films during melting and crystallization processes. Furthermore, we show that confocal Raman microscopy opens opportunities to apply the methodology to interrogate the structural features of PEO or other surface-supported polymer films as thin as 2 µm, a thickness regime beyond the reach of conventional thermal analysis techniques.

Koh, Miharu [Department of Chemistry, University o↗
DOE OSTI2026

Doping Metallic Grain Boundaries to Control Atomic Structure and Damage Tolerance

Grain boundaries often act as sites for crack and void nucleation during plastic deformation of metallic materials. While it is known that grain boundary character and structure can greatly influence this damage nucleation process, the current level of control over such details is limited. The objective of this project was to obtain a fundamental understanding of how metallic grain boundary structure can be controlled through intelligent doping, with the idea of inducing planned amorphous grain boundary phases or complexions. The effect of amorphous complexion structure on dislocation accommodation mechanisms was studied, to improve the field’s understanding of damage nucleation at a promising type of interface. Different microstructural descriptors were studied, as grain boundaries can have large variations and complexity of local structure, and influence the mechanical damage resistance of these features was tested. Nanocrystalline systems that contain amorphous grain boundary complexions were prioritized, as these have more extreme variations in interfacial structure and damage tolerance, with an emphasis on isolating the importance of complexion population (type, thickness, etc.), network topology, local structure, and local chemistry. This research used a combination of computational, experimental, and characterization techniques to isolate and understand the importance of nanoscale grain boundary structure and interfacial chemistry. Amorphous grain boundary complexions were found to clearly increase a material’s resistance to mechanical damage, with the local distribution of structural short-range order within the complexions found to be an important descriptor for damage and the proposed focus of future work in this area.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Symmetry-Based Structured Matrices for Efficient Approximately Equivariant Networks

There has been much recent interest in designing symmetry-aware neural networks (NNs) exhibiting relaxed equivariance. Such NNs aim to interpolate between being exactly equivariant and being fully flexible, affording consistent performance benefits. In a separate line of work, certain structured parameter matrices -- those with displacement structure, characterized by low displacement rank (LDR) -- have been used to design small-footprint NNs. Displacement structure enables fast function and gradient evaluation, but permits accurate approximations via compression primarily to classical convolutional neural networks (CNNs). In this work, we propose a general framework -- based on a novel construction of symmetry-based structured matrices -- to build approximately equivariant NNs with significantly reduced parameter counts. Our framework integrates the two aforementioned lines of work via the use of so-called Group Matrices (GMs), a forgotten precursor to the modern notion of regular representations of finite groups. GMs allow the design of structured matrices -- resembling LDR matrices -- which generalize the linear operations of a classical CNN from cyclic groups to general finite groups and their homogeneous spaces. We show that GMs can be employed to extend all the elementary operations of CNNs to general discrete groups. Further, the theory of structured matrices based on GMs provides a generalization of LDR theory focussed on matrices with cyclic structure, providing a tool for implementing approximate equivariance for discrete groups. We test GM-based architectures on a variety of tasks in the presence of relaxed symmetry. We report that our framework consistently performs competitively compared to approximately equivariant NNs, and other structured matrix-based compression frameworks, sometimes with a one or two orders of magnitude lower parameter count.

Samudre, Ashwin↗
DOE OSTI2025

Basin-Scale Structural Features Database

The Basin-Scale Structural Features database provides spatial datasets of faults, fractures, folds, and earthquakes compiled from public, authoritative sources (e.g., U.S. Geological Survey and State Geological Surveys) and aggregated into derivative forms to support subsurface assessments. Recognizing that characterizing basin-scale structural features requires interpreting data that are often ambiguous or lack key information, the source data were evaluated using a knowledge-data framework and geospatial fuzzy logic method (Justman et al., 2020) to represent both measured (observed) and predicted (inferred or potential) structural features as derivative datasets. This workflow employs conceptual models for known structural features and predicted structural features, incorporating geospatial data to estimate potential, even with limited data. The aim is to aid and support an understanding of basin-scale features and identify potential gaps in data and knowledge. As of 4/30/2025, the database includes resources for nine sedimentary basins: Appalachian, Denver, U.S. Gulf Coast, Illinois, Michigan, Permian, Sacramento, San Joquin and Williston. The database is organized by basin and then data category: 1) Faults, fractures, folds, 2) Earthquakes, 3) Topographic, 4) Structural contours and isopachs, 5) Geophysical, and 6) Structural feature density assessment maps.

basin scale↗