<p>Large-scale thermal use of shallow groundwater is often constrained in cities because temperature plumes can extend far beyond project boundaries and affect third-party water rights. Unidirectional Aquifer Thermal Energy Storage (UD-ATES) addresses this by reversing the conventional open-loop arrangement. The injection well is placed up-gradient and the production well down-gradient. During summer cooling, warmed return water is injected up-gradient; the resulting warm plume is carried by the natural groundwater flow to the down-gradient well and can be recovered in the following heating season. Conversely, during the heating season, cooled water is injected up-gradient; the resulting cold plume drifts down-gradient and can be recaptured for cooling in the next summer. This configuration is particularly suited to shallow, highly permeable aquifers with pronounced natural gradients, settings in which classical ATES suffers from advective losses, while also minimizing off-site thermal impacts that complicate permitting.</p>
<p>At the State Hospital Graz South site (Austria), we surveyed and characterized the aquifer and built a coupled groundwater-flow and heat-transport model to design a UD-ATES well pair tailored to local conditions. The optimized spacing between injection and production wells is <span class="inline-formula">∼463</span> m, aligning transport time with the seasonal load profile with a peak thermal power of 1.25 MW (60 L s<span class="inline-formula"><sup>−1</sup></span> by a <span class="inline-formula">Δ<i>T</i></span> of 5 K). Resulting temperature anomalies remain largely confined to the property, with the thermal signal decaying to below 1 K within a few hundred metres downstream. Despite an unavoidable imbalance between heating and cooling demand over the year, the system recovers a substantial fraction of the injected energy and markedly reduces the thermal footprint compared with a conventional open loop scheme. The thermal recovery factor amounts to 0.38. An expansion of the plant to a total peak thermal power of <span class="inline-formula">>3.5</span> MW using three pairs of wells appears to be feasible at the location in question. These findings support UD-ATES as a practical pathway to decarbonize large, space-constrained consumers in high-flow aquifers while safeguarding neighbouring groundwater uses.</p>
The proliferation of AI-powered search engines has shifted information discovery from traditional link-based retrieval to direct answer generation with selective source citation, creating new challenges for content visibility. While existing Generative Engine Optimization (GEO) approaches focus primarily on semantic content modification, the role of structural features in influencing citation behavior remains underexplored. In this paper, we propose GEO-SFE, a systematic framework for structural feature engineering in generative engine optimization. Our approach decomposes content structure into three hierarchical levels: macro-structure (document architecture), meso-structure (information chunking), and micro-structure (visual emphasis), and models their impact on citation probability across different generative engine architectures. We develop architecture-aware optimization strategies and predictive models that preserve semantic integrity while improving structural effectiveness. Experimental evaluation across six mainstream generative engines demonstrates consistent improvements in citation rate (17.3 percent) and subjective quality (18.5 percent), validating the effectiveness and generalizability of the proposed framework. This work establishes structural optimization as a foundational component of GEO, providing a data-driven methodology for enhancing content visibility in LLM-powered information ecosystems.
Manipulating dynamic objects remains an open challenge for Vision-Language-Action (VLA) models, which, despite strong generalization in static manipulation, struggle in dynamic scenarios requiring rapid perception, temporal anticipation, and continuous control. We present DynamicVLA, a framework for dynamic object manipulation that integrates temporal reasoning and closed-loop adaptation through three key designs: 1) a compact 0.4B VLA using a convolutional vision encoder for spatially efficient, structurally faithful encoding, enabling fast multimodal inference; 2) Continuous Inference, enabling overlapping reasoning and execution for lower latency and timely adaptation to object motion; and 3) Latent-aware Action Streaming, which bridges the perception-execution gap by enforcing temporally aligned action execution. To fill the missing foundation of dynamic manipulation data, we introduce the Dynamic Object Manipulation (DOM) benchmark, built from scratch with an auto data collection pipeline that efficiently gathers 200K synthetic episodes across 2.8K scenes and 206 objects, and enables fast collection of 2K real-world episodes without teleoperation. Extensive evaluations demonstrate remarkable improvements in response speed, perception, and generalization, positioning DynamicVLA as a unified framework for general dynamic object manipulation across embodiments.
Real-time prediction of rock slope stability in active mines remains a critical challenge due to complex geology, dynamic mining stress, and environmental factors. The Pulang Copper Mine, with its complex structural setting and ongoing subsidence, requires advanced monitoring to mitigate failure risk. This study aimed to develop and validate an IoT-driven Enhanced Transformer model for real-time prediction of the Factor of Safety (FoS) and stability classification, integrating numerical simulation, IoT data streaming, and deep learning to improve early-warning capability. A FLAC3D simulation replicated two years of mining (730 daily steps) at six strategic monitoring points, generating time-series data for displacement, velocity, acceleration, and FoS. An IoT framework streamed this data with <5-second latency. An Enhanced Transformer architecture with multi-head self-attention, multi-task learning ( λ =0.5), and advanced feature engineering was trained on the sequences. The Enhanced Transformer achieved superior performance, with testing R² ranging from 0.416 (Station 5, characterized by complex transitional kinematics) to 0.991 (Station 4). Testing MAE ranged [0.003596 (Station 2)–0.019071 (Station 5)], a reduction of up to 88% compared to the Standard Transformer. For four-class stability classification, the model attained a mean test accuracy of 0.993, with critical-class recall reaching 1.0—guaranteeing zero missed alarms for life-threatening critical and unstable conditions, the paramount objective for early-warning systems. The proposed IoT-Enhanced Transformer model provides a highly accurate, real-time solution for slope stability prediction, significantly outperforming conventional models.
<p>This study investigates the biomechanical properties of marram grass (<i>Calamagrostis arenaria</i>, formerly <i>Ammophila arenaria</i>) over a 12-month period on the island of Spiekeroog, Germany, to enhance the modeling of coastal dune dynamics. The research reveals significant seasonal variations in the stiffness and Young modulus of the vegetation, with higher values observed in winter, indicating increased mechanical resistance important for dune stability during storm events. In summer, increased flexibility and density are prominent, enhancing dune accretion. To account for these dynamics, the study emphasizes the importance of incorporating seasonally adjusted parameters into models, particularly accounting for the increased horizontal density, the presence of flower stems in summer, and the longer leaf lengths in winter. The differentiation among plant parts is highlighted, with flower stems providing the highest structural support due to their greater stiffness, while leaves contribute more to flexibility and dynamic responses. Interestingly, the minimal differences between green and brown leaves suggest that these can be treated similarly in modeling efforts, simplifying parameterization without compromising accuracy. Additionally, the study found no consistent evidence that wind exposure significantly affects the biomechanical properties of marram grass, suggesting that wind influence may not need to be factored into biomechanical models. The results also demonstrate that the biomechanical properties of marram grass are broadly transferable between fixed and dynamic dune systems, supporting their applicability across various coastal environments. The key outcome of this research is the detailed compilation of the biomechanical traits of marram grass's aboveground vegetation, reflecting the seasonal dynamics found in dune processes, which will serve as a valuable resource for future modeling efforts of dune vegetation and their surrogates.</p>
Boundaries in Hypernetwork Theory (HT) are non-structural tags that restrict visibility without altering the underlying hypernetwork. They attach to hypersimplices as annotations and participate in no identity, typing, or alpha/beta semantics. Projection over a boundary, B(H, b) = pi_b(H), is filtering only: it selects exactly those hypersimplices carrying b and preserves all axioms of the structural kernel. The backcloth remains immutable, and no new structure is created, removed, or inferred. This paper formalises boundaries as a simple and conservative scoping mechanism. It clarifies their syntax, their interaction with projection, and their use in producing identity-preserving subsystem views that support modular modelling and overlapping perspectives. The account also makes explicit why conservative scoping matters: boundaries provide reproducible view extraction, stable subsystem isolation, and safe model exploration without altering the global structure. Scoped operator application is defined as ordinary structural-kernel composition applied to projected views, ensuring that view-level reasoning remains local and does not modify the global hypernetwork. This establishes a disciplined separation between immutable structure and scoped analysis while retaining full compatibility with the structural kernel. The paper includes a worked example demonstrating how boundaries yield coherent, identity-preserving subsystem views and how scoped reasoning supports refinement within these views. The result is a precise and minimal account of boundaries that complements - but does not extend - the structural kernel and completes the scoping mechanism required for practical multilevel modelling with HT.
We analyze the behavior of stoquastic transverse-field quantum annealing (TFQA) on a structured class of Maximum Independent Set (MIS) instances, using the same decomposition framework developed in our companion work on the DIC-DAC-DOA algorithm (Beyond Stoquasticity). For these instances, we provide a structural explanation for the anti-crossing arising from the competition between the energies associated with a set of degenerate local minima (LM) and the global minimum (GM), and analytically derive the associated exponentially small gap. Our analysis proceeds in two steps. First, we reduce the dynamics to an effective two-block Hamiltonian $H_{core}$, constructed from the bare (decoupled) subsystems associated with the LM and GM. This reduction is justified analytically using the structural decomposition. Second, we reformulate the eigenvalue problem as a generalized eigenvalue problem in a non-orthogonal basis constructed from the bare eigenstates of the subsystems. This transformation enables a clean perturbative treatment of the anti-crossing structure, independent of the transverse field, unlike standard perturbation theory approach, which requires treating the transverse field as a small parameter. This paper serves as a supplementary companion to our main work on the DIC-DAC-DOA algorithm, where we demonstrate how appropriately designed non-stoquastic drivers can bypass this tunneling-induced bottleneck.
Data contamination has received increasing attention in the era of large language models (LLMs) due to their reliance on vast Internet-derived training corpora. To mitigate the risk of potential data contamination, LLM benchmarking has undergone a transformation from static to dynamic benchmarking. In this work, we conduct an in-depth analysis of existing static to dynamic benchmarking methods aimed at reducing data contamination risks. We first examine methods that enhance static benchmarks and identify their inherent limitations. We then highlight a critical gap-the lack of standardized criteria for evaluating dynamic benchmarks. Based on this observation, we propose a series of optimal design principles for dynamic benchmarking and analyze the limitations of existing dynamic benchmarks. This survey provides a concise yet comprehensive overview of recent advancements in data contamination research, offering valuable insights and a clear guide for future research efforts. We maintain a GitHub repository to continuously collect both static and dynamic benchmarking methods for LLMs. The repository can be found at this link.
Rameen Abdal, Or Patashnik, Ekaterina Deyneka
et al.
Recent advances in text-to-video generation have enabled high-quality synthesis from text and image prompts. While the personalization of dynamic concepts, which capture subject-specific appearance and motion from a single video, is now feasible, most existing methods require per-instance fine-tuning, limiting scalability. We introduce a fully zero-shot framework for dynamic concept personalization in text-to-video models. Our method leverages structured 2x2 video grids that spatially organize input and output pairs, enabling the training of lightweight Grid-LoRA adapters for editing and composition within these grids. At inference, a dedicated Grid Fill module completes partially observed layouts, producing temporally coherent and identity preserving outputs. Once trained, the entire system operates in a single forward pass, generalizing to previously unseen dynamic concepts without any test-time optimization. Extensive experiments demonstrate high-quality and consistent results across a wide range of subjects beyond trained concepts and editing scenarios.
The dynamic coupled hydro‐thermo‐mechanical behavior of the unlined structure in the saturated porous structure under extreme geotechnical and geology engineering (e.g., underground explosion, laser thermal rock breaking) has aroused extensive research interests in the constitutive modeling and transient dynamic responses prediction. Despite the current hydro‐thermo‐mechanical models that have been historically proposed, the model construction is still based on the classical thermoelastic coupling theory (Fourier heat conduction model). In the study of coupled heat transfer in extreme environments, the heat flux at a certain point is not only affected by the instantaneous heat source but also depends on the temperature gradient at that point and the effect of its historical heat flow. To address this deficiency, this work aims to construct a new hydro‐thermo‐mechanical coupling model by introducing the multi‐dual‐phase lag heat conduction law. The proposed model is applied to investigate the transient structural dynamic hydro‐thermo‐mechanical response of a cylindrical unlined tunnel in the poroelastic medium by applying the Laplace transformation approach. The influences of the parameters of heat flux lag and temperature lag on the wave propagation as well as the dimensionless responses of temperature, displacement, stress, and pore water pressure were evaluated and discussed.
Geologic carbon storage is undergoing rapid commercialization with many projects planned or in the early stages of construction and/or permitting. The Gulf of Mexico is widely recognized as a promising geographic region for carbon storage because the geology is well characterized from oil and gas exploration and there are several metro areas such as Corpus Christi and Victoria along the Texas coast which are principal sites for refineries, petrochemicals, and LNG production. Additionally, two major growth faults in this region, the Corsair and Clemente-Tomas Faults, run parallel to the Texas coast and create structural trapping opportunities for CO 2 storage. As part of the SEG EVOLVE Carbon Solutions program, this study develops a site-scale evaluation of the Miocene sands within the Matagorda Island region, offshore Texas. This study builds a reservoir model based on 3D seismic data, completes static and dynamic modeling to assess reservoir storage capacity, and develops a cost estimate for long-term storage. This study utilizes 3D seismic reflection data combined with well-log data to identify potential reservoir and sealing units within Miocene age formations as well as major structural features such as the Corsair and Clemente-Thomas faults. Key formations are exported from the seismic to construct a static geomodel to assess potential storage volumes and property distributions. Dynamic simulations are run to assess the trapping characteristics of the saline reservoirs for CCS potential. Results from 3D seismic mapping demonstrate the extensive lateral distribution of potential high quality reservoir units within Miocene saline aquifers. Additionally, the Corsair and Clemente-Thomas faults create structural traps that are likely capable of storing millions of tonnes of CO 2 . Simulation results combined with economic projections provide evidence that Matagorda Island is capable of hosting a hub-scale CCS project, resulting in a positive project revenue under current tax credit systems. While many CCS projects are
R. B. Medeiros, L. S. A. Dos Santos, J. R. L. Bezerra
et al.
The landscape cartography assesses the functional, dynamic, structural and morphological aspects of landscapes, regardless of their taxonomic scale. It seeks to use these units to support environmental and territorial planning and management. Thus, the present study sought to apply this line of analysis to the Pindaré River Basin, precisely in its lower course, located in the Brazilian state of Maranhão. The objective was to identify, classify, map and analyze the landscapes of the lower course through the correlation of variables related to geology, relief, soils, land use and land cover providing data to support and promote preservationist and conservationist public policies and actions in the area. The methodology identified four levels of landscape analysis, from morphometric aspects, geoforms and upper units to reaching the final landscape map, using field output, digital elevation models and satellite images to validate information. The procedures allowed to identify the landscape heterogeneity in a unique environment of saturated and periodically flooded soils contrasting with extensive pastures and little native vegetation. As a result, seven first-level landscape units were identified, coming up to fifty-eight sub-units in the final map. The work aims to apply the methodology in an area of the Maranhão State where few studies on landscape cartography have occurred. The target is to comprehend possible relationships between the functional and structural potential of landscapes and their relationship with the current intensity of land use, contributing to physical- territorial planning permeating geoecological sustainability.
The Middle Eocene, shallow, dolomitic, high salinity aquifer has significant importance as the main source of injection water at the present time in order to maintain reservoir pressure above the bubble point in maturing oil fields in southern Iraq until other sources of injection water become available. Therefore, in this study, the Dammam aquifer was studied in detail by integrating all available data, including 3D seismic, well information, well logs, and core data. A regional aquifer static model has been constructed to better understand subsurface geology and in order to be ready to be used in the construction of a sophisticated dynamic model to predict whether the Dammam aquifer can supply enough water for injection or not. More than 184 wells have been used in the present study. The structural framework was built according to 3D seismic cube and well tops. The average thickness is about 235 mm. In order to understand the lateral and vertical connectivity, a facies model was created in addition to the porosity and permeability models with input from the core and a Nuclear Magnetic Resonance (NMR) log. According to the facies change, the Dammam aquifer has been divided from bottom to top into five units (MD50, MD100, MD200, MD300, and MD400). The top of the Dammam formation varies from 700m in the southeast to 1000m in the north-west. The porosity in the Dammam formation is very high and varies from 12 to 45 PU, with an average porosity of 29 PU. In order to reduce uncertainty, the study recommends that a new rock core have to be cut, in addition to a number of NMR and Formation Micro Imager (FMI) logs needing to be run into selected wells
Logit dynamics are evolution equations that describe transitions to equilibria of actions among many players. We formulate a pair-wise logit dynamic in a continuous action space with a generalized exponential function, which we call a generalized pair-wise logit dynamic, depicted by a new evolution equation nonlocal in space. We prove the well-posedness and approximability of the generalized pair-wise logit dynamic to show that it is computationally implementable. We also show that this dynamic has an explicit connection to a mean field game of a controlled pure-jump process, with which the two different mathematical models can be understood in a unified way. Particularly, we show that the generalized pair-wise logit dynamic is derived as a myopic version of the corresponding mean field game, and that the conditions to guarantee the existence of unique solutions are different from each other. The key in this procedure is to find the objective function to be optimized in the mean field game based on the logit function. The monotonicity of the utility is unnecessary for the generalized pair-wise logit dynamic but crucial for the mean field game. Finally, we present applications of the two approaches to fisheries management problems with collected data.