Hard Boundary Research Articles

SIAT: Document-level Event Extraction via Spatiality-Augmented Interaction Model with Adaptive Thresholding

Document-level event extraction endeavors to automatically extract structural events from a given document. Many existing approaches focus on modeling entity interactions and decoding these interactions into events, assigning each entity as an event argument. However, these approaches encounter two primary limitations: they exclusively capture semantic dependencies to model entity interactions, overlooking the indication of the spatial distribution features of entities; they decode interactions imprecisely with a hard binary-classification boundary, potentially failing to calibrate micro differences in interactions. To overcome these limitations, we introduce a novel approach termed the S patiality-augmented I nteraction Model with A daptive T hresholding (SIAT). Our method addresses the first limitation by calculating the relative position encoding of entities to represent spatial interaction features. These features are then integrated with multi-granularity semantic interactions, enhancing the modeling of entity interactions for each entity pair. Furthermore, we introduce an adaptive event decoding mechanism, which establishes a more flexible decision boundary for different entity interactions. Additionally, an adaptive loss function for threshold learning is designed to further refine the model. Experimental results demonstrate that our proposed method achieves competitive performance compared to state-of-the-art methods on two public event extraction datasets while maintaining considerable training efficiency.

Near-α titanium alloy dwell load-induced deformation twinning to coordinate the deformation mechanism associated with crack initiation

In this study, we identified a specific phenomenon of coordinated deformation of twins in near-α titanium alloys during dwell fatigue (DF). The main crack source regions and internal cracks in the micro-texture region (MTR) and no-MTR samples with inconsistent orientation characteristics were characterized. The results demonstrate that the main cracks in all specimens are aligned with the (0001) basal plane, which is associated with basal slip. Notably, twins are found and confirmed to be involved in the DF crack initiation process in high Al content near-α Ti60 alloys where deformation twins are rare, and tend to nucleate at the DF basal cracks. Further in-situ dwell investigation reveals and proposes that the dislocation pile-up and prismatic-basal (PB) interfacial features between soft/hard grains lead to deformation twin nucleation and growth from hard grain boundaries. Concurrently, the pyramidal 〈c+a〉 dislocation slip is observed to be in concurrent operation with the induced twins. These findings suggest that deformation twins not only coordinate basal grain deformation but also hinder the initiation and propagation of basal cracks caused by basal 〈a〉 slip. Moreover, a high density of pyramidal 〈c+a〉 dislocations within the twin and their dissociation to form basal stacking faults (SFTs) with a specific nanometric spacing make the twinning process accompanied by significant lattice distortions inside the twin. These newly formed nanotwin boundaries and SFTs act as barriers to dislocation motion, enhancing the strength and DF lifetime of near-α Ti60 alloys and effectively reducing the dwell sensitivity of no-MTR alloys. Our findings extend the understanding of the coordinated roles of dislocation slip, twin nucleation and formation of basal SFTs in near-α titanium alloys in dwell fatigue.

Ultralow-frequency absorption mechanism of a hybrid membrane resonator with acoustic soft boundary condition

Hybrid membrane resonator (HMR) as a typical metamaterial-based absorber with acoustic hard boundary condition (AHBC) has demonstrated excellent noise absorption abilities, but the challenge lies in achieving broadband absorption of ultralow-frequency noise in the tens of Hz to 150 Hz regime. In this research, we investigate systematically the absorption characteristics, absorption mechanisms, and casual optimality of an HMR with three openings in lower surface of the cavity, which functioned as an acoustic soft boundary condition (ASBC). Compared to the well-studied HMR with AHBC, a new absorption mechanism, which states that most energy dissipates occur in the opening region rather than in the membrane, has been found first. As a result, the HMR with ASBC can demonstrate outstanding ultralow-frequency sound absorption performance with a very small thickness, and the full width at half maximum of the HMR can be enlarged over 7 times. Furthermore, the causal inequalities of the absorbers with ideal AHBC and ASBC are derived based on the Cauchy integral and causality principle. The causal optimality of the proposed absorber is also achieved. This research provides valuable guidelines for the design of excellent ultralow-frequency sound absorbers which could contribute to solving the major issue of noise reduction.

Geochemical anomaly separation based on geology, geostatistics, compositional data and local singularity analyses: A case study from the kuh panj copper deposit, Iran

This study combines geochemical anomaly separation with geostatistical approaches and compositional data analysis. To have a reasonable model for abnormal areas, suggesting the optimal drilling coordinates, one needs to consider some effective geological controllers on the mineralization such as lithology, alteration, and tectonic processes. The lithogeochemical samples at 608 locations were processed from the copper porphyry mineralization area of Kuh Panj, Iran, to illustrate the proposed approach. The heterogeneity of rock type domains was detected using contact analysis. Based on this, plurigaussian simulation made it possible to model the uneconomic dykes of the area. Copper-molybdenum grades and filler as geochemical components were transformed into multivariate normal data using isometric logarithmic ratio-flow anamorphosis. Hierarchical rock type-grade simulation reproduced the grade behavior at hard rock type boundaries. Selection of a dense and regular simulation grid minimized the smoothing effect caused by interpolation of the singularity index values. Finally, unsupervised machine learning identified the anomaly zone by clustering the results of the singularity analysis. Validation using drilling data and the geological map shows that the anomaly separation, considering the geological conditions and the compositional nature of the geochemical data, can provide a reliable model of the geochemical anomalies in the Kuh Panj deposit.

Unsupervised physics-informed deep learning for assessing pulmonary artery hemodynamics

Deep learning advancements have significantly benefited medical applications. One such helpful application is noninvasive fractional flow reserve (FFR) evaluation along the pulmonary artery tree, which aids in planning optimal balloon pulmonary angioplasty. This study proposes a surrogate model that employs an unsupervised physics-informed neural network (UPNN) to predict FFR based on pulmonary CT angiography. To ensure the UPNN strictly follows the unique solution of the governing equations, we implemented a hard boundary conditions enforcement approach. Subsequently, a finite difference convolutional filter was developed to enhance connectivity among neighboring points. This allows the neural networks to propagate boundary conditions into the unseen vessel interior and perceive the geometric structure of the computational domain. We also introduced regularization constraints on the three components contributing to pressure drop within the artery tree. A total of 4500 synthetic pulmonary artery trees were used to train the UPNN with impedance outlet boundary conditions. We found that the limits of agreement of FFR trained by UPNN ranged from −0.05 to 0.04 with computational fluid dynamics (CFD) simulation results. The testing results showed that the correlation coefficient between FFR predicted by UPNN and CFD were 96.1% and 94.7% for synthetic and patient-specific data, respectively. Using invasive FFR as reference, the accuracy of FFR predicted by UPNN and CFD was 81.4% and 84.8%, respectively. These results demonstrate that our UPNN-based surrogate model’s performance in evaluating FFR aligns closely with CFD simulations.

Phase and group speeds of airborne surface waves over porous layers and periodically rough hard surfaces.

Sound fields above porous layers or rough acoustically hard boundaries may include airborne surface waves. The surface wave properties depend on the effective surface admittance. Analytical expressions for surface wave speeds are derived from models for the acoustical properties of rigid porous media. Surface wave effects on measurements of level difference spectra over porous asphalt are investigated and predictions of phase, group speeds, and vertical attenuation of the surface waves over externally reacting hard backed layers corresponding to a porous asphalt are compared. Predictions of surface wave characteristics above an identical vertical slit medium are compared with data obtained over arrays of parallel aluminum strips on an acoustically hard surface. Group speeds of surface waves over lattices, parallel regularly spaced strips, and snow obtained by numerical differentiation of the phase speed spectra corresponding to admittance spectra deduced from complex excess attenuation are found to compare well with those estimated from time domain data. An effective admittance, deduced from a boundary element method simulation of the excess attenuation spectrum over regularly spaced ribs so that the frequency of the peak corresponds with that in the measured spectrum, is used to estimate the group speed of the associated surface wave.

An Emergent Quadruple Phase Ensemble in Doped Bismuth Ferrite Thin Films Through Site and Strain Engineering

AbstractIn ferroic materials, giant susceptibilities can be realized at artificially constructed phase boundaries through deterministic manipulation of the order parameter. Here, emergent ferroelectric structural phase evolution behavior is demonstrated through a synergistic combination of A‐site doping and strain engineering. Using chemical solution deposition derived (001)‐oriented Sm‐substituted bismuth ferrite (Bi1‐xSmxFeO3) films as a prototypical system, a morphotropic phase boundary comprising a coexistence of four distinct crystallographic phases is uncovered. These ferroelectric, polar, and nonpolar phases form a nanoscale mixture without the presence of crystallographically hard boundaries. The system thus possesses the ability to show both polarization rotation and extension, effectively releasing the polarization from its crystallographic constraint. Consequently, both robust ferroelectric properties and giant electromechanical responses are obtained. For instance, the optimized composition with x = 0.14 has a remnant polarization of 2Pr = 103 µC cm−2 and electromechanical response 175% that of undoped BFO. These findings showcase the tremendous potential of synthetic phase boundaries, particularly in the context of lead‐free functional multiferroics.

Soft No More: Gas Shielding Protects Soft Binaries from Disruption in Gas-rich Environments

Binaries in dense environments are traditionally classified as soft or hard based on their binding energy relative to the kinetic energy of surrounding stars. Heggie’s law suggests that stellar encounters tend to soften soft binaries and harden hard binaries, altering their separations. However, interactions with gas in such environments can significantly modify this behavior. This study investigates the impact of gas on binary softening and its consequences. We find that gas interactions can actually harden binaries, extending the soft–hard boundary to larger separations. This introduces a “shielding radius” within which binaries are likely to harden due to gas interactions, surpassing the traditional soft–hard limit. Consequently, a notable portion of binaries initially classified as “soft” may become “hard” when both gas and stars are considered. We propose a two-stage formation process for hard binaries: initial soft binary formation, either dynamically or through gas-assisted capture, followed by gas-induced hardening before eventual disruption. In environments with low gas density but high gas content, the shielding radius could exceed the typical hard–soft limit by 1 order of magnitude, leading to a significant fraction of originally soft binaries effectively becoming hard. Conversely, in high-gas-density environments, gas-induced hardening may dominate, potentially rendering the entire binary population hard. Gas hardening emerges as a crucial factor in shaping binary populations in gas-rich settings, such as clusters, star-forming regions, and possibly active galactic nucleus disks. This highlights the complex interplay between gas dynamics and stellar interactions in binary evolution within dense environments.

Relating absorbing and hard wall boundary conditions for a one-dimensional run-and-tumble particle

The connection between absorbing boundary conditions and hard walls is well established in the mathematical literature for a variety of stochastic models, including for instance the Brownian motion. In this paper we explore this duality for a different type of process which is of particular interest in physics and biology, namely the run-tumble-particle, a toy model of active particle. For a one-dimensional run-and-tumble particle (RTP) subjected to an arbitrary external force, we provide a duality relation between the exit probability, i.e. the probability that the particle exits an interval from a given boundary before a certain time t, and the cumulative distribution of its position in the presence of hard walls at the same time t. We show this relation for a RTP in the stationary state by explicitly computing both quantities. At finite time, we provide a derivation using the Fokker–Planck equation. All the results are confirmed by numerical simulations.

Shape from dots: a window into abstraction processes in visual perception

IntroductionA remarkable phenomenon in perception is that the visual system spontaneously organizes sets of discrete elements into abstract shape representations. We studied perceptual performance with dot displays to discover what spatial relationships support shape perception.MethodsIn Experiment 1, we tested conditions that lead dot arrays to be perceived as smooth contours vs. having vertices. We found that the perception of a smooth contour vs. a vertex was influenced by spatial relations between dots beyond the three points that define the angle of the point in question. However, there appeared to be a hard boundary around 90° such that any angle 90° or less was perceived as a vertex regardless of the spatial relations of ancillary dots. We hypothesized that dot arrays whose triplets were perceived as smooth curves would be more readily perceived as a unitary object because they can be encoded more economically. In Experiment 2, we generated dot arrays with and without such “vertex triplets” and compared participants’ phenomenological reports of a unified shape with smooth curves vs. shapes with angular corners. Observers gave higher shape ratings for dot arrays from curvilinear shapes. In Experiment 3, we tested shape encoding using a mental rotation task. Participants judged whether two dot arrays were the same or different at five angular differences. Subjects responded reliably faster for displays without vertex triplets, suggesting economical encoding of smooth displays. We followed this up in Experiment 4 using a visual search task. Shapes with and without vertex triplets were embedded in arrays with 25 distractor dots. Participants were asked to detect which display in a 2IFC paradigm contained a shape against a distractor with random dots. Performance was better when the dots were sampled from a smooth shape than when they were sampled from a shape with vertex triplets.Results and discussionThese results suggest that the visual system processes dot arrangements as coherent shapes automatically using precise smoothness constraints. This ability may be a consequence of processes that extract curvature in defining object shape and is consistent with recent theory and evidence suggesting that 2D contour representations are composed of constant curvature primitives.

A Stepwise Cosimulation Framework for Modeling Critical Elements in Copper Porphyry Deposits

The increased attention given to batteries has given rise to apprehensions regarding their availability; they have thus been categorized as essential commodities. Cobalt (Co), copper (Cu), lithium (Li), nickel (Ni), and molybdenum (Mo) are frequently selected as the primary metallic elements in lithium-ion batteries. The principal aim of this study was to develop a computational algorithm that integrates geostatistical methods and machine learning techniques to assess the resources of critical battery elements within a copper porphyry deposit. By employing a hierarchical/stepwise cosimulation methodology, the algorithm detailed in this research paper successfully represents both soft and hard boundaries in the simulation results. The methodology is evaluated using several global and local statistical studies. The findings indicate that the proposed algorithm outperforms the conventional approach in estimating these five elements, specifically when utilizing a stepwise estimation strategy known as cascade modeling. The proposed algorithm is also validated against true values by using a jackknife method, and it is shown that the method is precise and unbiased in the prediction of critical battery elements.

Simulation trouble and gender trouble

ABSTRACT Is it impossible to imaginatively simulate what it’s like to be someone with a different gender experience – to understand them empathically? Or is it simply difficult, a challenge requiring effort and dedication? I first distinguish three different sorts of obstacle to empathic understanding that are sometimes discussed: Missing Ingredient problems, Awkward Combination Problems, and Inappropriate Background Problems. I then argue that, although all three should be taken seriously, there is no clear reason to think that any of them are both genuinely intractable and also significant impediments to the kind of empathic understanding we might want. To that extent, the challenge to interpersonal understanding here may be more a matter of skill and work than a hard boundary.

Internally heated and fully compressible convection: Flow morphology and scaling laws

In stars and planets natural processes heat convective flows in the bulk of a convective region rather than at hard boundaries. By characterizing how convective dynamics are determined by the strength of an internal heating source we can gain insight into the processes driving astrophysical convection. Internally heated convection has been studied extensively in incompressible fluids, but the effects of stratification and compressibility have not been examined in detail. In this work, we study fully compressible convection driven by a spatially uniform heating source in two-dimensional (2D) and 3D Cartesian, hydrodynamic simulations. We use a fixed temperature upper boundary condition which results in a system that is internally heated in the bulk and cooled at the top. We find that the flow speed, as measured by the Mach number, and turbulence, as measured by the Reynolds number, can be independently controlled by separately varying the characteristic temperature gradient from internal heating and the diffusivities. Two-dimensional simulations at a fixed Mach number (flow speed) demonstrate consistent power at low wave number as diffusivities are decreased. We observe convection where the velocity distribution is skewed towards cold, fast downflows and that the flow speed is related to the length scale and entropy gradient of the upper boundary where the downflows are driven. We additionally find a heat transport scaling law which is consistent with prior incompressible work. Published by the American Physical Society 2024

Mixup Mask Adaptation: Bridging the gap between input saliency and representations via attention mechanism in feature mixup

The inherent complexity and extensive architecture of deep neural networks often lead to overfitting, compromising their ability to generalize to new, unseen data. One of the regularization techniques, data augmentation, is now considered vital to alleviate this, and mixup, which blends pairs of images and labels, has proven effective in enhancing model generalization. Recently, incorporating saliency in mixups has shown performance gains by retaining salient regions in mixed results. While these methods have become mainstream at the input level, their applications at the feature level remain under-explored. Our observations indicate that outcomes from naive applications of input saliency-based methods did not consistently lead to enhancements in performance. In this paper, we attribute these observations primarily to two challenges: ‘Hard Boundary Issue’ and ‘Saliency Mismatch.’ The Hard Boundary Issue describes a situation where masks with distinct, sharp edges work well at the input level, but lead to unintended distortions in the deeper layers. The Saliency Mismatch points to the disparity between saliency masks generated from input images and the saliency of feature maps. To tackle these challenges, we present a novel method called ‘attention-based mixup mask adaptation’ (MMA). This approach employs an attention mechanism to effectively adapt mixup masks, which are designed to maximize saliency at the input level, for feature augmentation purposes. We reduce the Saliency Mismatch problem by incorporating the spatial significance of the feature map into the mixup mask. Additionally, we address the Hard Boundary Issue by applying softmax to smoothen the adjusted mixup mask. Through comprehensive experiments, we validate our observations and confirm the effectiveness of applying MMA to saliency-aware mixup approaches at the feature level, as evidenced by the performance improvements on multiple benchmarks and the robustness improvements against corruption and deformation.

Trapping and scattering of a multiflagellated bacterium by a hard surface.

Thiovulum majus, which is one of the fastest known bacteria, swims using hundreds of flagella. Unlike typical pusher cells, which swim in circular paths over hard surfaces, T. majus localize near hard boundaries by turning their flagella to exert a net force normal to the surface. To probe the torques that stabilize this hydrodynamically bound state, the trajectories of several thousand collisions between a T. majus cell and a wall of a quasi-two-dimensional microfluidic chamber are analyzed. Measuring the fraction of cells escaping the wall either to the left or to the right of the point of contact-and how this probability varies with incident angle and time spent in contact with the surface-maps the scattering dynamics onto a first passage problem. These measurements are compared to the prediction of a Fokker-Planck equationto fit the angular velocity of a cell in contact with a hard surface. This analysis reveals a bound state with a narrow basin of attraction in which cells orient their flagella normal to the surface. The escape angle predicted by matching these near field dynamics with the far-field hydrodynamics is consistent with observation. We discuss the significance of these results for the ecology of T. majus and their self-organization into active chiral crystals.

The Intimacy of Held Solidarity: A Joint Memoir of Activism

This work reflects the slow intimacy of solidarity over time. We uncover some of what we know and have known with examples from anti-apartheid activism and the women’s health movement in South Africa. Solidarity is an expression of knowing trust. Knowing that you are often on each other’s side and leaning towards justice, demanding a relationship of knowing in instances of vulnerability. Being held in solidarity is a gesture of engaging in a shared humanity that at times gives the comfort of familiarity. In spaces where there are shared feminist and political goals and objectives, we assume safety and belonging only to find this is not to be true. Acting in solidarity demands consideration and risk. In moving towards solidarity, one turns one’s body and face towards the other. Solidarity at times demands resistance and speaking truths to power. An engagement of solidarity is conscious and intentional. We reflect engaging with each other over four decades, lessons on listening deeply, hearing seasons of life’s celebrations, joys, questions, and struggles. In this process, questions surface, asking for more on consideration. Our reflection is of collective experiences of slow intimacy, of soft and hard boundaries, remembering and longing as we hold each other in solidarity. We use our individual voices to show specific angles to these collective experiences.

Physics-informed neural networks for transonic flow around a cylinder with high Reynolds number

The physics-informed neural network (PINN) method is extended to learn and predict compressible steady-state aerodynamic flows with a high Reynolds number. To better learn the thin boundary layer, the sampling distance function and hard boundary condition are explicitly introduced into the input and output layers of the deep neural network, respectively. A gradient weight factor is considered in the loss function to implement the PINN methods based on the Reynolds averaged Navier–Stokes (RANS) and Euler equations, respectively, denoted as PINN–RANS and PINN–Euler. Taking a transonic flow around a cylinder as an example, these PINN methods are first verified for the ability to learn complex flows and then are applied to predict the global flow based on a part of physical data. When predicting the global flow based on velocity data in local key regions, the PINN–RANS method can always accurately predict the global flow field including the boundary layer and wake, while the PINN–Euler method can accurately predict the inviscid region. When predicting the subsonic and transonic flows under different freestream Mach numbers (Ma∞= 0.3–0.7), the flow fields predicted by both methods avoid the inconsistency with the real physical phenomena of the pure data-driven method. The PINN–RANS method is insufficient in shock identification capabilities. Since the PINN–Euler method does not need the second derivative, the training time of PINN–Euler is only 1/3 times that of PINN–RANS at the same sampling point and deep neural network.

Observation of Free-Boundary-Induced Chiral Anomaly Bulk States in Elastic Twisted Kagome Metamaterials.

Chiral anomaly bulk states (CABSs) can be realized by choosing appropriate boundary conditions in a finite-size waveguide composed of two-dimensional Dirac semimetals, which have unidirectional and robust transport similar to that of valley edge states. CABSs use almost all available guiding space, which greatly improves the utilization of metamaterials. Here, free-boundary-induced CABSs in elastic twisted kagome metamaterials with C_{3v} symmetry are experimentally confirmed. The robust valley-locked transport and complete valley state conversion are experimentally observed. Importantly, the sign of the group velocity near the K and K^{'} points can be reversed by suspending masses at the boundary to manipulate the onsite potential. Moreover, CABSs are demonstrated in nanoelectromechanical phononic crystals by constructing an impedance-mismatched hard boundary. These results open new possibilities for designing more compact, space-efficient, and robust elastic wave macro- and microfunctional devices.

Haldane model with chiral edge states using a synthetic dimension

We explicitly show that the differences, with respect to the appearance of topological phases, between the traditional Haldane model, which utilizes a honeycomb lattice structure, and the Haldane model imbued onto a brick-wall lattice geometry are inconsequential. A proposal is then put forward to realize the Haldane model by exploiting the internal degrees of freedom of atoms as a synthetic dimension. This leads to a convenient platform for the investigation of chiral edge states due to the hard boundaries provided by the hyperfine manifold. We make some cursory comments on the effects of interactions in the system. Published by the American Physical Society 2024

Physical informed neural networks with soft and hard boundary constraints for solving advection-diffusion equations using Fourier expansions

Deep learning methods have gained considerable interest in the numerical solution of various partial differential equations (PDEs). One particular focus is physics-informed neural networks (PINN), which integrate physical principles into neural networks. This transforms the process of solving PDEs into optimization problems for neural networks. To address a collection of advection-diffusion equations (ADE) in a range of difficult circumstances, this paper proposes a novel network structure. This architecture integrates the solver, a multi-scale deep neural networks (MscaleDNN) utilized in the PINN method, with a hard constraint technique known as HCPINN. This method introduces a revised formulation of the desired solution for ADE by utilizing a loss function that incorporates the residuals of the governing equation and penalizes any deviations from the specified boundary and initial constraints. By surpassing the boundary constraints automatically, this method improves the accuracy and efficiency of the PINN technique. To address the “spectral bias” phenomenon in neural networks, a subnetwork structure of MscaleDNN and a Fourier-induced activation function are incorporated into the HCPINN, resulting in a hybrid approach called SFHCPINN. The effectiveness of SFHCPINN is demonstrated through various numerical experiments involving ADE in different dimensions. The numerical results indicate that SFHCPINN outperforms both standard PINN and its subnetwork version with Fourier feature embedding. It achieves remarkable accuracy and efficiency while effectively handling complex boundary conditions and high-frequency scenarios in ADE.