ABSTRACT. Direct numerical simulations of homogeneous and isotropic Hall magnetohydrodynamic (MHD) turbulence are performed to study the influence of the Hall term, which becomes important at the ion inertial length scale. Homogeneous and isotropic Hall MHD simulation data are analyzed with special attention to turbulence intermittency. Intermittency is investigated through the scaling properties of structure functions. It is found that intermittency at scales smaller than the ion inertial length is more monofractal-like than that at larger MHD scales. This behavior is related to the modification of the magnetic energy spectrum from the well-known $k^{-7/3}$ scaling, where $k$ denotes the isotropic wavenumber.

ABSTRACT. In this paper, we employ two-dimensional, electromagnetic particle-in-cell simulations to investigate the lower hybrid wave instability driven by energetic ions. We focus on the effect of the energetic-ion density on the instability and its non-linear development. We find that the density affects the acceleration of electrons and bulk ions. Electrons are more strongly accelerated than bulk ions for the high-density case, while both are comparably accelerated for the low-density case. We also find that whistler waves can be enhanced in the non-linear stage for the low-density case. These results clarify how the energetic-ion density controls particle acceleration and the non-linear excitation of whistler waves in two dimensions.

ABSTRACT. The dispersion curve of a magnetosonic wave is separated into high- and low-frequency modes due to the presence of two ion species. Previous nonlinear analyses on these modes were based on the quasi-neutrality condition. Although this condition is well valid when the parameter $\Omega_{e}/\omega_{pe}$ is much smaller unity, it breaks down when $\Omega_{e}/\omega_{pe}$ becomes of order unity, where $\Omega_{e}$ and $\omega_{pe}$ are the electron cyclotron and plasma frequencies, respectively. In this paper, we extend the theory to be valid for $\Omega_{e}/\omega_{pe} \sim 1$, using Poisson's equation instead of the quasi-neutrality approximation. The new theory reveals the dependence of nonlinear waves on $\Omega_{e}/\omega_{pe}$. The width of the nonlinear solitary pulse of the high-frequency mode increases with increasing $\Omega_{e}/\omega_{pe}$. The heavy-ion speed accelerated by this pulse also increases with $\Omega_{e}/\omega_{pe}$. This theoretical prediction is confirmed by electromagnetic particle simulations.

ABSTRACT. Hydraulic hammer forging is one of the important manufacturing technology to produce high stiffness products. The hydraulic hammer forging utilizes the blow energy, and the product is then produced through several blows. Therefore, the determination of the energy distribution is important. In addition, to produce complex product shape, multi-stage forging is widely used. This study presents a method to determine the optimal energy distribution as well as the optimal intermediate die shape in the hydraulic hammer forging. Intermediate die shape, which is used in multi-stage forging, is optimized by the basis vector method. The numerical simulation is so expensive that the SAO using RBF network is adopted to identify the Pareto-frontier between the total energy and the distance between dies. The total energy and the maximum forging load is significantly reduced.

ABSTRACT. Caliber rolling is widely used in material manufacturers due to good mechanical properties and high productivity. In the caliber rolling, it is important to make the tool life long and it is important to minimize the rolling force, and process parameter optimization is a major topic. The distribution of equivalent strain after rolling is also a key factor, and it is preferable to distribute the effective strain as much as possible for avoiding the risk of void. The numerical simulation in caliber rolling is so expensive that sequential ap-proximate optimization (SAO) using radial basis function (RBF) network is adopted. It is found though the numerical result that the rolling force and the distribution of equivalent plastic strain are reduced.

ABSTRACT. Process parameters in plastic injection molding (PIM) such as the melt temperature, the packing pressure, and the cooling time have a significant influence on product quality and productivity. In thick part, the volume shrinkage and cycle time are critical factors for dimensional accuracy and production efficiency. In this study, the variable packing pressure profile as well as other process parameters are optimized so as to minimize the volume shrinkage and cycle time. It is found the numerical result that the volume shrinkage and the cycle time are reduced, compare with ones using the conventional process parameters.

ABSTRACT. This study proposes a scout search mechanism for the Covariance Matrix Adaptation Evolu-tion Strategy (CMA-ES) to improve exploration in topology optimization. The proposed method performs an additional search along the evolution path when a predefined condition is satisfied and relocates the search distribution if a better solution is found. The method was evaluated using a two-dimensional magnetic shield topology optimization problem. Although the proposed method altered the search behavior of the conventional CMA-ES, it did not improve the final objective value under the experimental conditions. Future work includes improving the scout search strategy and applying the proposed method to IPM motor topology optimization.

ABSTRACT. This study proposes a lightweight 24-byte binary descriptor for self-localization of a small lunar lander using correlation responses from PCA-based crater detection. It encodes the relative response magnitudes around each feature point and compensates for rotation using a quantized direction index. Experiments compared the proposed method with existing meth ods under rotation, scale changes, image disturbances, and enlarged search regions. The method remained stable under rotation, achieved a RANSAC inlier ratio comparable to that of ORB under disturbances, and showed less performance degradation as the search region increased.

ABSTRACT. Image-based navigation, which matches craters detected from onboard images with a preloaded lunar map, enables accurate and autonomous lunar landing. JAXA's Smart Lander for Investigating Moon (SLIM) achieved pinpoint landing through crater detection based on principal component analysis (PCA). This paper proposes a triangle descriptor built from craters detected by a PCA-based feature detector with Z-scale normalization. We evaluate it using an F1 score independent of subsequent correspondence-estimation steps such as k-nearest neighbor (k-NN), the ratio test, and Random Sample Consensus (RANSAC). On JAXA's lunar image database, the descriptor achieves the highest matching performance among the compared methods, outperforming AKAZE and SURF under rotation (F1: 83.4%) and combined disturbances (58.7%). Comparison with a method whose detector differs only in Z-scale normalization shows that rotation robustness stems from the descriptor itself, while robustness to combined disturbances stems from Z-scale normalization at the detection stage.

ABSTRACT. We propose a lattice Monte Carlo model for simulating amyloid fibril formation from monomers to mature fibrils with high computational efficiency. Each protein is represented by a particle on a simple cubic lattice and undergoes transitions among monomer, intermediate, and fibril states. The model demonstrates that the aggregation kinetics and fibril morphology are determined by the monomer-to-intermediate transition probability and the intermediate-to-fibril transition probability. This simple model provides an efficient framework for studying the universal mechanism of amyloid fibril formation.

ABSTRACT. In this study, an existing Excel-based cesium (Cs) transport model was ported to Python. By maintaining the original calculation procedure while substantially enhancing com-putational throughput, we enabled long-term continuous operation analyses within real-istic execution times, thereby advancing the modeling of Cs behavior in negative ion sources. As a result of achieving an approximately 60 times speedup, long-term contin-uous operation simulations spanning dozens of cycles can now be executed in an ex-tremely short time. This has dramatically improved the efficiency of parameter analyses, such as optimizing Cs oven settings and initial Cs coverage to maintain optimal Cs monolayer thickness on the plasma grid (PG) electrode.

ABSTRACT. Electromagnetic wave propagation in magnetized plasmas plays a crucial role in plasma heating and diagnostics for magnetic confinement fusion. In recent years, this research has focused on numerical investigations of novel electromagnetic wave phenomena, including millimeter-wave vortex propagation in plasma, using the finite-difference time-domain (FDTD) method.

This talk presents recent progress in the numerical simulation of wave–plasma interactions. First, the propagation characteristics of millimeter-wave vortex beams carrying orbital angular momentum (OAM) are discussed. The simulation results demonstrate that vortex waves can penetrate into magnetized plasma under conditions where conventional plane waves experience cutoff, and that the penetrated power in plasma strongly depends on the topological charge of the optical vortex.

The second part introduces recent studies on O–X mode conversion in non-uniform magnetized plasmas. Numerical simulations reveal how the O–X mode conversion efficiency depends on the propagation characteristics of optical vortex beams in magnetized plasma. These results provide important insights into the efficient excitation of Electron Bernstein Waves (EBWs) for overdense plasma heating.

Finally, the talk briefly introduces ongoing research on advanced numerical methods for computational electromagnetics, including FDTD and particle-in-cell (PIC) to plasma physics. Future perspectives on applying electromagnetic wave engineering to fusion energy research will also be discussed.

ABSTRACT. The ion heating through the merging of spherical tokamak plasmas is investigated by means of particle simulations. From macroscopic viewpoints, the ion heating is com-posed of compressional and viscous terms. From microscopic viewpoints based on observation of velocity distributions, it is found that the compressional heating is not genuine heating, but effective heating, while the viscous heating is thought to be a different type of effective heating.

ABSTRACT. Anomalous radiation structures associated with plasma radiative collapse were vis-ualized in the Large Helical Device using a two-dimensional radiation measurement and an autoencoder (AE). By defining the non-collapsed state as normal and the state of progressing collapse as anomalous, the anomalous radiation structures were visualized as the reconstruction error of the AE. The visualization results in different magnetic field configurations suggest that these anomalous radiation structures are related to the distance between the plasma and the vacuum vessel wall.

ABSTRACT. Gyrokinetic simulations have become an indispensable tool for predicting turbulent transport in magnetically confined plasmas with high fidelity. However, their high computational cost limits their direct application to integrated transport simulations, optimization, and real-time prediction. Therefore, reduced models are essential. Most existing approaches, however, rely primarily on interpolation of simulation databases or experimental data and consequently have limited extrapolation capability beyond the parameter range used for model construction. In contrast, we propose a physics-based mathematical modeling framework based on the underlying ``solution space'' of turbulent dynamics. Our previous works demonstrated that the time-evolution of gyrokinetic turbulence is confined to a low-dimensional solution space manifold. In the present study, this concept is extended to multiple magnetic field configurations to investigate whether a common solution space representation can be constructed across different magnetic geometries. Gyrokinetic simulation data obtained under different magnetic configurations are analyzed and projected onto a common solution space function. The present analysis explores whether such a common representation may serve as the basis for a universal solution space model of plasma turbulent transport. Rather than reproducing simulation outputs through data interpolation, the proposed framework aims to identify the underlying solution space structure governing turbulent transport. If successful, such a framework would provide a physically interpretable foundation for reduced models with enhanced extrapolation capability across magnetic field configurations.

ABSTRACT. MHD simulations have been performed for understanding the behaviors of the pellet plasmoids in Large Helical Device (LHD) plasmas with an $m/n = 1/1$ magnetic island and without magnetic islands. In the LHD plasma with the island, the density increase due to ablation becomes maximum within the island, which is almost quantitatively consistent between the simulation results and the experimental data. The density distribution in the plasma without islands is extended closer to the magnetic axis than the one in the plasma with the island in the simulation results and the experimental data. The density increase induced by the ablation becomes lower when the thermal conduction is present. As a result, the simulation result is slightly closer to the experimental data in the LHD without islands, and it is further from the experimental data in the LHD with an island.

ABSTRACT. This paper presents an eddy current analysis of a 3-D twisted multi-conductor model using an integral equation method. To validate the proposed method, the same model was analyzed using the 3-D finite element method, and the resulting AC resistance and inductance were compared. The proposed method reproduced the AC resistance with an error within 1.14 %, while the inductance was approximately 2.8 % larger than the finite element result. The current and voltage excitation results gave almost the same AC resistance and inductance.

ABSTRACT. In this study, we develop a surrogate model for rapidly predicting electric field distributionsin indoor Wi-Fi environments and evaluate its prediction accuracy and computation time.Although electromagnetic wave propagation analysis using the FDTD method is highly accurate,numerical computation is required for each condition, so the computation time increases when alarge number of conditions are considered. Accordingly, for radio waves in the 2.4 GHz band,we vary the lengths of concrete walls placed within a two-dimensional square domain measuring5.0 [m] on each side and generate electric field distribution data for 2500 conditions. The surrogatemodel is trained using the generated electric field distribution data. In the surrogate model, the |Ez|distribution along y = 1.0 [m] is predicted in the first stage, and the two-dimensional |Ez| distributionis predicted in the second stage using electric field amplitude values at 25 points extractedfrom the predicted distribution as input features. The mean absolute error was 6.825519×10−2, andthe maximum absolute error was 2.552067. Furthermore, a comparison of the computation timesconfirmed that the surrogate model can predict the electric field distribution 53.03 times faster thananalysis using the FDTD method.

ABSTRACT. Alloy 600 is a Ni-based alloy used as a structural material in facilities such as nuclear power plants. When the alloy is sensitized, Cr depletion occurs near grain boundaries, and these regions become ferromagnetic, although the alloy is usually paramagnetic at room temperature. Since the grain-interior region also becomes ferromagnetic below approximately 110 K, ferromagnetic regions near grain boundaries are considered to coexist with ferromagnetic grain-interior regions at low temperatures. In this study, a two-phase magnetic model consisting of grain-interior and grain-boundary sites was constructed, and the temperature dependence of the magnetization was calculated using the classical Monte Carlo method based on a Heisenberg model. The model containing grain-boundary sites exhibited a higher Curie temperature than the model consisting only of grain-interior sites, in qualitative agreement with the experimental results. The grain-boundary sites governed the total magnetization above approximately 110 K, whereas the grain-interior contribution became dominant below this temperature, masking that of the grain boundaries.

ABSTRACT. We propose a numerical method for estimating the higher-order modes generated in a miter bend of a corrugated waveguide. When the HE11 mode is launched, the HE11 mode remains dominant at the output plane. However, distortion of the spatial field distribution indicates that higher-order modes are also excited. We therefore use a mode expansion technique to quantitatively determine the fraction of higher-order modes.

ABSTRACT. This study quantitatively compares tool manipulation in the traditional Japanese martial art "Bo-no-te" using optical motion capture to support skill transmission. We evaluated stick-tip trajectories and dynamic grip positions between an expert and an intermediate practitioner across two phases: continuous rotation without releasing the hands (Sec1) and rotation with hand regripping (Sec2). In Sec1, the expert held the stick near its ends and smoothly slid the hands farther toward the ends, generating a circular trajectory with nearly constant angular velocity and a trajectory extent 20%–30% larger than that of the intermediate practitioner. In contrast, the intermediate practitioner fixed the grip near the center, resulting in a small, distorted polygonal trajectory. In Sec2, the expert gripped near the ends of the stick while sliding along the central section, thereby achieving a large rotational radius. Conversely, the intermediate practitioner gripped near the center and slid toward the ends, resulting in a smaller rotational radius because the grip position serves as the practical center of rotation. These findings establish novel quantitative metrics based on dynamic tool–hand interaction for traditional skill preservation.

ABSTRACT. The transfer of skilled techniques has become an urgent issue in the construction industry because of workforce aging and labor shortages. This study aimed to identify motion characteristics relevant to skill transfer and to propose a group-average position–velocity representation for visualizing skilled tool-handling motions. First, a web-based questionnaire sur-vey was conducted with 1,000 construction workers to identify the information considered important for skill transfer. Work procedures, whole-body motions, and posture and tool handling were the most frequently selected items. Based on these findings, roller-painting motions were measured in 11 skilled painters and 11 participants without painting experience using an optical motion-capture system. Representative position–velocity patterns were obtained by averaging mean-centered velocity data according to the vertical position of the roller for each group. The resulting patterns indicated that position-dependent deceleration before direction reversal is a characteristic of skilled roller-painting motion.

ABSTRACT. Selective robotic maize harvesting can benefit from image-plane cues describing both cob orientation and semantically meaningful landmarks. We evaluate two two-dimensional representations: oriented bounding box (OBB) detection, which encodes in-plane rotation, and the estimation of three ordered keypoints corresponding to the cob base, a candidate grasp center, and the tip. Because the keypoints are ordered, they also identify which end of the cob is the base, which an OBB does not. We used 60 RGB images from a Japanese maize field dataset with a fixed 48/6/6 training, validation and test split; the test set contained eight annotated cob instances across six images. YOLO11n-OBB, YOLO26n-OBB, YOLO11n-Pose and YOLO26n-Pose were each trained five times with predefined seeds under otherwise fixed training settings. On the internal test set, mAP50–95, reported as mean ± sample standard deviation, was 0.574 ± 0.028 for YOLO11n-OBB and 0.552 ± 0.030 for YOLO26n-OBB, 0.450 ± 0.047 and 0.456 ± 0.061 for the box output of the Pose models (Pose-B), and 0.827 ± 0.014 and 0.789 ± 0.055 for their keypoint output (Pose-P). We then evaluated the same twenty checkpoints, without retraining or hyperparameter tuning, on a separately collected Benin subset of 25 RGB images containing 41 annotated cob instances. External mean mAP50–95 fell to 0.149 ± 0.066 and 0.161 ± 0.066 for OBB, 0.234 ± 0.039 and 0.122 ± 0.048 for Pose-B, and 0.513 ± 0.035 and 0.305 ± 0.082 for Pose-P, lower than the internal values in all six cases. The external OBB ranking depended on the metric, whereas YOLO11n had higher mean external values across all reported Pose-B and Pose-P metrics. The metrics obtained on the small internal test set should therefore not be interpreted as estimates of field performance. The study covers two-dimensional perception only. We did not evaluate three-dimensional localization or physical grasping and cutting.

ABSTRACT. This study examines an object recovery robot system that employs 3D image measurements. The declining birth rate and aging population have rendered cleaning public environments a significant concern for public health and welfare. We developed a method for article collection that combines map generation for mobile robots, 3D spatial measurement, and image-based object recognition. Preliminary experimental validation suggests the feasibility of this method for controlled indoor environments, although significant challenges remain for robust outdoor deployment.

ABSTRACT. In modern mechanical systems, heat is often generated through various processes, and accurately measuring the temperature changes caused by such heat generation remains technically challenging. To address this issue, the authors have developed a temperature identification method applicable to heat sources by combining the fictitious domain finite element method and the adjoint variable method. However, the influence of the thermophysical properties of the heated material on the identification accuracy has not yet been sufficiently investigated. Since various materials are used in practical engineering applications, clarifying this influence is an important issue. Therefore, the objective of this study is to investigate the effect of thermophysical properties on the proposed temperature identification method and to evaluate its applicability to materials with different thermal characteristics. The results showed that the heat source temperature was accurately identified for materials with high thermal diffusivity, such as copper. In contrast, for materials with low thermal diffusivity, such as titanium, discrepancies appeared in the latter stage of the analysis. These findings indicate that the proposed temperature identification method is effective for materials with relatively high thermal diffusivity, while further improvement is required for accurate application to materials with low thermal diffusivity.

ABSTRACT. Identification of the shape, position and scale of underground cavities is important for implementing a countermeasure against the ground settlement induced by underground cavities. The commonly used geophysical methods for subsurface cavity detection methods often require dedicated survey equipment and on-site measurement and are therefore not suited for large-scale and continuous monitoring. This study proposes an inverse analysis method for cavity shape identification using surface settlement data, which can be obtained more easily than conventional geophysical survey data. The proposed framework combines finite element analysis of ground consolidation equation with sensitivity analysis based on the adjoint variable method. In addition, a parametric cavity shape update method is introduced to constrain shape updates and improve estimation accuracy. This paper investigates the influence of the initial cavity position on the identification results through numerical experiments.

ABSTRACT. Turbulence models often assume spatial locality, but in complex flows, their performance may be insufficient. Although turbulence is a dissipative system and its attractor exhibits low-dimensional properties, it is difficult to derive finite-dimensional equations that hold on the attractor. In this study, we develop a data-driven method for constructing high-precision time-evolution models on the attractor while maintaining computational efficiency by constraining each degree of freedom to local interactions. Here, "locality" is not limited to space but generally refers to the locality of interactions among state variables. Therefore, in the sense of minimizing model error, the model includes optimal long-range interactions in space. In this paper, due to the ease of optimization, we address the Kuramoto-Shivashinsky equation, a spatially one-dimensional dissipative system; however, this modeling method is similarly applicable to the Navier-Stokes equations.

ABSTRACT. Dense manual geological annotation is often costly or unavailable for seismic interface interpretation, particularly when large numbers of SEG-Y sections must be processed. This paper presents an unsupervised pseudo-label generation workflow for training a standard U-Net without manual annotations. Coarse seismic labels are first generated via KMeans clustering, while a simplified PINN-style component estimates a smooth velocity trend and derives auxiliary pseudo labels. These automatically generated cues are then used to train a dual-label U-Net for interface identification. On a multi-layered synthetic section with non-flat interfaces, the proposed workflow reduces boundary MAE from 23.46 to 12.17 pixels and improves layer mIoU from 0.777 to 0.877 relative to KMeans. The results suggest that velocity-trend-guided pseudo-label refinement can provide useful weak supervision for seismic interface identification.

ABSTRACT. Vortex-sheet dynamics is governed by singular integral operators and is commonly reg-ularized by vortex-blob methods. In many shear-layer configurations, however, vortical motion is not produced by a single interface alone but by the interaction of multiple sheets. This study extends a hyperfunction-based interpretation of the periodic Birkhoff-Rott equation to a two-layer vortex-sheet system. The coupled model is de-composed into self-induced terms, which retain the principal-value singularity of each sheet, and inter-layer terms, which describe the mutual induction between separated sheets. Based on Sato’s hyperfunctions, the singular self terms are reformulated through boundary values of analytic functions and contour-integral representations in the com-plexified Lagrangian parameter plane. The inter-layer terms are then incorporated as regular coupling operators, yielding a theoretical framework for two-layer vortex-sheet analysis without introducing a blob-core parameter directly into the singular kernel. This paper presents the formulation and its mathematical structure as a preliminary step toward numerical simulation. Time-evolution results and quantitative comparisons are left for future work.

ABSTRACT. Dynamic derivatives are essential for aircraft stability and control analysis, but their estimation by forced-oscillation experiments or full-order computational fluid dynamics (CFD) can be costly. This paper presents a non-intrusive reduced-order modeling approach based on operator inference for predicting the dynamic derivatives of an oscillating S002 airfoil under low-speed conditions. The method constructs an arbitrary Lagrangian-Eulerian (ALE) reduced-order model from CFD snapshot data and extracts the pitching-moment derivatives by Fourier analysis. The resulting model reduced the state dimension from n = 818,766 to r = 8 while retaining 99.98% of the snapshot energy. Over nine tested reduced frequencies, the ROM achieved a mean relative error of 1.56% for Cmα and a range-normalized RMSE of 2.22% for Cmα̇ + Cmq.

ABSTRACT. The air environment in hospital rooms is an important factor affecting patient comfort and well-being. Effective assessment of hospital room air environments requires both room-scale indicators, which capture overall thermal and ventilation conditions, and patient-level indicators, which reflect local exposure around the bed. In this study, we introduce a composite score based on four indicators: PMV and circulation as room-scale indicators, and draft and odor perception as patient-level indicators. Using indoor air environment data obtained by CFD analysis, we evaluate hospital room airflow conditions and perform design optimization with a surrogate model for air-conditioning parameters, including supply air velocity, outlet angle, and installation position. The results show that the proposed score enables integrated assessment of hospital room air environments and reveals that the dominant design parameters vary depending on airflow direction and the presence of curtains.

ABSTRACT. Physics-informed neural networks (PINNs) use automatic differentiation to impose differential-equation residuals, but good agreement in function values does not necessarily imply accurate derivatives. This paper examines derivative fidelity as a diagnostic issue for PINN reliability using one-dimensional benchmarks. Multilayer perceptrons are trained only on function values for sin(x) and exp(x), while second derivatives obtained by automatic differentiation are evaluated separately. Additional tests over training-point density, activation functions, endpoint-dense evaluation, and both RMSE and maximum-error diagnostics are conducted. The results show that visually accurate function approximation can coexist with substantially larger second-derivative errors, especially near high-curvature boundary regions. These findings suggest that PINN validation should distinguish value accuracy from derivative and residual reliability.

ABSTRACT. Topology optimization is gaining attention as an effective means of achieving structural weight reduction and improving design efficiency. This study proposes a topology optimization method aimed at minimizing acceleration for transient vibration problems in mechanical structures. The objective is to establish an automated and general purpose optimization method for structural design where vibration reduction is required, particularly for automotive components. Specifically, vibration acceleration-which directly affects product comfort-is defined as the objective function and derive the optimal material distribution within the structure.

ABSTRACT. X-ray phase imaging provides high sensitivity to light-element specimens, but phase retrieval and tomographic reconstruction must be solved as inverse problems. In high-speed or low-dose acquisition, incomplete projections, noise, and detector nonuniformity degrade reconstruction. We investigate physics-guided diffusion reconstruction, in which an unconditional denoising diffusion probabilistic model (DDPM) trained on clean CT images serves as an image prior and an X-ray forward model is imposed at inference time. This separation permits reuse of a trained prior across measurement conditions without retraining. We compare diffusion posterior sampling (DPS), which incorporates a likelihood-gradient update, with a DDPM–MAP update that explicitly corrects the clean estimate using a data-fidelity term. Simulations of low-dose CT and ring artifacts demonstrate image-quality improvement and the influence of update rules. The framework is intended for extension to X-ray phase-imaging inverse problems.

ABSTRACT. The clothing microclimate strongly affects thermal comfort and is crucial for apparel design and comfort evaluation. However, direct measurement is difficult due to narrow internal spaces, restricted sensor placement, and body movement interference. Therefore, this study attempts to estimate internal clothing temperature and humidity from clothing surface temperature and environmental conditions. One-dimensional transient coupled heat and moisture transfer simulations based on the finite volume method, using heat transfer coefficients identified from experimental data were conducted under 1,440 conditions, extracting quasi-steady values at 2 h as training data. A fully connected mul-tilayer perceptron trained on this dataset successfully predicted internal states of unseen test data with high accuracy, suggesting the utility of the surrogate model.

ABSTRACT. In this study, a physics-informed neural network approach based on Kirchhoff's boundary integral equation of the magnetic field for interior-region problems is developed. This approach is applied to the perturbative modeling of electrodynamic interaction between a relativistic charged particle beam and accelerator beam pipes with finite wall conductivity and verified in comparison with the available analytical solution.

ABSTRACT. In this paper, we apply an amplitude-based analysis using Multivariate Empirical Mode Decomposition (MEMD) to quantify the kinetic chain from pitching motion data. The kinetic chain transmits momentum from proximal to distal segments—legs, trunk, throwing arm—and is fundamental to pitching, yet hard to quantify from motion-capture joint angles. We decompose each pitch with MEMD and aggregate the dominant oscillatory mode’s amplitude into five kinetic chains. Over 30 pitches we consistently observe a clear proximal-to-distal peak order and an average amplification of about 4.2 times, quantitatively demonstrating sequential force transmission in pitching.

ABSTRACT. The effect of a prescribed through-thickness wall-stiffness profile on the apparent modulus of a closed-cell foam is difficult to infer from density alone. We combine body-fitted finite-element analysis with visualization of the resulting parameterized response. A foam generated from a neighbor distance field is discretized by a body-fitted tetrahedral mesh. Through-thickness stiffness profiles are then imposed on the same fixed mesh, allowing graded designs to be evaluated without re-meshing. For a foam mesh with 1.07 x 10^6 tetrahedra, sweeping the two grading amplitudes maps the apparent modulus 𝐸∗(𝑎1,𝑎2) under laterally constrained platen compression. Over the sampled amplitude interval, increasing 𝑎1 alone decreases 𝐸∗/𝐸0 from 0.231 to 0.182, whereas increasing 𝑎2 alone decreases it to 0.223; these profiles are not normalized to equal stiffness contrast or material cost. Mid-slab von Mises stress fields provide a qualitative comparison of representative designs.

ABSTRACT. 本セッションでは、高校生、高専生、大学生などの初学者を対象に、研究活動や探究活動の成果を論文としてまとめるための基礎を解説します。科学的方法と、論文の標準的な構成であるIMRADの枠組みについて学ぶとともに、円滑に執筆を進めるための生成AIの効果的な活用方法についても取り上げます。

参加登録・詳細はこちら：https://easychair.org/smart-program/JSST2026/writing_workshop.html

This session introduces beginners, including high school, technical college, and university students, to the fundamentals of turning the outcomes of research and inquiry-based learning into research papers. It covers the scientific method, the IMRAD framework as the standard structure of research papers, and the effective use of generative AI to support the writing process.

Registration and details: https://easychair.org/smart-program/JSST2026/writing_workshop.html