Computational Physics Research Articles (Page 1)

This manuscript presents and proves a reciprocity relation involving the Fourier transforms of a pair of square-integrable functions, expressed as a bilinear map. This reciprocity relation reveals a deep symmetry between the time (or spatial) and frequency domains. We explore its implications in theoretical and applied contexts such as signal processing, quantum mechanics, and computational physics. Additionally, we discuss the role of this relation in the bilinear nature of Fourier analysis.

Machine learning for modelling unstructured grid data in computational physics: A review

Etaloning artifacts introduce notable distortions in spectroscopic data, complicating downstream analysis and interpretation. We present an inverse modeling framework that integrates computational physics with deep learning to address this challenge. Our approach employs a two-phase transfer learning strategy: pretraining on over 30,000 simulated spectra generated using the transfer matrix method and fine-tuning on real experimental data. This extensive simulated data set enhances the model's ability to generalize across different sensor designs, significantly improving robustness and accuracy. Rigorous cross-validation across multiple devices demonstrates that the transfer learning approach reduces etaloning-induced distortions by up to 70%, ensuring substantial spectral accuracy and interpretability improvements. This study sets a new standard for achieving reliable spectral data by combining correction procedures with physics simulations.

To understand the dynamics of quantum many-body systems, it is essential to study excited eigenstates. While tensor network states have become a standard tool for computing ground states in computational many-body physics, obtaining accurate excited eigenstates remains a significant challenge. In this work, we develop an approach that combines the inexact Lanczos method, which is designed for efficient computations of excited states, with tree tensor network states (TTNSs). We demonstrate our approach by computing excited vibrational states for three challenging problems: (1) 122 states in two different energy intervals of acetonitrile (12-dimensional), (2) Fermi resonance states of the fluxional Zundel ion (15-dimensional), and (3) selected excited states of the fluxional and very correlated Eigen ion (33-dimensional). The proposed TTNS inexact Lanczos method is directly applicable to other quantum many-body systems.

Background This study presents an interdisciplinary pedagogical approach aimed at contextualizing the teaching of classical mechanics through the computational analysis of Break Dance movements. Situated within a constructivist framework, the research explores how culturally embedded practices—specifically urban dance—can serve as a medium for fostering conceptual understanding of physics in non-formal educational settings. By leveraging the artistic and kinesthetic dimensions of Break Dance, the study seeks to bridge the gap between abstract scientific content and learners’ lived experiences. Method A mixed-method, exploratory design was employed with a purposive sample of ten dancers (aged 13–30) affiliated with a community-based urban dance school in Bogotá, Colombia. Over the course of six three-hour sessions, participants engaged in movement analysis using Tracker video software, supported by pre- and post-intervention semi-structured interviews. The research design incorporated thematic analysis to interpret qualitative data, complemented by the kinematic study of body movement parameters such as angular velocity and center of mass. Results Findings indicate a marked cognitive shift among participants from intuitive and superficial conceptions to a more technical and applied understanding of physics principles. The integration of computational tools allowed dancers to visualize and internalize biomechanical variables relevant to their performance. Participants reported enhanced bodily control, injury prevention, and aesthetic execution, alongside increased motivation and collaborative learning. Conclusions The study concludes that embedding scientific content within culturally relevant, embodied practices—mediated by educational technologies—can significantly enhance learning outcomes in physics. The use of Tracker software not only demystified abstract concepts but also redefined physics as accessible and contextually meaningful. These results underscore the pedagogical potential of transdisciplinary, arts-integrated methodologies to foster inclusive, situated, and cognitively rich science education in non-traditional environments.

The rising energy demands of conventional AI systems underscorethe need for efficient computing technologies, such as brain-inspiredcomputing. Physical reservoir computing (PRC), leveraging the nonlineardynamics of physical systems for information processing, has emergedas a promising approach for neuromorphic computing. However, currentPRC systems are constrained by narrow responsive time scales and limitedperformance. To address these challenges, an ion-gel/graphene electricdouble layer (EDL) transistor-based ion-gating reservoir (IGR) wasdeveloped. This IGR achieves a highly tunable and ultrawide time-scaleresponse through the coexistence of fast EDL dynamics at the ion-gel/grapheneinterface and slower molecular adsorption dynamics on the graphenesurface. Consequently, the system demonstrates an exceptionally broadresponsive range, from 1 MHz to 20 Hz, while maintaining a high informationprocessing capacity and adaptability across multiple time scales.The IGR achieved deep learning (DL)-level accuracy in chaotic timeseries prediction tasks while reducing computational resource requirementsto 1/100 of those needed by DL. Principal component analysis revealsthe IGR’s superior performance stems from its high-dimensionality,driven by the ultrawideband responses of the EDL along with the ambipolarbehavior of graphene. The proposed IGR represents a significant stepforward in providing low-power, high-performance computing solutions,particularly for resource-constrained edge environments.

ABSTRACTElementary teachers feel underprepared to teach mathematics in an integrated manner, explaining the critical need to design professional development experiences centering on Science, Technology, Engineering, and Mathematics (STEM) in physical computing and mathematics activities. Not fully explained is the impact these professional development experiences have on teachers' mathematics understandings and how physical computing plays a role in changes to their understandings. In this study, we sought to examine such impact more fully, extending literature that examines individuals' concept definition (Tall and Vinner 1981) for “doing mathematics” (Schoenfeld 1992) situatively with a physical computing professional development experience. Results indicate the importance context and situated learning have when promoting changes in teachers' mathematics understandings. Nuances of teachers' mathematization explained some of the teachers' understanding development, suggesting further scholarship would benefit from examining how mathematization can frame teachers' experiences with physical computing projects.

Recent progress in magneto-ionics offers exciting potential to leverage its energy efficiency for implementing physical reservoir computing (PRC). In this work, we experimentally demonstrate the classification of temporal data using a perpendicularly magnetized magneto-ionic (MI) heterostructure. The device was specifically engineered to induce nonlinear ion migration dynamics, which in turn imparted nonlinearity and short-term memory (STM) to the magnetization. These key features for enabling reservoir computing were investigated, and the role of the ion migration mechanism, along with its history-dependent influence on STM, was explained. These attributes were utilized to distinguish between sine and square waveforms within a randomly distributed set of pulses. Additionally, two important performance metrics─STM and parity check capacity ─were quantified, yielding promising values of 1.44 and 2 for 24 virtual nodes, respectively, comparable to those of other state-of-the-art reservoirs. Our work paves the way for exploiting the relaxation dynamics of solid-state MI platforms and developing energy-efficient MI reservoir computing devices.

Physical computing can be used to teach students computer science concepts in a project-based learning environment. However, this approach can be challenging for novices, as it requires mastering electronics and coding skills simultaneously. This paper presents Boxy Board, a custom-designed physical computing kit aimed at novice students. Rooted in constructivist methodologies and cognitive load theory, Boxy Board aims to reduce cognitive barriers by removing electronic components (e.g., breadboards) while gradually introducing complexity, scaffolding electronic and coding concepts through a progressive disclosure approach. This paper outlines the design, prototyping, and implementation process of Boxy Board. We conducted a usability evaluation with sixth-grade students, using pre- and post- questionnaires and interviews. The study shows that Boxy Board is usable by children and that they enjoyed their experience with it. We provide recommendations for improvements, and we establish that it is possible to implement progressive disclosure successfully in a physical computing educational setting.

Porous materials such as zeolites and Metal-Organic Frameworks are widely used for molecular separations based on adsorption and enthalpy/entropy characteristics. Ideal adsorption solution theory (IAST) predicts mixture adsorption behaviour on the basis of pure component isotherms of adsorbents in porous media. Mixture data at all mole fractions are required for breakthrough simulations. The use of IAST avoids the expensive computations of mixtures with Monte Carlo methods. Matching outcomes from computational physics studies to experimentally measurable properties is the foundation of the materials design pipeline. Here, we report the regression of an Invertible Autoencoder (IAE) for the forward and backward mapping of pure and mixture isotherms. The invertible autoencoder is defined as a soft-invertible neural network, which can be used as mapping function. Pure component isotherms are modelled using a 3-site Langmuir-Freundlich model, with a broad range of equilibrium pressure and heterogeneity factors. A synthetic dataset is generated from pure component isotherms and mixture isotherms calculated with RUPTURA. The IAE predicts pure and mixture isotherms with high precision over a large fugacity range, for up to 6 components and 3-site isotherms. This work contributes to inverting the full design pipeline from physical gas separation to adsorbate design, enabling property-guided materials discovery.

Heightened sensory sensitivity and subsequent engagement among individuals with ASD.

ABSTRACT Current trends show an increase in population and needs for the STEM labor force. This aligns with the increase in student enrollment in STEM fields of study such as computer science, mathematics, and physics. Centered around student pathways to transfer or start a career, community colleges seek to provide students with resources to enable success in a sought out STEM field, often through support and engagement from faculty and financial initiatives. The Start SMART – Self Motivated, Academic, Reflective, Talented project, funded by a National Science Foundation S-STEM grant in 2018, provided Mid-Atlantic Community College with the opportunity to offer STEM-majoring students monetary, academic, and personal support. Those involved in the grant sought to improve upon previous iterations with increased variability in monthly activities and regular communication regarding student needs as well as tracking students’ needs while taking math courses. The research component of this project examined the influence of co-curricular activities on the development of self-regulated learners with specific attention to affective attributes, including grit, growth mind-set, and goal orientation, and the contributions of these attributes to student success. Inherent in this model was the need for the student to develop and practice metacognitive monitoring. The periodic monitoring of mind-set was crucial and movement toward self-regulated learning was possible through the students’ reflections. Students reflected on their experiences in the program to identify the impact of the program and its initiatives on student outcomes.

.SignificanceThe effects of optogenetic stimulation (OS) on in vitro neural network behavior were studied through a reservoir computing-based obstacle avoidance task, revealing its impact on the task-processing capabilities of the network. Furthermore, it is demonstrated that a minimal output of signals from 15 neurons in the network is sufficient to achieve stable task control, with a success rate exceeding 95%. The optogenetically enhanced biological reservoir computing frame could find applications in neuro-robotic control and brain-inspired intelligence.AimWe aim to utilize optogenetically controlled in vitro neural networks and the first-order reduced and controlled error (FORCE) learning algorithm to achieve obstacle avoidance in neuro-robotic systems.ApproachWe presented an all-optical biological reservoir computing framework that leverages optogenetics and calcium imaging to precisely regulate and record neuronal activities. A closed-loop system was developed incorporating the FORCE learning algorithm, which guided a virtual car through obstacle avoidance tasks.ResultsThe system demonstrated high accuracy and efficiency in navigating obstacles, achieving optimal performance after of training. OS significantly improved the obstacle avoidance success rate, enhancing the system’s adaptability and accuracy.ConclusionsThe results highlight the potential of optogenetically controlled biological neural networks in neuro-robotic systems, showcasing their capability to achieve accurate and efficient obstacle avoidance through physical reservoir computing.

Thermal system heat transfer simulation is essential in many fields, like electronics, aerospace, and energy. In real- time applications, such as interactive design, digital twins, or online control systems, traditional techniques like Finite Element Method (FEM) or Computational Fluid Dynamics (CFD) are both accurate and expensive, requiring extensive computation. These methods are precisely why we propose a thermal phenomenon surrogate modeling framework utilizing Generative Adversarial Networks (GANs). Our approach utilizes Conditional GAN (GCN) architecture where Generators output temperature fields based on the specified input parameters, pre-defined boundaries, conditions, material tech characteristics, and geometry. The Discriminator Network improves the fidelity and realism of the retrieved data by identifying whether the model-derived outputs are truly surpassing high-fidelity simulated outputs—model verification. We assembled a detailed dataset from high-res numerical simulations to train and validate the surrogate model. Our findings proved that a GAN-based surrogate model achieves spend value MSE and SSIM while outperforming conventional solvers by orders of magnitude, generating millisecond results, and maintaining high accuracy relative to ground truth simulations. Further, the model's robustness to unseen boundary conditions and geometric configurations enhances its versatility for various thermal analysis scenarios. This research underscores the transformative possibilities for real-time simulation in computational physics that deep generative models can provide. The proposed surrogate model strikes a viable compromise between efficiency and accuracy, expanding its scope for practical deployment in real-time thermal monitoring, design cycle optimization, and physics- based AI frameworks.

In-materia computing, harnessing material complexity for energy-efficient computation, drives breakthroughs in constructing physical reservoir computing (PRC), a promising paradigm for energy-efficient handling of dynamic temporal tasks. However, the integration of PRC components into complementary metal-oxide-semiconductor (CMOS)-compatible and very large-scale integration (VLSI)-scalable platforms remains challenging, particularly in two-terminal devices that utilize exotic material systems. Herein, the integration of hafnia-based hybrid ferroelectric-ionic field-effect transistors (FETs) is reported for in-materia PRC with all-FET structures. Hybrid FETs with dual long-term polarization switching and short-term ionic switching functionality are integrated into PRC via wafer-scale atomic layer deposition on a single wafer, guaranteeing CMOS compatibility and VLSI scalability with the deposition technique and materials in modern microelectronics. The proposed PRC system effectively processes multimodal biosignals, including electroencephalogram, electrocardiogram, and electromyogram, demonstrating superior performance compared to conventional two-terminal device-based systems by enabling adaptive temporal dynamics and tunable memory characteristics. These results pave the way for hardware-implemented dynamic neural networks that are highly energy- and area-efficient, thus advancing practical edge AI applications in healthcare and real-time signal processing.

Physical computing has emerged as a powerful tool for performing intelligent tasks directly in the mechanical domain of functional materials and robots, reducing our reliance on the more traditional CMOS computers. However, no systematic study explains how mechanical design can influence physical computing performance. This study sheds insights into this question by repurposing an origami-inspired modular robotic manipulator into an adaptive physical reservoir and systematically evaluating its computing capacity with different physical configurations, input setups, and computing tasks. By challenging this adaptive reservoir computer to complete the classical NARMA benchmark tasks, this study shows that its time series emulation performance directly correlates with the Peak Similarity Index (PSI), which quantifies the frequency spectrum correlation between the target output and reservoir dynamics. The adaptive reservoir also demonstrates perception capabilities, accurately extracting its payload weight and orientation information from the intrinsic dynamics. Importantly, such information extraction capability can be measured by the spatial correlation between nodal dynamics within the reservoir body. Finally, by integrating shape memory alloy (SMA) actuation, this study demonstrates how to exploit such computing power embodied in the physical body for practical, robotic operations. This study provides a strategic framework for harvesting computing power from soft robots and functional materials, demonstrating how design parameters and input selection can be configured based on computing task requirements. Extending this framework to bio-inspired adaptive materials, prosthetics, and self-adaptive soft robotic systems can enable next-generation embodied intelligence, where the physical structure can compute and interact with its digitalcounterparts.

Physical reservoir computing (PRC) is an innovative computational paradigm that leverages intrinsic nonlinearity of physical systems to efficiently perform complex tasks. It is discovered that the intrinsically disordered domain structure in multiferroic YMnO3 provides significant nonlinearity, making it a promising candidate for robust PRC with tuneability and functionality at high temperatures. This work explores the potential of YMnO3 single crystals for PRC. PRC performance of YMnO3 is systematically evaluated by analysing its nonlinear responses, phase shifts, and high dimensionality through benchmark tasks such as waveform generation (WG), memory capacity (MC), and second-order nonlinear autoregressive moving average (NARMA2) time-series prediction. This results demonstrate that YMnO3 single crystals exhibit superior performance in these tasks, achieving high accuracy and low power consumption (≈1.77 µW and ≈0.02 nW/domain). These crystals also performed well in practical application of low-power speech recognition. These findings establish YMnO3 as a viable platform for next-generation PRC technologies, addressing critical challenges in the field.

Physical neural networks (PNNs) are a class of neural-like networks that make use of analogue physical systems to perform computations. Although at present confined to small-scale laboratory demonstrations, PNNs could one day transform how artificial intelligence (AI) calculations are performed. Could we train AI models many orders of magnitude larger than present ones? Could we perform model inference locally and privately on edge devices? Research over the past few years has shown that the answer to these questions is probably "yes, with enough research". Because PNNs can make use of analogue physical computations more directly, flexibly and opportunistically than traditional computing hardware, they could change what is possible and practical for AI systems. To do this, however, will require notable progress, rethinking both how AI models work and how they are trained-primarily by considering the problems through the constraints of the underlying hardware physics. To train PNNs, backpropagation-based and backpropagation-free approaches are now being explored. These methods have various trade-offs and, so far, no method has been shown to scale to large models with the same performance as the backpropagation algorithm widely used in deep learning today. However, this challenge has been rapidly changing and a diverse ecosystem of training techniques provides clues for how PNNs may one day be used to create both more efficient and larger-scale realizations of present-scale AI models.

The output-side behaviors of typical digital computing systems, such as simulated neural networks, are generally unaffected by the act of observation; however, this is not the case for the burgeoning field of physical reservoir computers (PRCs). Observer dynamics can limit or modify the natural state information of a PRC in many ways, and among the most common is the conversion from analog to digital data needed for calculations. Here, to aid in the development of novel PRCs, we investigate the effects of bounded, quantized observations on systems' natural computational abilities. By utilizing a classical reservoir computing (RC) (an echo-state network) and some PRCs (a pneumatic artificial muscle and a soft tentacle), we show that observed state quantization effectively converts a system's natural memory into higher-order, nonlinear dynamics. Furthermore, this same effect can assist in reducing detectable system errors in the presence of noise. We demonstrate how these effects, imposed only through output-end observations, can improve timer task robustness, target different computational task types, and even encode the chaotic dynamics of a Lorenz attractor in a simple linear RC in a closed loop.

The Lanczos algorithm, introduced by Cornelius Lanczos, has been known for a long time and is widely used in computational physics. While often employed to approximate extreme eigenvalues and eigenvectors of an operator, interest in the sequence of basis vectors produced by the algorithm has been recently increased in the context of Krylov complexity. Although it is generally accepted and partially proven that the procedure is numerically stable for approximating the eigenvalues, there are numerical problems when investigating the Krylov basis constructed via the Lanczos procedure. In this paper, we show that the loss of orthogonality and the attempt of reorthogonalization fall short of understanding and addressing the problem. Instead, the sequence of numerical Lanczos vectors in finite-precision arithmetic escapes the true vector space spanned by the exact Lanczos vectors. This poses a real threat to the interpretation in view of the operator growth hypothesis.