We introduce and experimentally demonstrate an approach for selective mode excitation in non-Hermitian resonant systems using incoherent light. This method eliminates the need for precise phase control that is often required in coherent excitation schemes. Using this technique on a silicon photonic platform with coupled ring resonators, we successfully excite the topological edge state of a non-Hermitian Su-Schrieffer-Heeger model. Our work shows that incoherence-assisted excitation is a robust and passive strategy for topological state preparation, which broadens the scope of non-Hermitian topological photonics, thereby providing a practical and experimentally viable tool for selective mode excitation.

Quantum kernel methods are one of the most explored approaches to quantum machine learning. However, the structural properties and inductive bias of quantum kernels are not fully understood. In this work, we introduce the notion of entangled tensor kernels—a generalization of product kernels from classical kernel theory—and show that all embedding quantum kernels can be understood as an entangled tensor kernel. We discuss how this perspective allows one to gain insights into both the unique inductive bias of quantum kernels and potential methods for their dequantization.

Pulsed-eddy-current testing (PECT) is a highly efficient active detection technique used in nondestructive testing of conductive structures, transient electromagnetic exploration for mineral resources, and the detection of unexploded ordnance. To directly measure the induced magnetic field and improve the sensitivity of PECT systems, we use a free-induction-decay (FID) atomic magnetometer instead of pickup coils, enabling improved detection of weak low-frequency signals. The FID magnetometer employs an enhanced-optical-pumping scheme and an instantaneous phase-retrieval method, achieving a noise floor of 0.45 $\mathrm{pT}/\sqrt{\mathrm{Hz}}$ and bandwidth of 4.2 kHz. The synchronized magnetic-pulse pumping scheme not only significantly amplifies the spin polarization of the atomic ensemble to strengthen the FID signal and improve sensitivity, but also induces eddy-current magnetic fields in nearby metallic conductors. Using instantaneous phase retrieval, we extract the pulsed-eddy-current magnetic fields from the FID signal and determine their characteristic time constant $\left(\tau \right)$. The FID magnetometer is validated for measuring pulsed-eddy-current magnetic fields generated by multiple metals, including copper, aluminum, zinc, and lead. This work demonstrates the potential of the FID magnetometer for applications in magnetic induction tomography and remote-field eddy-current testing.

We investigate the nonequilibrium steady state of a one-dimensional (i.e., mechanical degree of freedom) run-and-tumble model of left-right oriented Brownian particles with an explicit coupling to a chemical degree of freedom and a macroscopic opposing force. This model enables us to explore the fundamental relationships between the energy process driving individual entities, the emergence of collective behavior, and macroscopic forces in active matter. Sufficiently large alignment interactions can break the right-left symmetry, leading to a macroscopic collective alignment of the particles, or flocking transition. Using a decoupling mean-field approximation, we find that collective alignment requires both sufficiently large alignment interactions and lattice hopping. Further, this collective alignment does not require chemical-to-mechanical coupling and can occur solely through passive mechanical hopping. If chemical-to-mechanical coupling is present, it can lead to a collective drift in the direction of particle alignment even against an opposing mechanical force. Interestingly, in the absence of an opposing force, the particles spontaneously align in a random direction, but in the presence of a force, they tend to align “uphill,” enabling macroscopic work output. The many-body system is formulated in terms of creation and annihilation operators for identical classical particles, which enables us to calculate a range of observables by exploiting techniques from many-body quantum systems.

This study extends the moment-based nonlinear inverse modeling (nLIM) framework from the stationary to the cyclostationary (CS) regime, introducing CS-Colored-nLIM. The proposed method constructs a periodic, colored-noise-forced, quadratic stochastic system from cyclostationary input data. In this formulation, the active nonlinear components can be empirically determined through a sparse identification scheme specifically developed for colored-noise-forced systems. Application to climate prediction demonstrates that CS-Colored-nLIM achieves predictive performance comparable to modern machine learning and statistical models, while offering greater flexibility with moderate computational demand, making it potentially applicable to a broader range of complex systems beyond climate dynamics.

The nature of a low-field helicon plasma mode transition is examined using experiment and theory. Tests in a radio frequency plasma source reveal an abrupt transition between helicon and capacitive discharge modes that occurs at a critical magnetic field strength. To characterize the mode transition, a global model for low-field $m=0$ helicon plasmas is developed based on power absorption via Landau damping. A general scaling law for the critical field strength is derived from the model and experimentally validated using measurements of electron density and temperature. The assumption of power absorption via Landau damping in the model leads to a critical condition for the ratio of wave phase velocity and electron thermal velocity. Measurements of wave phase velocity and electron temperature while operating on mixtures of argon and ${\mathrm{N}}_2\mathrm{O}$ support the theoretical prediction and show that the mode transition depends on electron properties rather than on the ion and neutral composition. Agreement between theory and experiment shows that Landau damping is the primary power absorption mechanism in low-field, conducting boundary helicon plasmas.

Astrophysical and geophysical fluids commonly generate organized magnetic fields, despite having enormous magnetic Reynolds numbers $\mathrm{Rm}$ and abundant small-scale turbulence. Flow-induced dynamo action produces these fields, with the “kinematic dynamo problem” devoted to determining the rate at which a flow exponentially amplifies weak magnetic fields. However, previous studies on high-Rm kinematic dynamos have generated flows via imposed volumetric forcing or oscillatory boundary conditions. In this article, we investigate a system with three important attributes: realistic flow conditions, fast dynamo action (operational for $\mathrm{Rm}\to \infty$), and a subharmonic spatiotemporal structure. We show that unsteady Taylor-vortex flow, a regime observed in laboratory experiments, gives rise to fast dynamos with timescales and length scales twice those of the flow at high $\mathrm{Rm}$. By numerically integrating a Floquet system driven by periodic oscillations of Taylor vortices, we solve the kinematic dynamo problem up to $\mathrm{Rm}=3.2\times {10}^6$, calculating the dynamo's growth rate as a function of Rm and streamwise wave number. We find the onset of instability and compute finite-time Lyapunov exponents, which identify the regions of Lagrangian chaos required for fast dynamo action. To our knowledge, unsteady Taylor-vortex flow produces the most physically motivated fast dynamo to date.

We identify a fundamental pathology in the likelihood for time delay inference that challenges standard inference methods. By analyzing the likelihood for time delay inference with Gaussian process light curve models, we show that it generically develops a boundary-driven “W” shape with a global maximum at the true delay and gradual rises toward the edges of the observation window. This arises because time delay estimation is intrinsically extrapolative. In practice, global samplers such as nested sampling are steered toward spurious edge modes unless strict convergence criteria are adopted. We demonstrate this with simulations and show that the effect strengthens with higher data density over a fixed time span. To ensure convergence, we provide concrete guidance, notably increasing the number of live points. Further, we show that methods implicitly favoring small delays, for example, optimizers and local Markov chain Monte Carlo, induce a bias toward larger $H_0$. Our results clarify failure modes and offer practical remedies for robust fully Bayesian time delay inference.

Syntax is an aspect of human language that sorts linguistic structures hierarchically. It is summarized by dependency trees with words as nodes and edges reflecting syntactic subordination. By merging trees from several sentences, we obtain syntax graphs or networks, which display distinct shapes depending on whether the language capability is well formed, still developing, or pathological. Such graphs make syntactic capacity quantifiable at a systemic level. What is the structure of syntax networks during ontogeny in typically developing children? Do cognitively challenged children develop language through alternative routes? We quantify and portray typical development of syntax networks in Dutch, and find that children affected by Down syndrome, hearing impairment, and specific language impairment track the typical developmental path but halt along the way. Through a broad dataset and a thorough mathematical quantification we (1) confirm and refine a proposed sharp transition in language development, (2) correlate network traits with syntax maturation, and (3) quantify aspects that fall short in atypical development—suggesting potential diagnostic tools. We find grounds to hypothesize a gap in syntax maturation that typically developing children bridge over seamlessly, but which divides atypically developing children regardless of their condition. Our quantitative analysis enables a rigorous visualization of linguistic development trajectories, an old (yet mostly qualitative) theme in linguistics. Similar works should allow to test specific linguistic concepts (such as development trajectories) on solid quantitative grounds—as done here. Future efforts should generalize to other languages and clinical conditions, seeking patterns and universalities in language development.

The probabilistic information flow and natural computational capability of a system with two magnetic skyrmions at room temperature have been experimentally evaluated. Probabilistic behavior has attracted attention for its potential to enable unconventional computing paradigms that achieve ultralow-energy computation. However, information propagation and computation in such systems are more complex than in conventional computers, making their visualization essential.

In this study, a two-skyrmion system confined within a square potential well at thermal equilibrium was analyzed using information thermodynamics. Transfer entropy and the time derivative of mutual information were employed to investigate the information propagation speed, the absence of a Maxwell’s demon in thermal equilibrium, and the system’s non-Markovian properties. Furthermore, it was demonstrated that the system exhibits a small but finite computational capability for the nonlinear XOR operation, potentially linked to hidden information in the non-Markovian system.

We introduce a picture to describe and interpret waveguide-QED problems in the non-Markovian regime of long photonic retardation times resulting in delayed coherent feedback. The framework is based on an intuitive spatial decomposition of the waveguide into blocks. Among these, the block directly coupled to the atoms embodies an effective lossy multimode cavity leaking into the rest of the waveguide, in turn embodying an effective white-noise bath. The dynamics can be approximated by retaining only a finite number of cavity modes, which grows with the time delay. This description captures the atomic as well as the field’s dynamics, even with many excitations, in both emission and scattering processes. As an application, we show that the recently identified non-Markovian steady states can be understood by retaining very few or even only one cavity mode.

We develop a general theory of frustration-free free-fermion systems and derive the necessary and sufficient conditions for such Hamiltonians. Assuming locality and translation invariance, we find that any band touching between the valence and the conduction bands is always quadratic or softer, which rules out the possibility of describing Dirac and Weyl semimetals using frustration-free local Hamiltonians. We further construct a frustration-free free-fermion model on the honeycomb lattice and show that its density fluctuations acquire an anomalous gap originating from the diverging quantum metric associated with the quadratic band-touching points. Nevertheless, an $O(1/L^2)$ finite-size scaling of the charge-neutral excitation gap can be verified even in the presence of interactions, consistent with the more general results we derive in an accompanying work [Masaoka et al., Phys. Rev. B 113, 195120 (2026)].

Thermodynamic phase transitions typically involve a transition in the microscopic ordering between a less and a more ordered state. After transitioning into an ordered state, localized structures known as defects may remain, which disrupt the overall macroscopic order. Kibble-Zurek theory predicts that for any second-order phase transition the density of defects that form should be determined by the scaling law for the system coherence time and the phase transition quench time. We have performed measurements of sound dissipation due to vortex mutual friction in thin channels of superfluid ${}^3\mathrm{He}$ where one spatial dimension is smaller than a characteristic length scale predicted by the Kibble-Zurek theory. In this regime, we find that the density of defects is not determined by quench time, but can be estimated by replacing the Kibble-Zurek length scale with a truncated length set by the channel size. This results in significantly higher defect densities than in a bulk system.

For electronic devices based on two-dimensional semiconductors (2DSs), few-layer transition-metal dichalcogenides offer enhanced electrical performance over their monolayer counterparts. Metal-semiconductor junctions (MSJs), which lie at the heart of such devices, incorporate not only a metal-semiconductor (M-S) interface but also multiple van der Waals semiconductor-semiconductor (S-S) interfaces when few-layer 2DSs are employed. However, the role of S-S interfaces within MSJs—particularly their interplay with the M-S interface and their potential device applications—remains poorly understood. Here, based on density-functional theory calculations, we conduct a systematic investigation of MSJs based on few-layer 2DSs ($\mathrm{Mo}{\mathrm{S}}_2$ and InSe) contacted with both two- and three-dimensional metals. We find layer-dependent band offsets between adjacent 2DS layers, which can change sign depending on the proximity to the metal contact. This spatial variation directly results in quantized band bending that deviates from classical models. These phenomena are attributed to the interplay and competition among three effects: (1) the decay of interface dipole with distance from the M-S interface, leading to layer-resolved band bending; (2) the interlayer quasibonding mixes up the energy levels between different 2DS layers, thereby affecting the band bending; and (3) screening by partially metallized 2DS layers within the junction, which can alter the direction of band bending. Leveraging these insights, we provide an understanding of the density-functional theory calculated Schottky barrier height values for few-layer 2DSs in MSJs. Additionally, we demonstrate that the interlayer band offset can suppress direct tunneling in vertical field-effect transistors based on 2DSs.

The evaluation of the Jones polynomial at roots of unity is a paradigmatic problem for quantum computers. In this work, we present experimental results obtained from existing noisy quantum computers for special cases of this problem, where it is classically tractable. Our approach relies on the reduction of the problem of evaluating the Jones polynomial of a knot at lattice roots of unity to the problem of computing quantum amplitudes of qudit stabilizer circuits, which are classically efficiently simulatable. More specifically, we focus on evaluation at the fourth root of unity, which is a lattice root of unity, where the problem reduces to evaluating amplitudes of qubit stabilizer circuits. To estimate the real and imaginary parts of the amplitudes up to additive error we use the Hadamard test, yielding non-Clifford circuits that nevertheless we can always efficiently compute the correct output of. Hence, we further argue that this setup defines a standard benchmark for near-term noisy quantum processors. Additionally, we study the benefit of performing quantum error mitigation with the method of zero-noise extrapolation.