A promising branch of neuromorphic computing aims to perform cognitive operations in hardware, leveraging the physics of efficient and well-established nanodevices. In this work, we present a reconfigurable classifier based on a network of electrically connected magnetic tunnel junctions that categorizes information encoded in the amplitude of input currents through the spin-torque-driven magnetization switching output configuration. The network can be trained to classify new data by adjusting additional programming currents applied selectively to the junctions. We experimentally demonstrate that a network composed of three magnetic tunnel junctions can learn to classify spoken vowels with a recognition rate that surpasses the performance of software multilayer neural networks with the same number of trained parameters in this task. These results, obtained with the same nanodevices and working principle employed in industrial spin-transfer torque magnetic random-access memories (STT-MRAMs), constitute an important step toward the development of large-scale neuromorphic networks based on well-established technology.

We present a cavity-electromechanical system comprising a superconducting quantum interference device which is embedded in a microwave resonator and coupled via a pickup loop to a $6$-$\text{μ}\mathrm{g}$ magnetically levitated superconducting sphere. The motion of the sphere in the magnetic trap induces a frequency shift in the SQUID-cavity system. We use microwave spectroscopy to characterize the system, and we demonstrate that the electromechanical interaction is tunable. The measured displacement sensitivity of ${10}^{-7}\mathrm{m}/\sqrt{\mathrm{Hz}}$ defines a path towards ground-state cooling of levitated particles with Planck-scale masses at millikelvin environment temperatures.

Superparamagnetic tunnel junctions are important devices for a range of emerging technologies, but most existing compact models capture only their mean switching rates. Capturing qualitatively accurate analog dynamics of these devices will be important as the technology scales up. Here we present results using a one-dimensional overdamped Langevin equation that captures statistical properties of measured time traces, including voltage histograms, drift and diffusion characteristics as measured with Kramers-Moyal coefficients, and dwell-time distributions. While common macrospin models are more physically motivated magnetic models than the Langevin model, we show that for the device measured here, they capture even fewer of the measured experimental behaviors.

Controlled engineering of vortex pinning sites in copper-oxide superconductors is a critical issue in manufacturing devices based on magnetic flux quanta. To address this, we employed a focused $\mathrm{He}$-ion beam ($\mathrm{He}$-FIB) to irradiate thin ${\mathrm{YBa}}_2{\mathrm{Cu}}_3{\mathrm{O}}_{7-\delta }$ films and create ultradense hexagonal arrays of defects with lattice spacings as small as 20 nm. Critical current and magnetoresistance measurements demonstrate efficient pinning by a matching field of 6 T visible in a huge temperature range from the critical temperature $T_c$ down to 2 K. These results show that $\mathrm{He}$-FIB irradiation provides excellent opportunities for the development and application of superconducting fluxonic devices based on Abrikosov vortices. In particular, our findings suggest that such devices can operate at temperatures far below $T_c$, where superconductivity is robust.

Environmental fluctuations degrade the performance of solid-state qubits but can in principle be mitigated by real-time Hamiltonian estimation down to timescales set by the estimation efficiency. We implement a physics-informed and an adaptive Bayesian estimation strategy and apply them in real time to a semiconductor spin qubit. The physics-informed strategy propagates a probability distribution inside the quantum controller according to the Fokker-Planck equation, appropriate for describing the effects of nuclear spin diffusion in gallium arsenide. Evaluating and narrowing the anticipated distribution by a predetermined qubit probe sequence enables improved dynamical tracking of the uncontrolled magnetic field gradient within the singlet-triplet qubit. The adaptive strategy replaces the probe sequence by a small number of qubit probe cycles, with each probe time conditioned on the previous measurement outcomes, thereby further increasing the estimation efficiency. The combined real-time estimation strategy efficiently tracks low-frequency nuclear spin fluctuations in solid-state qubits, and can be applied to other qubit platforms by tailoring the appropriate update equation to capture their distinct noise sources.

The growing demand for artificial intelligence has motivated research into nontraditional physical devices that enable efficient learning in various tasks. This requires the devices to exhibit natural nonlinear dynamics with minimal power consumption. Here we present the application of artificial spin ice (ASI) and an as-grown chiral helimagnet (CHM) as the nonlinear component in physical reservoir computing (RC) and deep neural networks (DNNs). Their complex nonlinear magnetodynamics can be easily characterized by the broadband coplanar waveguide–based ferromagnetic resonance technique, originating from the specifically geometrical frustration effect and intrinsic multiple magnetic interactions competition, respectively. On the basis of the experimentally obtained nonlinear magnetodynamic response curves of these two noncollinear spin textures, we build ASI- and CHM-based physical reservoirs for RC and use the absorption and differential ferromagnetic resonance spectra as the activation function and its derivatives to perform nonlinear transformation of inputs for DNNs. The results demonstrate that physical RC and DNNs can accomplish time-series prediction and image-recognition tasks, respectively, with high accuracy and low power consumption. Our findings provide valuable insights and a promising pathway toward neuromorphic hardware using abundant artificial or natural nontrivial magnetic systems.

In topological crystalline insulators, the presence of crystalline symmetries can lead to a filling anomaly—a mismatch in electron counts that causes quantized fractional charges at boundaries. These charges act as a critical probe of crystalline topology and are notably present even in electrically neutral systems, including artificial photonic and phononic systems. In these contexts, the boundary-localized fractional charges are represented by the local density of states, which is integrated over a specific frequency range (and is referred to as the integrated mode density). Here we report our measurement of the local density of states in an elastic wave lattice system. We propose an accurate approach to the measurement of the system’s Green’s function, achieved through piezoelectric transducer excitation and noncontact laser vibrometry detection. The experimental design includes both one-dimensional elastic plates and two-dimensional lattices with disclination defects. The experimental results conclusively show quantitative differentiation of these phases as well as the fractional charge, i.e., the fractional integrated mode density trapped in disclinations. Our approach provides valuable insights into the measurement of the local density of states in elastic systems, with implications for vibration enhancement and structural vibration control.

Cylindrical vector beams (CVBs) are radially polarized or azimuthally polarized beams in the radiative near field. They resist diffraction within the Fresnel zone and exhibit a tighter focusing spot compared to linearly polarized beams. Therefore, CVBs have found several applications, and multiple passive methods have been proposed to generate them. To date, passive antennas have been limited to generating standard CVBs like Gaussian or Bessel beams at microwave frequencies. Therefore, the topic of this paper is twofold. First, a passive method to generate arbitrarily defined CVBs is proposed. Second, the method is applied to realize two experimental CVB antennas. One antenna generates an orthogonal Bessel summation (OBS) beam, while the other generates the well-known radial Gaussian (RG) beam. The OBS beam is a superposition of multiple, weighted Bessel beams. Therefore, it cannot be readily realized using standard CVB antennas. It is unique in that it exhibits zero diffraction loss at a given plane tangential to the aperture, within its Fresnel zone, resulting in maximum coupling between circular apertures within the Fresnel zone. Key to the proposed method of generating CVBs is the spatially dispersive mode-converting metasurface. Unlike conventional metasurfaces, the normal power density does not have to be locally conserved across this metasurface.

Optical phase determination is an important and established tool in diverse fields such as astronomy, biology, or quantum optics. There is increasing interest in using a lower number of total photons. However, different noise sources, such as electronic readout noise in the detector, and shot noise, hamper the phase estimation in regimes of very low illumination. Here, we report a study on how the quality of phase determination is affected by these two sources of noise. To that end, we experimentally reconstruct different wave fronts by means of a point-diffraction interferometer for different mean intensities of illumination, up to $15$ photons per pixel. Our interferometer features a skipper-CCD sensor, which allows us to reduce the readout noise arbitrarily, thus enabling us to separate the effect of these two sources of noise. For two cases of interest—a spatial qudit encoding phase, consisting of $d=6$ uniform phase regions, and a more general continuous phase—we see that reducing the readout noise leads to a clear improvement in the quality of reconstruction. This can be explained by a simple noise model that allows us to predict the expected fidelity of reconstruction and shows excellent agreement with the measurements.

We demonstrate parametric all-optical modulation in a periodically poled lithium niobate microring resonator on a chip. It uses quantum Zeno blockade between two distinct waves, a signal and a pump, through their sum-frequency generation at a large per-photon efficiency of $8.2$ MHz. With nanosecond pump pulses with $6$-mW peak power, $85.7\mathrm{\%}$ modulation extinction is observed, achieving an efficiency increase of more than 30 times compared with previous implementations. With only $2$-mW pump peak power, $43.0\mathrm{\%}$ modulation extinction is observed for a twice-as-strong signal of $4$ mW. This demonstrates that optical transistors with cascadability and fan-out are possible with just parametric nonlinear optics. These results, together with inherent advantages in such photonic integrated circuits, open the door to scalable technology for all-optical and quantum information processing.

Magnetocardiography (MCG) has emerged as a sensitive and precise method to diagnose cardiovascular diseases, providing more diagnostic information than traditional technology. However, the sensor limitations of conventional MCG systems, such as large size and cryogenic requirement, have hindered the widespread application and in-depth understanding of this technology. In this study, we present a high-sensitivity, room-temperature MCG system based on the negatively charged nitrogen-vacancy ($\mathrm{N}$-$V$) centers in diamond. The magnetic cardiac signal of a living rat, characterized by an approximately 20-pT amplitude in the $R$ wave, is successfully captured through noninvasive measurement using this innovative solid-state spin sensor. To detect these extremely weak biomagnetic signals, we utilize sensitivity-enhancing techniques such as magnetic flux concentration. These approaches have enabled us to simultaneously achieve a magnetometry sensitivity of $9\text{pT}{\text{Hz}}^{-1/2}$ and a sensor scale of 5 mm. By extending the sensing scale of the $\mathrm{N}$-$V$ centers from cellular and molecular level to macroscopic level of living creatures, we have opened the future of solid-state quantum sensing technologies in clinical environments.

Trap-mediated recombination influences the performance of a wide range of electronic devices. The well-known Shockley-Read-Hall (SRH) expression for inorganic semiconductors is often invoked to describe the recombination rate in organic materials, although without a clear understanding of how its parameters relate to the underlying material properties or how it should be modified to account for the finite lifetime of exciton intermediates in, for example, the doped emissive layer of an organic light-emitting diode (OLED). Here, we formalize SRH recombination for organic semiconductors based on diffusive trapping and Langevin recombination. We show that including the exciton state suppresses the recombination rate in host-guest systems with type II energy level alignment whenever the interfacial gap between the host and guest molecular orbitals is comparable to the exciton energy. These results quantify the balance between bimolecular and trap-mediated recombination in doped OLED emissive layers, and indicate that devices with type II host-guest pairings can, in principle, beat the thermodynamic limit of their neat guest counterparts.

This paper details a class of metal-based space-time metasurfaces for application in wireless communications scenarios. Concretely, we describe space-time metasurfaces that periodically alternate their properties in time between three spatial states: “air,” “conductor,” and “grating.” We analyze the physics of these metastructures via a computationally efficient analytical technique based on the use of Floquet-Bloch series, integral equations, and circuit models. By doing so, we reveal features of these spatiotemporal metasurfaces: scattering parameters, field profiles, diffraction angles, and the nature of the space-time harmonics. The results, corroborated with a self-implemented numerical finite-difference time-domain approach, show the potential application of these space-time metasurfaces as beamformers acting in reflection, in transmission or both. The amplitude and direction of the diffracted orders can be electronically controlled with the parameters of the metasurface. Moreover, the intrinsic ability of time-modulated diffractive metasurfaces to mix and multiply frequencies is tested. We show how two different modulations can lead to the same diffraction angle but with different mixed output frequencies.

The cocktail party effect is the capability to focus one’s auditory attention on particular audio sources while ignoring other audio sources. We propose an experimental strategy to reproduce this ability by designing a time-dependent metasurface composed of independent active mirrors. Each active mirror is a programmable acoustic unit cell capable of hearing, computing, and re-emitting acoustic signals: each of them acts as a convolution filter. The proper configuration of the metasurface temporal filters allows one to establish an acoustic communication link between groups of individuals immersed in the noisy environment: a multiple-user multiple-input, multiple-output acoustic system is built.

Hybrid quantum photonic systems connect classical photonics to the quantum world and promise to deliver efficient light-matter quantum interfaces while leveraging the advantages of both, the classical and the quantum, subsystems. However, combining efficient, scalable photonics and solid-state quantum systems with desirable optical and spin properties remains a formidable challenge. In particular, the access to individual spin states and coherent mapping to photons remains unsolved for hybrid systems. In this paper, we demonstrate all-optical initialization and readout of the electron spin of a negatively charged silicon-vacancy center in a nanodiamond coupled to a silicon nitride photonic crystal cavity. We characterize relevant parameters of the coupled emitter-cavity system and determine the silicon-vacancy center’s spin-relaxation and spin-decoherence rate. Our results mark a key step towards the realization of a hybrid spin-photon interface based on silicon nitride photonics and the silicon-vacancy center’s electron spin in nanodiamonds with potential use for quantum networks, quantum communication, and distributed quantum computation.

Nanomechanical computers promise robust, low-energy information processing; however, to date, electronics have generally been required to drive gates, and logical operations have generally involved bits with different oscillation frequencies. This limits the scalability of nanomechanical logic. Here we demonstrate an acoustically driven logic gate that has a single frequency of operation. Our gate uses the bistability of a nonlinear mechanical resonator to define logical states. These states are efficiently coupled into and out of the gate via nanomechanical waveguides, providing the mechanical equivalent of electrical wires and allowing purely mechanical information transfer. Since the inputs and output all share the same frequency, they are compatible with cascaded chains of gates. Our architecture is CMOS compatible, and with miniaturization could allow an energy cost that approaches the fundamental Landauer limit. Together this presents a pathway towards large-scale nanomechanical computers.

The increasing scale of neural networks needed to support more complex applications has led to an increasing requirement for area- and energy-efficient hardware. One route to meeting the budget for these applications is to circumvent the von Neumann bottleneck by performing computation in or near memory. However, an inevitability of transferring neural networks onto hardware is the fact that nonidealities, such as device-to-device variations or poor device yield impact performance. Methods, such as hardware-aware training, where substrate nonidealities are incorporated during network training, are one way to recover performance at the cost of solution generality. In this work, we demonstrate inference on hardware-based neural networks consisting of 20 000 magnetic tunnel junction (MTJ) arrays integrated on CMOS chips in a form that closely resembles scalable and market-ready spin transfer-torque magnetoresistive random access memory (STT-MRAM) technology. Using 36 dies, each containing a MTJ-CMOS crossbar array with its own nonidealities, we show that even a small number of defects in physically mapped networks significantly degrades the performance of networks trained without defects and show that, at the cost of generality, hardware-aware training accounting for specific defects on each die can recover to comparable performance with ideal networks. We then demonstrate a robust training method that extends hardware-aware training to statistics-aware training, producing network weights that perform well on most defective dies regardless of their specific defect locations. When evaluated on the 36 physical dies, statistics-aware trained solutions can achieve a mean misclassification error on the MNIST dataset that differs from the software-baseline by only 2%. This statistics-aware training method could be generalized to networks with many layers that are mapped to hardware suited for industry-ready applications.

In this paper, we present an ab initio methodology to account for electron-phonon interactions in two-dimensional (2D) materials, focusing on transition-metal dichalcogenides (TMDCs). It combines density functional theory and maximally localized Wannier functions to acquire material data and relies on the linearized Boltzmann transport equation (LBTE) and the nonequilibrium Green’s functions (NEGF) method to determine the transport properties of materials and devices, respectively. It is shown that both the LBTE and NEGF methods return very close mobility values, without the need to adjust any parameters. The excellent agreement between the two approaches results from the inclusion of nondiagonal entries in the electron-phonon scattering self-energies. The NEGF solver is then used to shed light on the “current versus voltage” characteristics of a monolayer $\mathrm{Mo}\mathrm{S}$${}_2$ transistor, highlighting how the interactions with phonons impact both the current magnitude and its distribution. The mobility of other TMDCs is considered as well, demonstrating the capabilities of the proposed technique to assess the potential of 2D channel materials in next-generation logic applications.

Coupled-bistable-oscillator machines have recently generated significant interest due to their observed ability to rapidly produce high-quality solutions to nondeterministic-polynomial-time-complete optimization problems. While the dynamics of such systems are often derived in the literature, it has hitherto been unclear why exactly the system dynamics perform optimization so well. This paper answers this question by presenting a complete equivalence between coupled-oscillator machines and the primal-dual method of Lagrange multipliers. This equivalence explains how coupled-oscillator solvers implement the correct optimization constraints and find high-quality solutions. The equivalence also provides precise mathematical meaning to each system component and enables the principled design of systems that implement more-sophisticated optimization algorithms. We simulate the system dynamics and demonstrate (1) that its performance is competitive with performance of the best-known digital algorithms, (2) that the circuit is robust with regard to large component variations, hinting that the traditional shortcomings of analog computation may be less important for these applications, and (3) that the circuit consumes extremely low amounts of power (on the order of milliwatts) and energy (approximately $100$ nJ) per optimization even for problems of 2000 variables.

A wide-field magnetometer utilizing nitrogen-vacancy (N-$V$) centers in diamond that does not require microwaves is demonstrated. It is designed for applications where microwaves need to be avoided, such as magnetic imaging of biological or conductive samples. The system exploits a magnetically sensitive feature of N-$V$ centers near the ground-state level anticrossing (GSLAC). An applied test field from a wire has been mapped over an imaging area of $\approx 500\times 470\text{μ}{\mathrm{m}}^2$. Analysis of the ground-state level anticrossing (GSLAC) lineshape allows us to extract the vector information of the applied field. The device allows micrometer-scale magnetic imaging at a spatial resolution dominated by the thickness of the N-$V$ layer (here, $50\text{μ}\mathrm{m}$). For a pixel size of $4\text{μ}\mathrm{m}\times 3.8\text{μ}\mathrm{m}$, the estimated photon-shot-noise-limited sensitivity is $4.8\text{μ}\mathrm{T}/\sqrt{\mathrm{Hz}}$. Two modalities for visualizing the magnetic fields, static and temporal, are presented, along with a discussion of technical limitations and future extensions of the method.