Typical mammal brains have some form of random connectivity between neurons. Reservoir computing, a neural network approach, uses random weights within its processing layer along with built-in recurrent connections and short-term, fading memory, and is shown to be time and training efficient in processing spatiotemporal signals. Here we prepared a niobium oxide-based thin film memristor device with intrinsic structural inhomogeneity in the form of random nanopores and performed computational tasks of XOR operations, image recognition, and time series prediction and reconstruction. For the latter task we chose a complex three-dimensional chaotic Lorenz-63 time series. By applying three temporal voltage waveforms individually across the device and training the readout layer with electrical current signals from a three-output physical reservoir, we achieved satisfactory prediction and reconstruction accuracy in comparison to the case of no reservoir. This work highlights the potential for scalable, on-chip devices using all-oxide reservoir systems, paving the way for energy-efficient neuromorphic electronics dealing with time signals.

We present a Wilson-Cowan reservoir computer (WC-RC) that treats a retinotopic excitatory-inhibitory neural field as a structured reservoir. Travelling waves and bounded oscillations provide an interpretable spatiotemporal basis, while a two-stage sampler (40 sites × 200 steps) exports 8000 features per input-a [Formula: see text] reduction with unchanged integration cost. On MNIST and Fashion-MNIST, recurrent readouts trained on these features achieve strong performance within this fixed export budget: Att-LSTM reaches [Formula: see text], while GRU and vanilla LSTM yield similar accuracies (all with tight Wilson 95% confidence intervals). A simple MLP readout performs markedly worse ([Formula: see text]), whereas a compact ridge classifier still attains non-trivial performance ([Formula: see text]), indicating that the exported WC-RC representation is partially linearly decodable but benefits further from temporal modelling. Selective suppression of lateral couplings ([Formula: see text]) shows that reinstating wave dynamics improves recurrent models while degrading the MLP, supporting a functional role for propagating dynamics. A complementary neighbour-coupling ablation, implemented via a diffusion-like scaling factor ψ, shows that increasing lateral spread beyond the baseline regime reduces late-time spatial variance and degrades recurrent-readout accuracy, indicating that useful computation depends on balanced wave dynamics rather than maximal smoothing. At matched exported-feature and readout budgets, ESN baselines underperform ([Formula: see text]), whereas compact CNNs achieve higher accuracy but with larger parameter and MAC budgets and without wave interpretability. Supplementary analyses confirm numerical fidelity and relate performance gains to propagation coherence. We outline a fixed-point streaming-convolution mapping for FPGA/ASIC deployment, positioning WC-RC as an interpretable, energy-aware reservoir for neuromorphic vision.

Biological systems use neural circuits to integrate input information and produce outputs. Synaptic convergence, where multiple neurons converge their inputs onto a single downstream neuron, is common in natural neural circuits. However, understanding specific computations performed by such neural blocks and implementating them in hardware requires further research. This work focuses on synaptic convergence in a simplified circuit of three spiking artificial neurons based on diffusive memristors. Numerical modelling and experiments reveal input voltage combinations that enable targeted activation of spiking for specific neuron configurations. We analyse the statistical characteristics of spiking patterns and interpret them from a computational perspective. The numerical simulations match experimental measurements. Our findings contribute to development of universal functional blocks for neuromorphic systems.

We investigate the emergence of cooperative and competitive behaviors in multiagent reinforcement learning systems, where autonomous agents, modeled as active particles, learn to capture targets dynamically regenerated in geometrically distinct arenas. Using proximal policy optimization, we study circular, elliptical, and square geometries with varying target distributions. Our results show that cooperation frequently induces spontaneous symmetry breaking with agents exploring different arena regions, whereas competition tends to preserve symmetry and produce overlapping space distributions. In both fixed and mobile-target simulations, symmetry breaking occurs. The ability to break symmetry and reach optimal strategies depends sensitively on the arena geometry, agent number, and location of target-rich regions. Moreover, in a "blind" situation where agents lose direct target perception, symmetry breaking can still emerge in rare instances, highlighting that memory and interagent interactions alone may suffice to induce asymmetry. Introducing stochasticity through reward noise further facilitates this process, allowing agents to escape metastable symmetric states with moderate performance and converge toward fully asymmetric configurations with optimal performance. Together, these findings establish symmetry breaking as a key organizing principle in learning-based multiagent systems and reveal how geometry, reward structure, and stochasticity jointly govern the emergence of collective intelligence.

Spatiotemporal trajectories are ubiquitous across physical, biological, and social systems, yet their inherent complexity has hindered systematic comparison and interpretation. Here, we introduce a robust and versatile framework for trajectory categorization, utilizing a minimal set of geometric indices to quantify spatial occupation, co-occupancy, and temporal coordination. This representation defines a low-dimensional feature space enabling mechanism-agnostic classification, exposing the hidden organizational structure of movement patterns. We validate the efficacy and broad utility of our approach across three disparate domains: Hamiltonian particle dynamics, reinforcement learning agents, and human movement in a museum. In all cases, high-dimensional data collapse into distinct, well-separated clusters corresponding to specific organizational motifs. The emergence of these conserved geometric structures across diverse systems demonstrates that spatiotemporal organization is governed by a low-dimensional logic, recoverable without knowledge of governing equations or internal decision processes. By providing a unified basis for analyzing movement from particles to humans, our method establishes a powerful tool for uncovering latent system motivations and interactions.

Many biological, chemical, and physical systems are underpinned by stochastic transitions between equilibrium states in a potential energy landscape. Here, we consider such transitions in a minimal model with two possible competing pathways, both starting from a local potential energy minimum and eventually finding the global minimum. There is competition between the distance to travel in state space and the height of the potential energy barriers to be surmounted, for the transition to occur. One pathway has a higher energy barrier to go over, but requires traversing a shorter distance, whereas the other pathway has a lower potential barrier but it is substantially further away in configuration space. The most likely pathway taken depends on the available time for the transition process; when only a relatively short time is available, the most likely path is the one over the higher barrier. We find that upon varying temperature the overall most likely pathway can switch from one to the other. We calculate the statistics of where the barrier crossing occurs and the distribution of times taken to reach the potential minimum. Interestingly, while the configuration space statistics is complex, the time of arrival statistics is rather simple, having an exponential probability density over most of the time range. Taken together, our results show that empirically observed rates in nonequilibrium systems should not be used to infer barrier heights.

Recently created diffusive memristors have garnered significant research interest owing to their distinctive capability to generate a diverse array of spike dynamics, which are similar in nature to those found in biological cells. This gives the memristor an opportunity to be used in a wide range of applications, specifically within neuromorphic systems. The diffusive memristor is known to produce regular, chaotic, and stochastic behaviors, which leads to interesting phenomena resulting from the interactions between the behavioral properties. The interactions, along with the instabilities that lead to the unique spiking, phenomena are not fully understood due to the complexities associated with examining the stochastic properties within the diffusive memristor. In this work, we analyze both the classical and the noise-induced bifurcations that a set of stochastic differential equations, justified through a Fokker–Planck equation used to model the diffusive memristor, can produce. Finally, we replicate the results of the numerical stochastic threshold phenomena with experimentally measured spiking.

Diffusive memristors show great promise as fundamental components for brain-inspired neuromorphic computing. By relying on the drift and diffusion of charge carriers, which form conductive filaments for charge transport, these devices offer high nonlinearity, tunability, fast switching between resistive states, and low power consumption. Their ability to generate a wide range of nonlinear dynamics, driven by the complex interplay of thermal, electrical, and mechanical effects, mimics the behavior of biological neurons. In this paper, we simulate spiking dynamics in an artificial neuron based on diffusive memristors with two independent conducting filaments. We uncover instabilities that lead to self-sustained spike generation and show that external voltage bias allows the coexistence of two characteristic spiking modes. Noise, either inherent or externally added, facilitates switching between these spiking regimes. The model predictions align well with our experimental measurements, offering the way for the development of neuromorphic devices for parallel signal processing.

We explore competitive dynamics in multiagent active matter systems using reinforcement learning. In our study, two active Brownian particles (referred to as predators) were trained using either simultaneous or sequential protocols to capture ten passive Brownian particles (preys). The training results depend on the agent, and generally one agent tends to overperform the other. To assess the effectiveness of the two protocols, we examined two policies: (i) a natural policy, where updates to the reinforcement learning parameters of both predators were stopped at a fixed time, even if one agent performed suboptimally; and (ii) a hybrid policy, where we combined the reinforcement learning parameters recorded when each agent achieved its optimal performance. If limited to natural training, simultaneous training appears to be the better option. However, when hybrid training is also allowed, sequential training becomes the preferred choice.

In this study we propose a new class of artificial neurons and memristors made of active chiral particles. We formulate a single-particle model to simulate active chiral particle behavior in a two-terminal device, with resistance depending on the particle position. We create a dynamical phase map connecting particle trajectories and memristor electrical properties to applied voltage and particle's self-propulsion parameters. Analysis of spiking modes in artificial neurons, with and without noise, shows the memristor switches between high- and low-resistance states, exhibiting stable limit cycles in the position-voltage phase response.

We investigate a minimal architecture for quantum reservoir computing based on Hamiltonian encoding, in which input data are injected via modulation of system parameters rather than state preparation. This approach circumvents many of the experimental overheads typically associated with quantum machine learning, enabling computation without feedback, memory, or state tomography. We demonstrate that such a minimal quantum reservoir, despite lacking intrinsic memory, can perform nonlinear regression and prediction tasks when augmented with post-processing delay embeddings. Our results provide a conceptually and practically streamlined framework for quantum information processing, offering a clear baseline for future implementations on near-term quantum hardware.

Diffusive memristors owing to their ability to produce current spiking when a constant or slowly changing voltage is applied are competitive candidates for development of artificial electronic neurons. These artificial neurons can be integrated into various prospective autonomous and robotic systems as sensors, e.g. ones implementing object grasping and classification. We report here Ag nanoparticle-based diffusive memristor prepared on a flexible polyethylene terephthalate substrate in which the electric spiking behaviour was induced by the electric voltage under an additional stimulus of external mechanical impact. By changing the magnitude and frequency of the mechanical impact, we are able to manipulate the spiking response of our artificial neuron. This functionality to control the spiking characteristics paves a pathway for the development of touch-perception sensors that can convert local pressure into electrical spikes for further processing in neural networks. We have proposed a mathematical model which captures the operation principle of the fabricated memristive sensors and qualitatively describes the measured spiking behaviour. Employing such flexible diffusive memristors that can directly translate tactile information into spikes, similar to force and pressure sensors, could offer substantial benefits for various applications in robotics.

The roadmap is organized into several thematic sections, outlining current computing challenges, discussing the neuromorphic computing approach, analyzing mature and currently utilized technologies, providing an overview of emerging technologies, addressing material challenges, exploring novel computing concepts, and finally examining the maturity level of emerging technologies while determining the next essential steps for their advancement.

Recently created diffusive memristors have garnered significant research interest owing to their distinctive capability to generate a diverse array of spike dynamics which are similar in nature to those found in biological cells. This gives the memristor an opportunity to be used in a wide range of applications, specifically within neuromorphic systems. The diffusive memristor is known to produce regular, chaotic and stochastic behaviors which leads to interesting phenomena resulting from the interactions between the behavioral properties. The interactions along with the instabilities that lead to the unique spiking phenomena are not fully understood due to the complexities associated with examining the stochastic properties within the diffusive memristor. In this work, we analyze both the classical and the noise induced bifurcations that a set of stochastic differential equations, justified through a Fokker-Planck equation used to model the diffusive memristor, can produce. Finally, we replicate the results of the numerical stochastic threshold phenomena with experimentally measured spiking.

In ferromagnet/superconductor bilayer systems, dipolar fields from the ferromagnet can create asymmetric energy barriers for the formation and dynamics of vortices through flux pinning. Conversely, the flux emanating from vortices can pin the domain walls of the ferromagnet, thereby creating asymmetric critical currents. Here, we report the observation of a superconducting diode effect (SDE) in a NbSe 2 /CrGeTe 3 van der Waals heterostructure in which the magnetic domains of CrGeTe 3 control the Abrikosov vortex dynamics in NbSe 2 . In addition to extrinsic vortex pinning mechanisms at the edges of NbSe 2 , flux-pinning-induced bulk pinning of vortices can alter the critical current. This asymmetry can thus be explained by considering the combined effect of this bulk pinning mechanism along with the vortex tilting induced by the Lorentz force from the transport current in the NbSe 2 /CrGeTe 3 heterostructure. We also provide evidence of critical current modulation by flux pinning depending on the history of the field setting procedure. Our results suggest a method of controlling the efficiency of the SDE in magnetically coupled van der Waals superconductors, where dipolar fields generated by the magnetic layer can be used to modulate the dynamics of the superconducting vortices in the superconductors.

In ferromagnet/superconductor bilayer systems, dipolar fields from the ferromagnet can create asymmetric energy barriers for the formation and dynamics of vortices through flux pinning. Conversely, the flux emanating from vortices can pin the domain walls of the ferromagnet, thereby creating asymmetric critical currents. Here, we report the observation of a superconducting diode effect in a NbSe_2/CrGeTe_3 van der Waals heterostructure in which the magnetic domains of CrGeTe_3 control the Abrikosov vortex dynamics in NbSe_2. In addition to extrinsic vortex pinning mechanisms at the edges of NbSe_2, flux-pinning-induced bulk pinning of vortices can alter the critical current. This asymmetry can thus be explained by considering the combined effect of this bulk pinning mechanism along with the vortex tilting induced by the Lorentz force from the transport current in the NbSe_2/CrGeTe_3 heterostructure. We also provide evidence of critical current modulation by flux pinning depending on the history of the field setting procedure. Our results suggest a method of controlling the efficiency of the superconducting diode effect in magnetically coupled van der Waals superconductors, where dipolar fields generated by the magnetic layer can be used to modulate the dynamics of the superconducting vortices in the superconductors.

Gamma photons with an average energy of 1.25 MeV are well-known to generate large amounts of defects in semiconductor electronic devices. Here we investigate the novel effect of gamma radiation on diffusive memristors based on metallic silver nanoparticles dispersed in a dielectric matrix of silica. Our experimental findings show that after exposure to radiation, the memristors and artificial neurons made of them demonstrate much better performance in terms of stable volatile resistive switching and higher spiking frequencies, respectively, compared to the pristine samples. At the same time we observe partial oxidation of silver and reduction of silicon within the switching silica layer. We propose nanoinclusions of reduced silicon distributed across the silica layer to be the backbone for metallic nanoparticles to form conductive filaments, as supported by our theoretical simulations of radiation-induced changes in the diffusion process. Our findings propose a new opportunity to engineer the required characteristics of diffusive memristors in order to emulate biological neurons and develop bio-inspired computational technology.

Zero-dimensional graphene quantum dots (GQD) dispersed in conducting polymermatrix display a striking range of optical, mechanical, and thermoelectricproperties which can be utilized to design next-generation sensors and low-costthermoelectric. This exotic electrical property in GQDs is achieved byexploiting the concentration of the GQDs and by tailoring the functionalizationof the GQDs. However, despite extensive investigation, the nonlinearresistivity behavior leading to memristive like characteristic has not beenexplored much. Here, we report electrical characterisation of nitrogenfunctionalized GQD (NGQD) embedded in a polyaniline (PANI) matrix. We observe astrong dependence of the resistance on current and voltage history, themagnitude of which depends on the NGQD concentration and temperature. Weexplain this memristive property using a phenomenological model of thealignment of PANI rods with a corresponding charge accumulation arising fromthe NGQD on its surface. The NGQD-PANI system is unique in its ability tomatrix offers a unique pathway to design neuromorphic logic and synapticarchitectures with crucial advantages over existing systems.

By consistently applying the formalism of quantum electrodynamics, we developed a comprehensive theoretical framework describing the interaction of single microwave photons with an array of superconducting transmon qubits in a waveguide cavity resonator. In particular, we analyze the effects of microwave photons on the array’s response to a weak probe signal exciting the resonator. The study reveals that high quality factor cavities provide a better spectral resolution of the response, while cavities with moderate quality factors allow better sensitivity for a single-photon detection. Remarkably, our analysis showed that a single-photon signal can be detected by even a sole qubit in a cavity under the realistic range of system parameters. We also discuss how the quantum properties of the microwave radiation and electrodynamical properties of resonators affect the response of qubits’ array. Our results provide an efficient theoretical background for informing the development and design of quantum devices consisting of arrays of qubits, especially for those using a cavity where an explicit expression for the transmission or reflection is required.

Memristive devices are promising elements for energy-efficient neuromorphic computing and future artificial intelligence systems. For diffusive memristors, the device state switching occurs because of the sequential formation and disappearance of conduction pillars between device terminals due to the drift and diffusion of Ag nanoparticles in the dielectric matrix. This process is governed by the application of the voltage to the device contacts. Here, both in experiment and in theory we demonstrate that varying temperature offers an efficient control of memristor states and charges transport in the device. We found out that by raising and lowering the device temperature, one can reset the memristor state as well as change the residual time the memristor stays in high and low resistive states when the current spiking is generated in the memristive circuit at a constant applied voltage. Our theoretical model demonstrates a good qualitative agreement with the experiments and helps to explain the effects reported.