A preprint on arXiv proposes a method called Quantum Krylov Learning for determining the Hamiltonian of a quantum system directly from measurements, rather than by fitting candidate microscopic models to observed data. The authors note that, in many settings, direct access to the full system is unavailable, motivating an approach that can work under restricted access.
OutlookSpeculative
If Quantum Krylov Learning reduces the measurement overhead for Hamiltonian estimation under partial access, it could be used within two years to characterize near-term quantum hardware more efficiently than existing Hamiltonian learning protocols.
A preprint on arXiv introduces Walshness, a new metric intended to characterize how readily a quantum state can be represented by a neural network. It addresses the limited understanding of neural quantum state efficiency, which the authors attribute in part to nonlinear parameterization and sign structure. The metric is framed as an intrinsic property of the quantum state itself.
OutlookPlausible
Walshness could become a practical pre-screening tool for choosing neural-network ansätze in variational Monte Carlo simulations within two years.
A preprint introduces graph neural networks that predict whether a quantum error correction decoder will fail, using only syndrome data as input. The method is intended to support shot post-selection without requiring additional decoder executions, and is aimed at high-rate qLDPC codes.
OutlookPlausible
Within two years, GNN-based post-selection filters could be integrated into existing high-rate qLDPC decoding pipelines, reducing the number of shots needed to reach a target logical error rate without adding decoder calls.
A preprint on arXiv introduces a finite-measurement theory for inferring the symmetry group of a quantum learning model from candidate transformations, using both observable-level and task-level information. It identifies transformations that no accessible observable can distinguish with a stabilizer condition.
OutlookPlausible
If the inference procedure is practical, quantum ML developers could automatically select symmetry-preserving ansätze from limited measurement data, lowering the sample requirements for training on near-term devices.
D-Wave has opened a beta program for a gate-model quantum computing simulator through its Leap cloud service. The simulator supports up to 21 qubits and includes ideal and hardware emulation modes that model D-Wave's dual-rail superconducting cavity qubit architecture. It is intended to let users test mid-circuit error detection using hardware-level erasure information.
OutlookPlausible
The simulator could let researchers prototype and validate error-correction protocols for dual-rail erasure qubits before D-Wave has physical gate-model hardware, seeding a software ecosystem around its architecture.
The Quantum Internet Alliance has received €47.5 million from the European Commission. The funding is intended to support construction of a full-stack quantum network prototype.
OutlookPlausible
This could make a metropolitan-scale quantum network testbed available to early application developers within two years.
A preprint posted to arXiv introduces QuLoC, a compression pipeline for large language models that combines SVD-based low-rank approximation with outputs from photonic quantum circuits. The quantum outputs are used as gating signals intended to preserve downstream task performance after parameter reduction. The abstract describes the method but does not report experimental benchmarks or hardware demonstrations.
OutlookSpeculative
If QuLoC can be simulated on classical hardware or run on small photonic processors, it could produce a benchmark showing improved downstream accuracy over standard low-rank SVD on a small language model within two years.
A theoretical preprint analyzes quantum circuits restricted to local operations on a 2D lattice under constant-rate local stochastic noise. It reports that such noisy planar quantum circuits can compute functions outside AC0, the class of constant-depth classical circuits with unbounded fan-in AND, OR, and NOT gates. The abstract indicates the result holds even though the classical circuits are noiseless and have no geometric constraints.
OutlookPlausible
This could reframe near-term quantum advantage experiments to target a separation against constant-depth classical circuits, giving noisy 2D devices a well-defined benchmark they might meet before full fault tolerance.
A preprint on arXiv introduces SQD-Agent, an LLM-driven agentic framework designed to help researchers translate quantum chemistry problems into executable hybrid quantum-classical workflows. The abstract frames the core barrier as the need for expertise in quantum algorithms and hardware-specific details, which the framework aims to reduce.
OutlookSpeculative
If SQD-Agent can validate the quantum circuits it generates against classical quantum chemistry benchmarks, it could give computational chemists without quantum algorithm expertise a usable entry point for small hybrid variational calculations on available hardware within two years.
A preprint on arXiv presents a comparison of hybrid quantum-classical self-supervised learning models for fingerprint recognition, with classical and quantum variants matched for width. The authors frame self-supervised learning as a way to avoid large labeled enrollment datasets and see hybrid models as a route to richer representations. They note that earlier quantum self-supervised learning work has examined only a single contrastive objective.
OutlookPlausible
The width-matched setup could provide a reusable reference for testing whether hybrid quantum-classical SSL actually improves fingerprint representations beyond classical SSL, helping practitioners decide within two years whether quantum layers are worth the overhead.
Researchers refined a finite-size classical soundness bound for Aasnaess's graph-distributed construction, making it scale linearly with the number of players. Combined with standard disjoint-player repetition, this yields a two-round interactive test for shallow-circuit quantum advantage. The result is aimed at enabling experimentally testable demonstrations at finite circuit sizes.
OutlookPlausible
This could allow near-term quantum hardware teams to run the two-round verification test on existing shallow circuits within the next two years, producing a quantum advantage claim backed by explicit classical bounds.
Infleqtion and Riverlane announced a strategic memorandum of understanding to integrate Riverlane's Deltaflow real-time decoder with Infleqtion's neutral-atom quantum processor and qLDPC software. The collaboration aims to close the latency gap between qubit error detection and processing so that error correction cycles can run dynamically on neutral-atom hardware.
OutlookPlausible
Infleqtion's neutral-atom platform could demonstrate real-time quantum error correction cycles within two years, yielding a logical qubit whose lifetime extends beyond its physical constituents.
Researchers introduced a framework for quantum inference in which the protocol's output is itself a quantum state rather than a classical measurement result. They identified tasks such as quantum purity amplification, random purification, approximate cloning, and density matrix exponentiation as instances of this coherent-output setting. The authors report that these protocols can achieve an exponential sample-complexity advantage over standard classical-output inference.
OutlookPlausible
This framework could make density matrix exponentiation and purity amplification practical on small quantum processors by reducing the number of physical copies needed, enabling proof-of-principle demonstrations that were previously sample-limited.
A preprint identifies three bottlenecks for deep reinforcement learning in quantum circuit optimisation: replay buffers ignore the reliability of temporal-difference targets, curriculum-based architecture search demands a full quantum-classical evaluation after every edit, and noiseless trajectories are discarded when retraining under hardware noise. The work is framed around replay-buffer engineering for noise-aware circuit optimisation.
OutlookPlausible
Redesigned replay buffers that retain noiseless trajectories and weight temporal-difference targets by reliability could let RL-based circuit optimisers adapt to hardware noise without a full quantum-classical evaluation at every step, making noise-aware compilation cheaper on near-term devices.
A preprint on arXiv proposes an analog quantum kernel method using Rydberg atoms instead of gate-based circuits, and applies it to estimating non-Markovianity from sparse temporal data. The approach is framed as noise-enhanced, suggesting the authors deliberately incorporate or exploit noise rather than treating it only as an obstacle.
OutlookPlausible
Within the next two years, this could let neutral-atom quantum processors characterize non-Markovian noise in other quantum systems directly from sparse time-series data, avoiding full process tomography.
An arXiv preprint reports an experimental quantum-computing implementation of a Markov Chain Monte Carlo routine, using quantum amplitude estimation as the engine for faster sampling-based mean estimates when the target distribution is encoded in a quantum state. The authors position the result as a step toward the quadratic speedup known for certain sampling tasks.
OutlookPlausible
If the state-preparation circuits remain stable on current noisy processors, this could enable early quantum-assisted Bayesian inference on small, low-dimensional models in risk analysis or clinical statistics within two years.
A preprint on arXiv proposes a quantum reservoir computing scheme for superconducting devices in which the system is measured repeatedly during its natural dissipative dynamics, and the resulting measurement record is used as a feature stream for time-series prediction. The abstract frames the approach as using the nonlinear and memory properties of quantum dynamics rather than conventional digital quantum circuits. No experimental results are reported in the abstract.
OutlookPlausible
This could make existing cloud-accessible superconducting processors usable as small-scale temporal machine learning testbeds within two years.
Researchers have introduced a geometry-conditioned neural-network quantum state for molecular electronic structure formulated in second quantization. The method aims to share wavefunction coefficient representations across molecular geometries, so that a single trained network can describe a potential energy surface rather than being retrained for each nuclear configuration.
OutlookPlausible
Within two years, this approach could be benchmarked against established potential energy surface datasets for small molecules, providing accuracy comparable to multi-reference methods at a fraction of the repeated training cost.
A new arXiv preprint introduces a single-circuit quantum generative adversarial network method for image generation. It uses quantum fidelity landscape information to calibrate prior distributions, moving away from patch-based decomposition. The authors note that existing patch-based QGAN methods can harm global consistency in generated images.
OutlookPlausible
This could make single-circuit QGANs a practical baseline for low-resolution image generation on current NISQ hardware, enabling more consistent comparisons with classical GANs on small datasets.
A preprint introduces QASM-Eval, a dataset intended to train and evaluate large language models on OpenQASM-3 code. The abstract frames the dataset around NISQ-era constraints, arguing that useful quantum programs need hardware-facing features beyond ordinary gate sequences, such as mid-circuit measurement and classical feedback for quantum error correction, and precise timing for dynamical decoupling. The abstract text is cut off after those examples.
OutlookPlausible
Within two years, an LLM fine-tuned on QASM-Eval could reliably generate OpenQASM-3 snippets that include mid-circuit measurement and timing controls, making it easier for experimental teams to prototype QEC and dynamical decoupling routines on available hardware.