Explore documented scientific discoveries, original research and new scientific developments. Search the library by topic, category or date.
arXiv:2610.40840v1 Announce Type: new Abstract: Data encoding is a central challenge in quantum machine learning (QML), as it determines how classical information is mapped onto quantum states and influences the expressive power and efficiency of parameterized quantum circuits.
arXiv QuantumRead article →arXiv:2610.40827v1 Announce Type: new Abstract: Quantifying non-Gaussianity is essential for assessing the computational power of bosonic quantum systems. We introduce a framework to witness and quantify non-Gaussianity from displaced parity measurements, and we directly apply it to experimental data obtained with superconducting cavities.
arXiv QuantumRead article →arXiv:2610.40810v1 Announce Type: new Abstract: The static limit of the Kubo susceptibility describes the adiabatic response of an isolated quantum system, in which energy-level occupations remain unchanged, and generally differs from the isothermal susceptibility, where these occupations readjust to thermal equilibrium.
arXiv QuantumRead article →arXiv:2610.40808v1 Announce Type: new Abstract: Driven-dissipative quantum devices are routinely operated with time-dependent drives and exposed to temporally correlated classical noise. When these perturbations vary on timescales comparable to the system's intrinsic dynamics, rotating-wave and Markovian approximations can fail, making effective static descriptions unreliable.
arXiv QuantumRead article →arXiv:2610.40805v1 Announce Type: new Abstract: Quantum many-body scarred dynamics in Rydberg-atom arrays is characterized by revivals, in which the system repeatedly returns to its initial state despite its many-body complexity. But how extensively does the wave function explore the space of allowed configurations between revivals?
arXiv QuantumRead article →arXiv:2610.40776v1 Announce Type: new Abstract: Extracting late-time hydrodynamic transport properties directly from microscopic quantum models, especially in dimensions beyond one, is believed to be classically intractable. This owes, in part, to the extensive growth of entanglement in higher dimensions and often a mismatch between geometric and tensor network locality.
arXiv QuantumRead article →arXiv:2610.40762v1 Announce Type: new Abstract: Quantum synchronization holds promise in advancing quantum technologies such as sensing, communication, and computation. A key challenge in this rapidly developing field is to identify mechanisms that can promote the emergence of synchronous dynamics.
arXiv QuantumRead article →arXiv:2610.40709v1 Announce Type: new Abstract: Cooling a quantum system with $N$ thermal machines costs more than Landauer's bound. For swap-based collision models the excess is set by thermodynamic length, which is optimal for qubits and was suggested by numerics to be optimal in general.
arXiv QuantumRead article →arXiv:2610.40693v1 Announce Type: new Abstract: Electromagnetic analysis based on the finite element method is widely used in antenna design. At higher frequencies, however, the required mesh refinement makes solving large sparse linear systems the main contributor to computational cost.
arXiv QuantumRead article →arXiv:2610.41104v1 Announce Type: new Abstract: Self-distillation methods for large language models with an exponential moving average (EMA) teacher already retain a history of student updates. Muon nevertheless stores a separate momentum buffer, increasing persistent training memory.
arXiv MLRead article →arXiv:2610.41074v1 Announce Type: new Abstract: Low-precision training reduces computation but introduces additional rounding error. We theoretically study how stochastic rounding (SR) affects prediction risk as model size, training time, and precision change. We consider a two-layer linear network with a fixed random first layer, trained by mini-batch SGD on fresh Gaussian data with power-law spectra.
arXiv MLRead article →arXiv:2610.41073v1 Announce Type: new Abstract: Reinforcement learning (RL) post-training of language models keeps fp32 master weights in the optimizer and serves rollouts from a bf16 copy produced by round-to-nearest conversion. At the learning rates in common use, most per-step updates are smaller than half a bf16 spacing and the conversion discards them: on Qwen3-1.7B only 0.4-0.5% of served weights change per…
arXiv MLRead article →arXiv:2610.41059v1 Announce Type: new Abstract: Equivariant linear maps cannot mix non-isomorphic irreducible representations, whereas nonlinear activations can. We develop a degree-resolved version of this nonlinear interaction picture and use the resulting block inventory to construct Specht Triple, a residual on unordered graph triples.
arXiv MLRead article →arXiv:2610.41055v1 Announce Type: new Abstract: Federated reinforcement learning supports cooperative vehicular routing without centralizing trajectory data, but model updates may reveal private information and incur communication overhead over intermittent vehicle-to-infrastructure links, particularly when post-quantum authentication is required.
arXiv MLRead article →arXiv:2610.41046v1 Announce Type: new Abstract: Understanding which regulatory sequence motifs a genomic deep learning model has learned, with statistical error guarantees, is essential for trusting its use in downstream tasks such as variant effect prediction. Existing approaches to identifying which units in a trained network encode human- understandable concepts fall into two groups.
arXiv MLRead article →arXiv:2610.41041v1 Announce Type: new Abstract: Continual reinforcement learning (CRL) requires agents to operate under environmental changes while reusing prior knowledge whenever possible. We argue that model-based reinforcement learning provides a promising framework for this setting because learned world models (WMs) can, in principle, support change detection, knowledge reuse, adaptation, and acquisition of new…
arXiv MLRead article →arXiv:2610.41039v1 Announce Type: new Abstract: As Mixture-of-Experts (MoE) large language models are increasingly deployed in distributed environments, from federated learning to edge-cloud systems, their architecture harbors an underexplored landscape of security vulnerabilities.
arXiv MLRead article →arXiv:2610.41038v1 Announce Type: new Abstract: There exists little mathematical theory explaining why Transformer architectures are empirically successful even for non-Markovian data. To this end, this paper studies next-token prediction via stochastic optimal control theoretically trained Transformers for such sources, where we consider either stationary and ergodic non-Markovian sources or those which are hidden…
arXiv MLRead article →arXiv:2610.41023v1 Announce Type: new Abstract: Omniprediction concerns the task of making predictions that are simultaneously accurate with respect to different losses. Earlier results by [CV09] show that in an online learning setting, $O(\log(|\mathcal{F}||\mathcal{L}|))$ regret is achievable simultaneously against every predictor in a finite predictor class $\mathcal{F}$ and for every loss in a finite class…
arXiv MLRead article →arXiv:2610.41022v1 Announce Type: new Abstract: Latent world models allow reinforcement learning agents to learn policies from imagined trajectories, yet determining the optimal depth of model reliance remains a fundamental challenge.
arXiv MLRead article →arXiv:2610.41015v1 Announce Type: new Abstract: Expert scientists excel at weighing empirical findings from multiple sources that vary in both statistical reliability and epistemic relevance. To what extent are language models, which are increasingly driving autonomous discovery systems, capable of such evidence deliberation?
arXiv MLRead article →arXiv:2610.41014v1 Announce Type: new Abstract: Production large language models sometimes reward-hack, exploiting flaws in how tasks are scored rather than performing the tasks as intended, sometimes in highly sophisticated and adversarial ways.
arXiv MLRead article →arXiv:2610.41004v1 Announce Type: new Abstract: In multi-task settings, combining multiple modalities allows task representations to draw on signals that a single modality alone would not capture. However, individual tasks may depend disproportionately on particular modalities, a structure that conventional fusion-based architectures do not model explicitly, instead enforcing a unified representation space.
arXiv MLRead article →arXiv:2610.41001v1 Announce Type: new Abstract: Membership queries let a learner label arbitrary points and give near-optimal $\tilde O(n\log(1/\eps))$ label complexity for halfspaces under every marginal. A pool learner may label only sampled points. We ask when unlabeled data can replace membership queries. For homogeneous halfspaces the answer depends only on the angular law of $X/\|X\|$.
arXiv MLRead article →