arXiv:2606.25362v1 Announce Type: cross Abstract: Sequential contextual stochastic programs model real-time decision systems in which each time epoch commits to an action under uncertainty whose consequences propagate into future decisions. In many practical contexts, these programs require obtaining solutions rapidly as new information becomes available. These problems can be represented through scenario approximations to be solved by off-the-shelf optimization solvers, which achieve high decision quality offline but typically run in seconds to minutes per instance, falling short of the sub-second responses that peak periods of planning require. This paper develops a learning-based optimization proxy: a scenario-embedded neural network trained offline on solver-generated labels, paired online with a decoder that enforces feasibility, replacing the per-epoch solve with a single forward pass. The framework is specialized to omnichannel order fulfillment, where each arriving order requires a sub-second assignment of products to distribution centers and carrier services under stochastic delivery times and future demand. A two-stage contextual stochastic program is introduced to formulate this problem, and its contextual sample average approximation (C-SAA) supplies the offline labels, while a composite training loss combines label imitation, a constraint-violation penalty, and self-supervised cost alignment. In a calibrated simulator built from JD.com transactional records, a detailed computational study is provided. The proxy reduces decision latency by roughly 2800x relative to the online finite-sample C-SAA reference and improves over it by 3.3% in realized fulfillment cost. Relative to established fulfillment policies, the proxy lowers total realized cost by at least 10.7% and roughly halves the late-delivery rate.
Science Journals
arXiv:2606.25460v1 Announce Type: cross Abstract: Recent advances in sequence modeling have significantly improved ASR systems, bringing them close to human-level recognition accuracy and enhancing robustness across diverse acoustic conditions and languages. In contrast, Forced Alignment has not experienced comparable progress, and traditional HMM-GMM frameworks remain widely adopted and highly competitive. To address this gap, we propose an end-to-end, fully differentiable neural architecture specifically designed for phoneme alignment. The model consists of an encoder that processes the input signal and a decoder that produces alignment decisions. The encoder is structured into two complementary branches: one dedicated to phoneme identity verification and the other to phoneme boundary detection. The decoder is implemented as a trainable module based on differentiable soft dynamic programming. The entire system is optimized end-to-end using a novel contrastive loss that encourages clear separation between steady-state phoneme regions and transition boundaries. The proposed approach outperforms the current state of the art in phoneme alignment on hand-annotated English benchmarks, achieves strong word-level generalization results, and demonstrates generalization on unseen languages.
arXiv:2606.25502v1 Announce Type: cross Abstract: Discovering solid-state materials with target properties remains a central challenge in computational materials science. Existing approaches -- high-throughput screening, surrogate optimization, and generative models -- require extensive evaluations or training data and extrapolate poorly to unseen compositions. Here we develop a first-principles inverse-design framework, integrating reverse-mode automatic differentiation (AD) into KKR-CPA -- the Korringa--Kohn--Rostoker method with the coherent potential approximation -- where atomic compositions are continuous variables to be optimized. Reverse-mode AD yields gradients of objective functions with respect to composition at a cost independent of the number of candidate elements, enabling gradient-based optimization to identify materials from compositional spaces spanning dozens of elements. In this framework, any computable quantity can serve as the objective. We demonstrate this generality through two contrasting applications, magnetic alloys and half-metals, yielding candidates such as (Lu$_{0.553}$Yb$_{0.447}$)(Co$_{0.759}$Fe$_{0.241}$)$_2$Fe$_3$ and FeZr(Sb$_{0.94}$Te$_{0.06}$). Our framework offers a physically grounded route from a target property to the material that realizes it.
arXiv:2606.25511v1 Announce Type: cross Abstract: The cavity-magnon-qubit system has recently been demonstrated as a new platform for preparing macroscopic quantum states in magnonic systems. Here, we propose to prepare a two-mode magnonic cat state, which is also a non-Gaussian entangled state, based on this practical system involving two yttrium-iron-garnet (YIG) spheres and a superconducting qubit coupled to a common microwave cavity. By adiabatically eliminating the cavity and resonantly driving the qubit, an effective magnon-qubit conditional-displacement interaction is achieved. Further working in the magnon-magnon strong-coupling regime and considering two identical magnon frequencies and coupling strengths to the cavity, two hybridized magnon modes are formed, of which the bright mode is prepared in a cat state after a projective measurement on the qubit, while the dark mode remains in its initial vacuum state. Such a state corresponds to a two-mode cat state of two original magnon modes, which share strong non-Gaussian entanglement. We also discuss practical dissipation and dephasing effects on the cat state. The results indicate that strong nonclassicality and non-Gaussian entanglement are present in the two-mode cat state using fully feasible parameters.
arXiv:2606.25615v1 Announce Type: cross Abstract: In this paper, we consider a parabolic problem with constraint written as a differential inclusion, driven by a multiplicative colored noise and involving a p-Laplace operator (for $p \geq 2$), nonlinear random source terms and subject to Neumann boundary conditions on a bounded Lipschitz domain of $R^d$ with $d \geq 1$. This contribution aims at proving existence and uniqueness of a solution for such a multivalued problem. On one hand, the existence result is proved by the analysis of a semi-implicit time discretization scheme constructed on a smoother version of our problem, itself obtained by a regularization "\`a la Moreau-Yosida" of the subdifferential term. The key point of our approach consists in finding a clever relation between the time step denoted $\tau$ and the Moreau-Yosida regularization parameter denoted $\epsilon$ in view to pass simultaneously to the limit with respect to $\tau$ and $\epsilon$. On the other hand, the uniqueness of the solution is proved by standard arguments.
arXiv:2606.25666v1 Announce Type: cross Abstract: Classical invisibility cloaks are designed to suppress selected scattering signatures and thereby make an object appear absent to external electromagnetic probes. However, the suppression of a classical scattering observable does not, by itself, establish that all information about the concealed object has been removed from the detected quantum state of light. Here we formulate the detectability of classically cloaked objects as a quantum-state distinguishability problem. Treating a linear passive cloak as an effective Gaussian quantum channel acting on the accessible detected modes, we show that local quantum undetectability requires the detected first and second moments to be independent of the hidden-object parameter. In this framework, quantum Fisher information provides an operational criterion for whether the concealed parameter remains estimable from the detected output state. We derive displacement- and covariance-level detectability conditions and show that a nonzero parameter imprint surviving in the detected Gaussian state leads to a nonzero accessible quantum Fisher information. To connect the criterion with a physical cloaking model, we analyze a regularized cylindrical transformation-optical cloak in the Born limit and compare the scaling of the classical scattering response with the derivative-based quantum sensitivity. The analysis shows that reducing a scattering amplitude is not equivalent to eliminating local quantum-state sensitivity. Loss, environmental noise, and finite numerical aperture degrade the accessible information, but quantum undetectability is reached only when the parameter imprint is removed from the detected state or projected entirely outside the accessible subspace. These results provide a Gaussian-channel framework for assessing when classical cloaking does, and does not, imply quantum-state undetectability.
arXiv:2606.25813v1 Announce Type: new Abstract: Traditional particle-based kinetic methods, such as DSMC, suffer from prohibitive computational cost in low-signal flows, where the deviation from thermodynamic equilibrium is small and statistical noise overwhelms the signal of interest. The Variance-Reduced BGK-DSMC scheme is further advanced and implemented to support this class of flows in the open-source gas-kinetics framework PICLas. Modified versions of flow estimators and collision operators enhancing stability are developed. The Shakhov and Ellipsoidal Statistical models for BGK are demonstrated, along with entirely new features such as adaptive equilibria, variable particle weights and domain axisymmetry. The implementation is validated using synthetic benchmarks, 1D, 2D and axisymmetric simulations. Comparison of VRBGK to BGK simulations shows exact agreement of the models. A further comparison with an analytical solution of thermal transpiration in a microchannel showcases the low-signal efficiency of the method as well as newly proposed features.
arXiv:2606.25816v1 Announce Type: new Abstract: Motivated by applications for robust guarding, we consider a variant of the multiple-watchmen problem that ensures that every point within a polygon $P$ is seen from more than one direction: we search for two routes $W_1,W_2$, such that every point $p\in P$ is contained in a segment $\overline{w_1w_2}\subseteq P$ such that $w_1\in W_1$ and $w_2\in W_2$. We call such routes segment watchman routes. We show that finding the two routes that are optimal with respect to the min-max criterion is weakly NP-hard even in simple polygons, and that finding the routes that are optimal with respect to the min-sum criterion is NP-hard in polygons with holes. Moreover, we present sufficient conditions for routes to be segment watchman routes, and provide a polynomial-time $2$-approximation under both the min-max criterion and the min-sum criterion, both in simple polygons. Finally, we show how to generalize our results for $k$ watchmen.
arXiv:2606.25768v1 Announce Type: cross Abstract: Multi-mode interferometers for bosons in Gaussian states are important systems for quantum metrology with precision beyond the standard quantum limit and for bosonic quantum computing. However, there is a lack of theoretical foundation for generic $N$-mode Gaussian interferometry. In this work, we study quantum metrology and quantum control in multi-mode bosonic systems with quadratic Hamiltonians, exploiting the fundamental Sp$(2N,R)$ symmetry of such interferometers. We show that the optimal quantum control to maximize sensitivity requires aligning squeezing and displacement in the same direction. We propose Sp$(2N,R)$ echo, a multi-mode generalization of the SU$(1,1)$ interferometry, to achieve the sensitivity of phase estimation set by the quantum Fisher information. In addition, we introduce a geometrical means for reversing many-body dynamics with Sp$(2N,R)$ dynamical symmetry, such as dynamics of the bosonic Kitaev chain. Our schemes are readily realizable in optical, atomic, and mechanical platforms.
arXiv:2606.25808v1 Announce Type: cross Abstract: We propose a predict-optimize-explain framework that uses gradient-based sample generation to interpret various portfolio models by identifying macroeconomic conditions that induce specified portfolio outcomes. Unlike traditional feature-importance methods, this approach directly probes decision pipelines (predictive models coupled with portfolio optimization) by constructing economically meaningful what-if questions. We focus on four such questions: under what macroeconomic conditions a predict-then-optimize pipeline closes or reverses its return gap with a predict-and-optimize pipeline; what conditions lead a pipeline to diversify rather than concentrate its allocation; when a pipeline trained on calm markets overtakes one trained through crises; and what conditions would let a pipeline match a benchmark return. These examples illustrate how our framework uncovers key behavioral differences between various decision pipelines. Beyond these cases, the proposed framework is flexible and can support a wide range of probing questions tailored to specific portfolio objectives. Our findings highlight the value of integrating prediction, optimization, and explanation to produce more robust and transparent portfolio strategies.
arXiv:2606.25820v1 Announce Type: new Abstract: We present an operator learning framework based on a coarse data-driven multiscale space for nonlinear flow in random heterogeneous porous media. The multiscale space is constructed from local representative fine-scale solution snapshots, yielding an accurate low-dimensional representation of the solution manifold. This multiscale basis serves as the trunk of a neural operator, while a branch network predicts the corresponding reduced coefficients from the input permeability field. Unlike Galerkin projection methods, the neural operator learns a global nonlinear mapping from permeability fields to solution coefficients, providing greater flexibility, improved accuracy, and eliminating the need for online nonlinear coarse-grid solves and coefficient evaluations. Numerical results show that the proposed approach achieves good accuracy and substantially lower computational cost than projection-based methods for nonlinear flow in high-contrast heterogeneous media.
arXiv:2606.25828v1 Announce Type: new Abstract: Direct motional coupling of opposite-charge trapped-ion pairs could provide a pathway to extend ultra-low temperatures and quantum control to negative ions that lack the suitable electronic energy structures required for direct laser cooling. Because positive and negative ions cannot be confined within a single electrostatic potential well, I investigate a configuration where single ions are trapped in close proximity within separate potential wells to couple their motion. I analytically and numerically evaluate the electrostatic trapping requirements. As a concrete implementation, I present an optimized segmented surface Paul trap design to couple an antimatter hydrogen molecular ion ($\bar{H}_2^-$) and a beryllium ion ($^9 Be^+$). A motional coupling frequency of 5 kHz can be achieved at an ion-ion separation of $35 \mu m$, with an ion height of $50 \mu m$, axial trap frequencies of 4 MHz, and static trap voltages with a magnitude of $\approx 20 V$. Finally, I outline three applications for this technique: quantum logic spectroscopy of $\bar{H}_2^-$ for matter-antimatter comparisons, the preparation of cold neutral deuterium atoms via near-threshold photo-detachment of $D^-$ for optical trapping, and quantum information processing using equal-mass opposite-charge ion pairs.
arXiv:2606.24892v1 Announce Type: new Abstract: Peer review is central to scientific quality control, yet it can undervalue papers that later achieve substantial citation impact. While frontier large language models have shown promise in automating aspects of peer review, they primarily mimic human reviewer preferences rather than predict long-term scientific value. We introduce ReviewGuard, a two-stage framework that aligns LLM-generated reviews with citation-based estimates of long-term scientific impact rather than contemporaneous reviewer judgments. On 20,861 AI/ML papers from OpenReview augmented with Semantic Scholar citation data, ReviewGuard achieves a Spearman correlation of \r{ho} = 0.776 with future citations on rejected-then-published papers, outperforming human reviewers (\r{ho} = 0.492) and a supervised Expert model (\r{ho} = 0.681). Under the same decision threshold, ReviewGuard flags 10.2% of high-impact rejected papers, compared with 1.8% for human reviewers, corresponding to a 5.6x improvement. Our results demonstrate that impact-aligned reinforcement learning can provide editors with a complementary signal for identifying high-potential work, without replacing human judgment.
arXiv:2606.25829v1 Announce Type: new Abstract: Reliable perception in adverse conditions remains challenging for autonomous systems, as cameras and LiDAR degrade in poor lighting or weather. Millimetre-wave FMCW radar is robust to such conditions, but its elevation collapse limits geometric reasoning. We observe that vehicle chassis occlude radar rays and form a distinctive geometric shadow, and its consistency can enable us to infer useful information about objects whose returns intersect this shadow. Motivated by this observation, we propose a method to recover the 3D, in-plane inclination of nearby slender vertical objects from this cue. The object inclination is retrieved without assumptions about the wider scene, but through an analytical, closed-form mapping between its radar return boundaries and the opening angle. Validation in simulation and experimentation on a Navtech CTS350-X radar shows that inclinations can be estimated under practical conditions, with segmentation of the object in the radar scan emerging as the main bottleneck. This work highlights chassis shadows as a novel geometric cue, extending the role of 2D rotating radar beyond localisation and toward 3D scene reconstruction.
arXiv:2606.25817v1 Announce Type: cross Abstract: The IC4Stars (Intensity Correlation for Stars) project aims to measure the diameter of the white dwarf star Sirius B, using Intensity Interferometry. In this work we present our latest efforts and the milestones achieved in the last year. We report laboratory characterization of the single-photon detectors, TDC and synchronization electronics. We describe an observation campaign where we demonstrated on-sky time resolution below~35~ps~RMS, synchronizing two TDCs using the White Rabbit protocol and a~30~m telecom fiber. We developed the data acquisition of the raw time tags, and an algorithm to compute the second-order correlation function.
arXiv:2606.25852v1 Announce Type: new Abstract: Group-based reinforcement learning effectively post-trains LLM agents for long-horizon, sparse-reward tasks by deriving step-level credit from trajectory outcomes. However, this ties a step's credit to its rollout's final outcome: semantically near-identical intermediate steps receive opposite credit depending on whether their trajectory eventually succeeded or failed. Such semantic credit inconsistency sends conflicting gradients to similar actions and wastes the partially-correct progress inside failed rollouts. Motivated by this, we propose Semantic Consistency Policy Optimization (SCPO), a value-free reward-shaping method that mitigates this inconsistency by recovering step-level credit from successful siblings in the same rollout group. Concretely, SCPO scores each failed step against a successful sibling and adds positive step-level credit for new progress along that sibling. On ALFWorld and WebShop, SCPO matches or exceeds strong group-based baselines, reaching 93.7+/-4.1 percent success on ALFWorld and 74.8+/-2.0 percent on WebShop at 1.5B parameters, with gains concentrated on the hardest multi-step tasks.
arXiv:2606.24898v1 Announce Type: new Abstract: Looped language models turn hidden states into runtime state: each state is decoded for prediction and fed back into future computation. This creates a basic supervision question: which state variables does cross-entropy actually control? We show that dense per-loop cross-entropy controls the variables exposed by the readout, not every variable active in the recurrent transition. Hidden-state scale gives a concrete failure mode. Scale-invariant readouts such as RMSNorm and LayerNorm hide radial scale from the immediate cross-entropy loss, while pre-norm residual recurrence continues to carry and update that same scale. Thus per-loop loss can make early exits usable without controlling recurrent scale. In 44M and 129M looped transformers without inter-loop normalization, per-loop cross-entropy through RMSNorm readouts still drives final hidden-state norms into the thousands or tens of thousands. Scale-visible readouts and explicit norm penalties keep norms in the tens, and scale-removing recurrence is the complementary architectural fix. The resulting design rule is simple: dense supervision trains exits; recurrent scale control requires either making scale visible to a loss or removing it from the loop. Consistent with this rule, scale-controlled variants achieve lower perplexity at matched inference-depth operating points in our variable-depth benchmarks.
arXiv:2606.24933v1 Announce Type: cross Abstract: Recent advances in quantum machine learning have motivated efficient models for sequential data processing. In this paper, we propose Self-Modulating Quantum Fast Weight Programmers, or Self-Modulating QFWP, which extends Quantum Fast Weight Programmers by introducing adaptive modulation over both newly generated fast-weight updates and historical fast-weight memory. Numerical results show that the proposed mechanism improves convergence stability and prediction performance across varying model settings, including different numbers of qubits and input sequence lengths. We further provide theoretical arguments explaining how self-modulation balances new information injection with memory retention, thereby enhancing temporal information propagation. These results suggest that Self-Modulating QFWP is a compact and effective framework for quantum machine learning on time-series data.
arXiv:2606.25945v1 Announce Type: cross Abstract: The increasing number of orbital and sub-orbital launches makes it necessary to investigate the environmental impacts of launch vehicles and incorporate eco-design considerations into their development. In response, the European Space Agency has promoted Life Cycle Assessment (LCA) as a standardization methodology to mitigate environmental impacts of present and future space missions. This need is further amplified in the NewSpace, where numerous configurations and innovative technologies are explored, reinforcing the importance of integrating environmental considerations. At early design stages, launch vehicle architecture can be formalized through a multi-physics optimization problem based on Multidisciplinary Design Analysis and Optimization (MDAO) methods, where disciplines such as propulsion, aerodynamics, structure, and trajectory are coupled to obtain trade-offs among candidate configurations. This paper proposes a methodology to integrate an LCA discipline within an MDAO framework for launch vehicle design. The approach relies on parametric life-cycle inventories depending on design and coupling variables, covering component and propellant production as well as transport to the launch site. Launch emissions are evaluated from optimized trajectory profiles and characterized in terms of climate change impact. The methodology is illustrated on a representative expendable launch vehicle, where multi-objective optimizations assess trade-offs between performance and environmental indicators. Results highlight antagonistic behaviors among environmental impact categories, emphasizing the importance of carefully defining environmental objectives in eco-design studies. The generic nature of the methodology lays the foundation for integrating LCA into early-stage launch vehicle design, enabling exploration of trade-offs between performance, cost, and environmental considerations.
arXiv:2606.25946v1 Announce Type: cross Abstract: Van der Waals materials are ideally suited for the implementation of high-frequency nanophononic resonators with atomically flat interfaces. Here, we present two versatile van der Waals-based nanophononic architectures: First, we introduce self-supporting nano-domes of WSe$_2$ as a scalable platform for the simultaneous generation of hundreds of high-quality nanoacoustic resonators with resonance frequencies in the 100 GHz range. Second, we engineer self-supporting nano-drums that reach record-high working frequencies for 2D-semiconductor transducers beyond 1 THz. Through optical pump-probe spectroscopy experiments and photoelastic linear chain model calculations, we gain a detailed understanding of the intricate interplay between phononic mode hybridization across heterostructures, the differences between modes close to the center and edge of the acoustic Brillouin zone, and the temporal structure of the photoelastic response. Both architectures have potential applications in low-cost nanoacoustic probing and the ultrafast modulation of quantum emitters in two-dimensional semiconductors. While nano-drums surpass the THz frequency barrier, nano-domes appear as an accessible, low-cost alternative for developing scalable nanophononic technologies.
arXiv:2606.25954v1 Announce Type: cross Abstract: This theoretical note studies the finite axiomatizability of strict majority reasoning in finite social decision frames. Moss and Pedersen (2026) introduce a coherence criterion that characterizes exactly when qualitative majority judgments are representable by a finitely additive measure. The question addressed here is whether that coherence criterion can be replaced, in the finite setting, by any bounded finite fragment. We prove that it cannot. For every $k\ge 1$, we construct a maximal standard frame whose shortest coherence violation has length exactly $2k+2$. Hence there is no uniform finite bound on the incoherence index of social decision frames, resolving Conjecture 5.7 stated by Moss and Pedersen (2026). The construction is geometric, in the sense that it proceeds via orthogonality and dimension in rational vector spaces, and self-contained: it isolates a symmetric family of half-sized voting blocs and extends it to a maximal frame in which every shorter balanced obstruction is excluded. Along the explicit infinite sequence of universe sizes obtained in the construction, this also establishes the middle-layer family predicted by Conjecture B.25 by Moss and Pedersen (2026). Together with the soundness and completeness theorem for the Moss-Pedersen minimal logic for strict majorities, this establishes that measurable social decision frames are not finitely axiomatizable in that language.
arXiv:2606.24916v1 Announce Type: new Abstract: Memory bandwidth accounts for 30-40% of total power consumption in standalone virtual reality (VR) headsets, yet existing systems typically store the entire 360-degree frame at a uniform resolution regardless of viewer gaze. This paper presents SPORT (Spherical-PSNR Optimized tRuncaTion), a bit-truncation framework that reduces display-path memory power by storing only the most significant bits of pixels outside the user's field of view (FoV). Specifically, a new bit-truncation framework is developed to use weighted-to-spherically-uniform PSNR (WS-PSNR) directly in the optimization constraint, eliminating the metric inconsistency that arises when standard PSNR is used for a WS-PSNR quality target. Also, gaze-predictive tile classification compensates for the 9.33 ms end-to-end pipeline latency, reducing boundary misclassifications by 5.2 percentage points at a cost of only 0.01 ms. In addition, the developed SPORT-B variant, which keeps the FoV lossless, achieves 47.9% memory power saving and 47.9% bandwidth reduction across different 4K video sequences while satisfying all three per-region WS-PSNR thresholds and maintaining SSIM = 1.000 in the attended region. The full adaptive variant SPORT-A reaches 51.6% power saving, 3.1percentage points more than a PSNR-based optimizer at equal measured quality. SPORT is validated on the TrunMEM360 flexible SRAM Application-Specific Integrated Circuit (ASIC) fabricated in SkyWater 130 nm CMOS, confirming byte-exact silicon-software agreement, with WS-PSNR and SSIM matching within 0.1 dB and 0.001. CACTI-based analysis confirms 48.72% DRAM leakage reduction and 36.4%/36.7% read/write energy reduction. The total motion-to-photon latency of 9.33 ms satisfies the 20 ms VR comfort budget with a 53.3% safety margin.
arXiv:2606.24927v1 Announce Type: new Abstract: Inspired by Nuel Belnap's "How a computer should think," which gave rise to the four-valued logic FDE, we contemplate, in this article, how a computer might think if we add a fifth value for unknowable or cannot be known. We devise two new five-valued logics, UKN1 and UKN2, called the logics of the unknowable. These are different from the five-valued logic FDEe of the FDE-family. The main difference is in the number of designated truth values. While FDEe takes two designated values, UKN1 and UKN2 have three. The four-valued reducts of these logics are also different from FDE. This is due to the fact that instead of taking one of the non-designated values as neither true nor false, as in FDE, we have interpreted this as not known yet. This value although denoted by the same letter $n$, behaves differently from its namesake in FDE.
arXiv:2606.24934v1 Announce Type: new Abstract: Cloud GPU tenants receive a model name and a region, but cannot directly inspect the physical accelerator that runs their job. We present a software-only attestation primitive for this setting. A CUDA probe measures an SM-by-memory-region latency matrix using physical SM labels and dependent global loads. A streaming reducer commits sufficient statistics, configuration, code hashes, network evidence, and a compressed raw data archive into a certificate that a verifier can check without a GPU. The certificate supports three claims. First, the per-SM latency map is a stable physical fingerprint. Over a six-hour full-load RTX 5090 run, its median temporal jitter is 0.09 cycles, while shape-only leave-one-out classification separates distinct Blackwell dies with 100.0% accuracy. Second, cache-bypassing HBM sweeps recover hardware-class topology across generations, including a unified Volta V100 memory domain, a two-way Hopper H200 L2 split, and a Blackwell B200 two-die NV-HBI package whose 74/74 SM partition carries a 30-cycle, 15.5 ns cross-die penalty. Third, public network landmarks bind the same certificate to a coarse location. In the B200 run, 169 RIPE Atlas probes place the server within 44 km of its claimed datacentre and reject all 11 decoy sites. Together, these measurements check cloud-GPU identity, class, and coarse location without privileged access or a vendor key.
arXiv:2606.24947v1 Announce Type: new Abstract: The increasing integration of distributed energy resources (DERs) is crucial for power system decarbonization, yet unlocking DERs' flexibility is challenged by their inherent uncertainties and modelling complexity. As traditional optimization methods struggle with such uncertainty and complexity of DERs, reinforcement learning (RL) has emerged as a promising alternative for DER management. However, standard RL methods suffer from sample inefficiency and sub-optimality when trained from scratch. Inspired by the training paradigms in large language models, this paper proposes a Supervised Reinforcement Learning (SRL) framework for learning DER coordination policies. This framework first pre-trains a policy on demonstration data in a supervised-learning fashion, which is then further fine-tuned using RL. Furthermore, we propose a two-step fine-tuning process: offline fine-tuning for enhancing policy performance and online fine-tuning for adapting it to the real-world dynamics. Experiments demonstrate that RL implementations based on the proposed framework significantly outperform all benchmarks, achieving high cost efficiency even under low-quality demonstration data.