arXiv:2607.09238v1 Announce Type: new
Abstract: A new family of ionic basis sets, denoted i{\sigma}XZ1, is presented for molecular calculations on systems containing monovalent ions. The basis sets extend the SIGMA family by explicitly accounting for the different electronic structure of cations and anions. Auxiliary basis sets for resolution-of-the-identity calculations are also developed for the Coulomb term. The performance of the proposed basis sets is assessed for alkali-halide clusters. Compared with conventional basis sets, i{\sigma}XZ1 provides an accurate description of structural, energetic, and electronic properties while showing remarkable robustness against near-linear dependencies, allowing stable calculations on systems containing up to 792 NaCl units. These features make the i{\sigma}XZ1 family a reliable and efficient alternative for large-scale calculations on ionic systems.
Science Journals
arXiv:2603.01115v2 Announce Type: replace
Abstract: General vision foundation models (VFMs) have been primarily developed on natural images, and their utility for medical image segmentation is therefore often considered to depend on costly adaptation or domain-specific fine-tuning. In this paper, we revisit this assumption from a different perspective: rather than requiring VFM segmentors to relearn visual regularities, we investigate whether the low-level visual priors necessary for anatomical delineation already lie dormant within general VFMs. We observe that frozen VFMs, despite lacking medical supervision, encode transferable visual regularities. These properties are not exclusive to natural images but are also fundamental to medical image understanding. Motivated by this observation, we propose Latent Universal Medical PriOrs for Segmentation (LUMOS), a novel framework that amplifies general VFM priors to conventional medical segmentors. LUMOS consists of two key components: (1) Pathfinder that distills visual cues from a frozen vision foundation model, and (2) Inspiror that sparks the conventional medical networks with spatial guidance from distilled visual regularities. In this way, the segmentor is relieved from learning complex visual regularities entirely from limited medical annotations and can instead focus on task-specific anatomical delineation. Across diverse medical datasets and token-based VFMs, LUMOS shows that general VFMs can serve as spatial prior generators when their frozen token spaces preserve patch-level pattern relevance. DINO provides stable matched-backbone gains, while SigLIP exposes VFM-specific sensitivity caused by its different token granularity and representation objective.
arXiv:2606.20858v2 Announce Type: replace
Abstract: The temporal structure of reward composition in reinforcement learning (RL) is typically hand-designed and held fixed throughout training, leaving the progression of motivational priorities largely unexplored. In this work, we propose an evolutionary framework for discovering developmental reward schedules, in which three distinct biologically inspired motivational components -- agency, novelty, and reactivity -- are combined through time-varying weights that dynamically shift over the course of training. Evaluated on two sparse-reward MiniGrid tasks: DoorKey-6x6 and KeyCorridorS3R1, our framework compares the generalizability of four evolutionary algorithms: CMA-ES, xNES, DE, and L-SHADE against an extrinsically motivated baseline (our main comparison point), and three additional hand-designed methods. On DoorKey-6x6, all evolved methods outperform the non-evolved baselines, with L-SHADE achieving the best performance -- an approximate relative mean improvement of 11.4% over the extrinsic only baseline. On KeyCorridorS3R1, CMA-ES achieves the best overall performance, with the remaining evolved methods showing weaker and less reliable generalization capability compared to the extrinsic only baseline. Interestingly, the discovered schedules diverge from our defined developmental ordering, with novelty consistently emerging as the dominant early signal during training, across both tasks. Collectively, our results position evolutionary optimization as a promising approach for developmental reward schedule discovery in deep reinforcement learning, and suggest that what evolution finds to be optimal in computational settings may differ from what it finds to be optimal in biology. The code for this project can be found at: https://github.com/alannadels/Evolutionary_RL.git.
arXiv:2606.28225v2 Announce Type: replace
Abstract: Temporal link prediction (TLP) is typically evaluated by predictive performance on unseen edges, but this criterion can conflate predictive accuracy with recovery of the underlying causal mechanism. In stochastic models, Fisher information governs the Cram\'{e}r--Rao (CR) bound on parameter estimation error: higher Fisher information permits more accurate parameter recovery. We show that, under comonotonicity conditions between Fisher information and entropy, binary logistic models exhibit an estimation--prediction tradeoff: regimes with higher Fisher information, and hence smaller CR bounds, also have higher irreducible predictive entropy. To study this tradeoff in TLP, we introduce a probabilistic causal generator for temporal graphs with transient edges and known ground-truth causal structure, and validate the phenomenon empirically.
arXiv:2607.09065v1 Announce Type: new
Abstract: Software engineering (abbrev. SE) has continuously evolved through increasingly powerful forms of reuse, from source code and libraries to components and services. Recent advances in AI agents have introduced a potentially new reusable artifact: skills. Emerging agent skill repositories and marketplaces enable developers to package, share, and reuse SE expertise as reusable skills. This trend raises a fundamental question: what SE activities are being encapsulated into reusable skills? Existing studies primarily focus on a broad range of skills acquisition, safety, or benchmarking, while lacking a systematic understanding of SE-specific skills and their coverage across the software development lifecycle. To address this gap, we conduct the first large-scale empirical study of SE skills in public repositories and marketplaces. We collect and analyze a large corpus of SE skills, examining the activities they encapsulate, lifecycle coverage, evolution characteristics, and evaluation mechanisms. Our findings reveal that SE activities are increasingly becoming reusable artifacts via skills and suggest promising research opportunities for skill recommendation and engineering-oriented structuring, as well as the need for mechanisms to encapsulate high-context SE activities into reusable skills. Overall, our study provides the first activity-centric characterization of SE skills and reveals how SE activities are increasingly being transformed into reusable skills. These findings offer new insights into skill reuse, ecosystem development, and the future of agent-centric SE.
arXiv:2607.09405v1 Announce Type: new
Abstract: Similarity search is a primary application of embedding models trained by contrastive learning. For one of the most popular contrastive learning loss functions, InfoNCE, we show that the population risk with $k$ negative samples is $O(1/k)$ close to an expected cross-entropy which quantifies deviation between i) a softmax similarity search over unseen data using the learned embedding function, and ii) an idealised softmax search over the same data but using similarity implicitly represented in the positive sample generator. This complements existing interpretations of InfoNCE in the $k\to\infty$ limit which are phrased in terms of mutual information, and alignment versus uniformity in embeddings. To quantify generalisation performance, we introduce a new continuity bound for the InfoNCE loss, obtained via G\^{a}teaux differentiation. The bound preserves the structure of averaging over negative samples present in the loss function and features an ``inverse temperature'' parameter which can be tuned to account for the algorithmic temperature. For embedding functions which are Lipschitz in a parameter, this yields a simple demonstration that the averaging effect of $k$ negative samples in the InfoNCE loss carries over to stabilisation of the generalisation error as $k$ grows.
arXiv:2607.09143v1 Announce Type: new
Abstract: Fusing standard RGB frames with asynchronous event streams has emerged as a definitive paradigm for robust perception in degraded environments. Although unified backbones have recently gained traction in multi-modal vision, adapting them to the RGB-Event domain remains fundamentally challenging. Existing architectures either resort to decoupled dual encoders that double computational overhead, or adopt generic unified designs that fail to resolve implicit geometric parallax and cross-spectral aliasing under the extreme representational divide between dense intensity grids and sparse kinematic spikes. To transcend these bottlenecks, we present Evita, the first unified backbone specifically engineered for dedicated dense RGB-Event parsing. To achieve profound modal synergy, Evita explicitly embeds a suite of intrinsic co-learning modules directly into every encoder layer. Specifically, it features Geometric Parallax Rectification for adaptive spatial alignment, Harmonic Spectral Resonance for texture transfer exclusively in the complex frequency domain, and Transient Global Routing for event-driven asymmetric attention. To guarantee robust feature extraction against spatial misalignments and decouple representations from specific event encodings, we construct N-ImageNetV2 alongside a stochastic event representation mixing pretraining protocol, empowering the network to seamlessly accommodate arbitrary event formats in downstream tasks. Extensive evaluations across the DELIVER, DDD17, and DSEC benchmarks confirm that Evita establishes new state-of-the-art metrics while delivering a superior accuracy-latency trade-off for real-time multimodal perception.The code are publicly available at: https://github.com/chaineypung/Evita.
arXiv:2210.12643v2 Announce Type: replace-cross
Abstract: We consider the algorithmic decision problem that takes as input an $n$-vertex $k$-uniform hypergraph $H$ with minimum codegree at least $m-c$ and decides whether it has a matching of size $m$. We show that this decision problem is fixed parameter tractable with respect to $c$. Furthermore, our algorithm not only decides the problem, but actually either finds a matching of size $m$ or a certificate that no such matching exists. In particular, when $m=n/k$ and $c=O(\log n)$, this gives a polynomial-time algorithm, that given any $n$-vertex $k$-uniform hypergraph $H$ with minimum codegree at least $n/k-c$, finds either a perfect matching in $H$ or a certificate that no perfect matching exists.
arXiv:2607.08665v2 Announce Type: replace
Abstract: Routing among large language models (LLMs) trades response quality against serving cost, motivated by the reported gap between deployed routers and a per-instance oracle. Recent analysis shows that test-time resampling can recover per-instance selection headroom that no single-commit router captures; however, that guarantee holds only under an idealized oracle equipped with correctness labels and an unconstrained budget, neither of which a deployed system has. To the best of our knowledge, no previous work treats resampling the committed model and rerouting to an alternative model as competing uses of a single per-query cost budget. Therefore, this work formulates budget-aware test-time model selection: given a per-query budget and an imperfect verifier, allocate each unit of budget between resampling and rerouting so that expected correctness is maximized. An online resample-or-reroute (RoR) allocation policy driven by estimated marginal correctness per unit cost is proposed, and its behavior is grounded in the recoverability asymmetry between selection and sampling. Replay experiments on newly regenerated multi-draw correctness tensors from an eleven-model open-weight pool over four benchmarks of differing difficulty show that the proposed RoR policy attains a favorable cost-quality Pareto front relative to single-route, one-commit-router, budget-aware best-of-K, cascade, and random-allocation baselines for the tested pools, with the largest gains on the most heterogeneous benchmark; an ablation further shows the gains are verifier-gated, shrinking as verifier quality degrades, and robustness replays under a provider price vector and a label-free agreement verifier delineate where the conclusions carry over.
arXiv:2606.24955v2 Announce Type: replace
Abstract: Power forecasting models deployed in real-world energy markets must operate under nonstationary conditions, where data distributions continually evolve due to weather variability, infrastructure upgrades, and changing consumption behaviors. In practice, these models face strict operational constraints: historical data may be limited or unavailable for repeated retraining, and uninterrupted long-term service is often required. This paper addresses these challenges by proposing the paradigm of Continuous Power Forecasting, which views power forecasting as a continual learning problem rather than a static offline task. Based on an adaptive continual learning framework for regression, we systematically investigate the practical effectiveness of six representative continual learning approaches from three methodological categories. These approaches are evaluated under different realistic assumptions regarding data accessibility and update policies. Experimental validation on real-world power datasets demonstrates that continual learning enables forecasting models to self-adapt to distributional drift, accumulate knowledge over time, and mitigate catastrophic forgetting without relying on large-scale historical data storage. Beyond performance gains, our study provides practical insights into the stability and adaptation behaviors of different continual learning approaches under realistic operational constraints. Overall, this work illustrates how continual learning can be pragmatically integrated into industrial power forecasting pipelines, offering a scalable and sustainable solution for long-term deployment in dynamic environments.
arXiv:2607.09349v1 Announce Type: new
Abstract: Retrieval-augmented generation evaluation checks whether model claims are factually grounded in retrieved documents. It does not check whether retrieved evidence is attributed to the correct entity.
A clinical RAG response can pass every automated check (zero hallucinations, near-perfect faithfulness, real citations) while presenting drug Y's clinical evidence as evidence about queried drug X. We term this deceptive grounding (DG): a failure invisible to faithfulness, hallucination, and citation checks because every claim is sourced from a real document, about the wrong entity.
Using a controlled factorial benchmark across 13 models, we find DG rates spanning 8-87% at peak adversarial conditions. Medical and biomedical fine-tuned models reach up to 86.7%; domain specialization amplifies the failure rather than mitigating it.
A controlled ablation identifies the mechanism: removing entity-specific clinical evidence from retrieved documents eliminates entity-attribution failure entirely, shifting all failures to confabulation. The two failure modes respond to the same trigger, taking different paths.
Production measurement across 740 drug-disease pairs finds 7.8% overall DG in a deployed RAG system, rising to 13.6% for recently approved drugs. Entity-attribution verification (checking that cited evidence applies to the queried entity) detects DG at 97.0% precision and 98.7% DG recall (IPW-adjusted human gold standard); no existing framework implements it.
arXiv:2607.09518v1 Announce Type: new
Abstract: A central challenge for digital health interventions aimed at improving habitual behaviors is deciding when to deliver an intervention prompt. For many daily habits, such as tooth brushing or eating, individuals tend to act around a usual time of day, but this timing is not fixed and can shift as routines evolve. When intervention timing is selected in advance and held constant throughout a study, it can gradually become misaligned with behavior, causing interventions to potentially arrive after the behavior has already occurred or too early to be effective. In this work, we address this habitual timing misalignment in digital health interventions by proposing an online decision-making framework that continuously adapts intervention timing as individual behavior patterns change. Rather than treating intervention timing as a static design choice, our framework adapts it over time and integrates it into a sequential process that determines both when and whether to deliver an intervention. Using data from a deployed oral health intervention trial as a case study, we evaluate our approach using both observed data and simulated settings to assess how well different intervention timing strategies align with the timing of brushing events. Across these evaluations, we measure performance using a coverage-based metric that captures whether an intervention is delivered sufficiently close to a subsequent brushing event. We find that adaptive intervention timing consistently improves coverage compared to fixed intervention times based on user-provided input. The proposed framework is currently deployed in an ongoing randomized controlled trial of a digital oral health intervention, with preliminary results that are consistent with and further support our prior evaluations.
arXiv:2607.09522v1 Announce Type: new
Abstract: Locally Checkable Proofs (LCPs) enable the verification of global graph properties using locally checkable certificates assigned by a prover. Recently, this framework was extended to Locally Checkable Proofs-with-Errors (LCPE), where an adversary may corrupt some certificates. Existing LCPE algorithms, however, are designed for the LOCAL model, whose unbounded communication makes them unsuitable for direct implementation in the bandwidth-restricted CONGEST model.
We initiate the study of efficient CONGEST implementations of LCPE through the \textsc{unique-leader} verification problem on trees. The main challenge is that tolerating $\varepsilon$ certificate errors requires each node to reason about its $(2\varepsilon+1)$-hop neighborhood, whose exact topology may require up to $O(\Delta^{2\varepsilon+1}\log n)$ bits to communicate. To overcome this bottleneck, we introduce \emph{local graph sketches}, together with the notions of \emph{imagined trees} and \emph{imagined certifications}, which encode precisely the information needed for verification using only $O(\varepsilon^2\log n)$ bits per node. Based on these sketches, we design an LCPE algorithm that tolerates up to $\varepsilon$ adversarial certificate errors and constructs the required sketches in $O(\varepsilon^2)$ communication rounds in the CONGEST model.
We complement our algorithm with a matching impossibility result: even in the strictly more powerful LOCAL model, and even with unbounded certificate size, no verification scheme with view distance at most $\varepsilon$ can tolerate $\varepsilon$ adversarial certificate errors. Since every CONGEST algorithm can be simulated in LOCAL, this lower bound immediately applies to CONGEST, showing that a view distance exceeding $\varepsilon$ is unavoidable.
arXiv:2510.04487v5 Announce Type: replace
Abstract: While accuracy is a critical requirement for time series forecasting, an equally important desideratum is reasonable forecast volatility across forecast creation dates (FCDs). Even highly accurate models can produce erratic revisions between FCDs, undermining trust and disrupting downstream decision-making. To improve the volatility of forecast revisions, state-of-the-art models like MQCNN, MQT, and SPADE employ a powerful yet underexplored neural network architectural design: forking-sequences. This architectural design jointly encodes and decodes the entire time series across all FCDs, producing an entire multi-horizon forecast grid in a single forward pass. This approach contrasts with conventional neural forecasting methods that process FCDs independently, generating only a single multi-horizon forecast per forward pass. In this work, we formalize the forking-sequences design and motivate its broader adoption by introducing a metric for quantifying excess volatility in forecast revisions and by providing theoretical and empirical analysis. We theoretically motivate three key benefits of forking-sequences: (i) reduced forecast volatility through ensembling; (ii) gradient variance reduction, improving the statistical efficiency of the training procedure; and (iii) improved inference computational efficiency. We validate the benefits of forking-sequences compared to baseline window-sampling on the M-series benchmark, using 16 datasets from the M1, M3, M4, and Tourism competitions. We observe median sCRPS improvements across datasets of 46.2%, 49.3%, 28.6%, 24.7%, and 6.4% for RNN, LSTM, CNN, Transformer, and State Space-based architectures, respectively. We then show that forecast ensembling during inference can reduce median forecast volatility by 13.2%, 13.0%, 10.9%, 10.2%, and 11.2% for these respective models trained with forking-sequences, while maintaining accuracy.
arXiv:2605.02574v3 Announce Type: replace-cross
Abstract: Gaussian Process (GP) models provide a flexible framework for prediction and uncertainty quantification. For most covariance functions, however, exact GP prediction with $n$ points scales as $\mathcal{O}(n^3)$, making it prohibitively expensive for large datasets or large numbers of prediction points. While nearest neighbor-based prediction can work well in certain settings, non-pathological circumstances (like measurement noise, for example) can severely restrict its efficiency. This work presents a complementary approach where one conditions on carefully designed linear combinations of data, which is particularly effective in the setting of jointly predicting many values in large connected regions of the data domain. For kernel functions that are smooth away from the origin and simple prediction domains, this method can be exponentially convergent in the number of linear combinations $r$ used for conditioning. The procedure costs $\mathcal{O}(T r^2)$ work, where $T$ is the cost of solving a linear system with the data covariance matrix, and so in many cases can be computed in linear or near-linear cost by exploiting rank structure in well-behaved covariance matrices. At the cost of $\mathcal{O}(nr^2)$ additional precomputation work, this approach can also provide predictions at arbitrary points of a designated region in $\mathcal{O}(1)$ online work, making it particularly attractive for problems where prediction points are not known in advance. After establishing favorable theoretical properties, we provide several example applications to problems in prediction and matrix approximation.
arXiv:2603.21883v2 Announce Type: replace
Abstract: Real-space singularities govern a broad spectrum of wave phenomena, yet they remain largely unexplored in elastic wave systems. Here, we report hybrid topological singularities that emerge on the surfaces of finite-sized solids due to the full vectorial character of elastic waves. These textures fuse spin-field singularities with displacement-field singularities and exhibit unique non-pairwise topological charge dynamics. Moreover, a subset of these singularities imprint dislocation lines onto the radiated acoustic field, generating robust acoustic vortices in free space from an otherwise achiral source and structure. Our results establish elastic waves as a powerful platform for engineering real-space singularities and open avenues for singular phononics and the exploration of rich topological defects in elastic media.
arXiv:2603.13334v5 Announce Type: replace
Abstract: Lipschitz-based robustness certification bounds a network's sensitivity through concrete numerical computation rather than symbolic reasoning, and so scales efficiently. It is increasingly used even where verifiable guarantees matter. Yet, as with most prior work on robustness certification and verification, soundness is typically proved against a semantic model assuming exact real arithmetic. Deployed networks instead execute in floating-point, creating a gap between certified properties and executed behaviour.
As motivating evidence, we give counterexamples showing that real arithmetic robustness guarantees can fail under floating-point execution, even for previously verified certifiers. We then develop a formal, compositional theory relating real arithmetic Lipschitz-based sensitivity bounds to floating-point execution under standard rounding-error models for feed-forward ReLU networks. We derive sound conditions for floating-point robustness, including bounds on certificate degradation and sufficient conditions for the absence of overflow. We also give an efficient floating-point Gram iteration algorithm for Lipschitz bounds and prove that it never under-estimates the true norm. Separately, when a model is certified pre-deployment, we show how measuring its actual deviation against a high-precision execution can substantially reduce certificate degradation.
We formalise the theory and its soundness, and implement an executable certifier, evaluated across dense networks spanning image, tabular, and many-class classification. To our knowledge, ours is the first method for soundly accounting for floating-point effects in Lipschitz-based robustness certification, and, done efficiently, the first floating-point-sound robustness checking procedure of any kind to certify models' entire test set -- seven those with 500,000 examples -- while retaining enough precision to be practical.
arXiv:2605.28998v2 Announce Type: replace-cross
Abstract: Improving the image contrast of objects immersed in weakly scattering media can be achieved using various strategies. One common approach is to reject events associated with scattered photons in favor of the detection of ballistic photons. While this is traditionally done via time gating or spatial filtering, we propose a different approach based on probing the object with spatio-temporally entangled photon pairs. We show that coincidence detection, followed by post-selection on spatially correlated events, allows us to isolate ballistic from scattered bi-photons, thereby enhancing image contrast relative to a single-photon detection strategy, and simultaneously removes events due to background light. Our predictions are obtained via numerical simulations and confirmed by experiments conducted in two configurations where either both photons or only one illuminates the scene. In both scenarios, correlation post-selection shows an improvement in image contrast at the expense of higher shot noise due to the lower number of events. The latter can be partially compensated for by appropriately combining events from several post-selection windows. Our findings will enable extending imaging through scattering media into the quantum imaging framework in settings where adaptive optics, time gating, and spatial filtering are impractical.
arXiv:2604.14530v2 Announce Type: replace
Abstract: We study the problem of constructing concurrent objects in a setting where $P$ processes run in parallel and interact through a shared memory that is subject to write contention. Our goal is to transform hardware primitives that are subject to write contention into ones that handle contention gracefully.
We give contention-resolution algorithms for several basic primitives, and analyze them under a relaxed, roughly-synchronous stochastic scheduler, where processes run at roughly the same rate up to a constant factor with high probability. Specifically, we construct read/write registers and CAS registers that have latency $O(\log P)$ w.h.p. under our scheduler model, using $O(1)$ hardware read/write registers and, in the case of our CAS construction, one hardware CAS register. Our algorithms guarantee performance even when their operations are invoked by an adaptive adversary that is able to see the entire history of operations so far, including their timing and return values. This allows them to be used as building blocks inside larger programs; using this compositionality property, we obtain several other constructions (LL/SC, fetch-and-increment, bounded max registers, and counters).
To complement our constructions, we give a trade-off showing that even under a perfectly synchronous schedule and even if each process only executes one operation, any algorithm that implements any of the primitives that we consider, uses space $M$, and has latency at most $L$ with high probability must have expected latency at least $\Omega(\log_{ML} P)$.
arXiv:2607.03198v2 Announce Type: replace
Abstract: World models -- compressed latent representations of an environment that support action-conditioned prediction and planning -- are typically presented as a product of modern self-supervised learning. This paper argues that the functional anatomy of a world model was independently developed, deployed, and formally analyzed decades earlier in the model-order-reduction (MOR) and control literature, under different names and for a different purpose: the real-time operation of physical systems. We trace the anatomy across three communities. Low-dimensional models of turbulence built on proper orthogonal decomposition (POD) supplied latent dynamics learned from data of a chaotic environment; eigenface methods in early computer vision supplied the encoder-decoder half, including a primitive runtime validity check; and measurement-based POD frameworks for facility thermal control assembled the complete loop -- POD coefficients as latent state, parametric dependence on actuator setpoints as action conditioning, modal reconstruction as decoding, and, critically, a priori analytical error bounds as a verification layer that certified when the model's predictions could be trusted in closed loop. We then examine what each tradition possesses that the other lacks: MOR contributes verification, physical grounding, and extreme data efficiency; learned world models contribute nonlinear representation, transferability, and horizon. We argue that the outstanding obstacle to deploying world models in systems that cannot fail -- power, thermal, process control -- is not predictive fidelity but verifiability, and we outline a research agenda for physics-grounded, verifiable world models that unifies the two lineages.
arXiv:2605.15516v2 Announce Type: replace
Abstract: Liquid-cooled exascale supercomputers dissipate heat through cooling plants organized as multiple parallel subloops, but how to allocate coolant distribution units (CDUs) across subloops and how to distribute flow among them has not been systematically addressed for facilities at this scale. This paper presents a three-layer optimization framework that jointly determines the integer partition of CDUs across subloops, the continuous flow fraction allocation, and the per-timestep co-design optimization of total flow rate and supply temperature subject to per-subloop thermal safety constraints. The Modelica simulation model is built based on the data of Frontier exascale supercomputer at Oak Ridge National Laboratory. By developing a reduced-order surrogate model, all 611 feasible partitions of 25 CDUs are evaluated across the full year operational dataset of 49,353 timesteps. Three progressively richer operational strategies are compared, ranging from flow control optimization to full three-layer co-design optimization with dynamically adjusted flow fractions. The optimal design within the surrogate optimization problem is a two-subloop plant achieving 35.48% annual cooling energy savings, only 0.18% above the current three-subloop Frontier design at 35.30%. Most of the savings are delivered by supervisory co-optimization of total flow rate and supply temperature; the distinct role of flow fraction optimization is design robustness rather than additional raw savings. Flow fraction optimization compensates for any feasible CDU-to-subloop assignment, reducing the design sensitivity by 93% and providing a low-cost software-only pathway to near-optimal performance on the existing Frontier hardware. The framework is transferable to other liquid-cooled high-performance computing plants.framework is transferable to other liquid-cooled high-performance computing plants.
arXiv:2607.09248v1 Announce Type: new
Abstract: We study non-clairvoyant scheduling for batched Large Language Model (LLM) inference under a hard Key-Value (KV) cache memory budget. Each request has a known prompt length but an unknown response length, and its memory footprint comprises a fixed prompt component together with a response component that grows with each decoded token. At each decoding round, the scheduler chooses a feasible batch of active requests; evicting a request discards its accumulated cache states, wasting prior computation. The goal is to minimize total completion time against the optimal clairvoyant schedule that knows all response lengths.
We present the first constant-competitive algorithm for arbitrary prompt lengths and arbitrary response lengths with no additional assumptions. Rather than relying on a single universal scheduling policy, our algorithm is built on a novel regime-aware routing framework. Specialized sub-schedulers handle different memory-growth geometries, while a meta-scheduler time-shares the memory budget across them and dynamically routes each job as its execution progressively reveals its behavior. This framework also yields constant-competitive guarantees for makespan and for total completion time under online arrivals.
arXiv:2605.28215v2 Announce Type: replace
Abstract: In-context learning (ICL) enables multimodal large language models (MLLMs) to classify images from a few labelled examples. Yet, how these models use the provided context remains opaque. While Chain-of-Thought prompting is widely used, recent work argues that it may not reflect true internal computation.
In this paper, we systematically evaluate the concept-based explainability of frozen MLLMs under few-shot ICL using five conditions of increasing formal rigour, ranging from baseline classification to Description Logics (DL) axiom generation.
Evaluating four state-of-the-art MLLMs via an independent LLM-as-a-judge pipeline, we demonstrate that explaining is genuinely harder than predicting alone. Surprisingly, forcing models to generate formally structured, concept-based explanations degrades predictive accuracy monotonically (from 93.8% to 90.1%), contradicting the assumption that explicit reasoning universally aids performance.
However, when models successfully articulate class-discriminative visual features, explanation quality strongly correlates with correct predictions. Our findings suggest that while MLLMs excel at visual classification, they lack the specific instruction-tuning required for formal, machine-verifiable explainability.
arXiv:1805.10330v4 Announce Type: replace
Abstract: Consider a network that evolves according to a reversible, nearest neighbours dynamics. Is the dynamics allowed to vary the size of the network? On the one hand it seems that, being the principal carriers of information, nodes cannot be destroyed without jeopardising bijectivity. On the other hand, there are plenty of bijective functions from the set of graphs to the set of graphs that are non-vertex-preserving. The question has been settled negatively -- for three different reasons. Yet, in this paper we do obtain reversible local node creation/destruction -- in three relaxed settings, whose equivalence we prove for robustness. We motivate our work both by theoretical computer science considerations (reversible computing, cellular automata extensions) and theoretical physics concerns (basic formalisms towards discrete quantum gravity).
arXiv:2606.03631v3 Announce Type: replace
Abstract: Multivariate time series classification (MTSC) is pivotal in high-stakes domains, such as clinical diagnosis and industrial fault detection, where safe deployment necessitates transparent decision-making. However, isolating the temporal segments that drive model predictions is challenging because discriminative signals in real-world time series are typically sparse, heterogeneous, and heavily obscured by background noise. This paper, therefore, proposes AnchorMoE, an interpretable-by-construction classification framework. Built upon a Mixture-of-Experts (MoE) architecture, AnchorMoE encodes multi-view representations of local patches and routes them to specialized experts, ensuring that the final prediction is formulated as an exact additive decomposition over the input segments, facilitating ante-hoc transparency rather than relying on post-hoc estimations. To maintain the reliability of this decomposition under sparse signal distributions, we introduce a geometric orthogonality constraint that penalizes representational redundancy, compelling distinct experts to specialize in heterogeneous predictive patterns. Furthermore, an uncertainty-aware reliability gate is designed to dynamically calibrate the contribution of each segment, effectively suppressing residual background noise. Extensive experiments on real-world and synthetic benchmarks demonstrate that AnchorMoE achieves highly competitive classification performance while faithfully grounding its decisions in the raw time series.