Optimization and Control
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Wednesday 2 September
A basic question in adaptive control is how rapidly a discrete-time nonlinear plant with unknown parameters may grow while remaining stabilizable. When the nonlinear growth exponent is known and only a scalar coefficient is unknown, existing theory identifies 4 as the exact critical exponent for stabilizability. We determine how this critical exponent changes when the growth exponent is also unknown. For any positive disturbance bound, one feedback law stabilizes all plants in some neighborhood of a nominal parameter pair if and only if the nominal exponent is below $3\sqrt{3}/2$. At equality, every compact parameter set whose exponents do not exceed this value remains stabilizable. If the exponent belongs to a known finite set, the critical exponent remains 4 for every nondegenerate compact coefficient interval. Hence finite exponent sets and arbitrarily short exponent intervals can have different critical exponents. Uncertainty in the growth rate therefore changes the range of growth that feedback can stabilize, not merely the size of the uncertainty set. The critical exponent remains $3\sqrt{3}/2$ under a known positive state-dependent multiplier bounded above and away from zero. For a system whose coefficient and exponent are known functions of an unknown parameter, a compact parameter family is stabilizable when its exponents do not exceed this value. At an interior parameter point where the Jacobian of these functions has rank two and the exponent is at least this value, no compact neighborhood is stabilizable.
Neurons and Cognition
Prediction emerges in RNNs trained for perception
The brain is highly proficient at making sense of noisy and ambiguous sensory inputs. Predictive processing hypothesises that this ability relies on prediction. However, it is unclear why the brain would have evolved to predict the sensory world, a computationally expensive process, in order to aid perception. Here we use simulations to argue that prediction naturally emerges in systems optimised for perception. We train recurrent neural networks (RNNs) to denoise a tokenised version of Bach's compositions at a range of noise levels. Afterwards, we enquire whether the states of the networks contain predictive information about the next token. We test this by freezing the RNN weights and training a linear readout from its states on prediction. We compare the performance of the linear readout with that of an independently trained linear benchmark model. The results show that the linear readout from the RNNs outperforms the benchmark model at moderate levels of noise, indicating that the networks rely on a predictive mechanism to support perception. We further show that the responses of the RNNs to sensory inputs are proportional to prediction error. Together, the results demonstrate that neural signatures of predictive processing emerge, without any further training constraints, from optimisation of perception.
Read the full abstract
Quantum Physics
Spin squeezing by geometric focusing in vacuum Rabi oscillations
We show that vacuum Rabi oscillations can directly generate spin squeezing through geometric focusing on the Bloch sphere. Starting from a coherent spin state resonantly coupled to a cavity initially in the vacuum state, quantum fluctuations are focused by the curvature of the Bloch sphere as the collective spin approaches the atomic ground state, producing squeezing transverse to the direction of motion. The squeezing timescale is set by the collective Rabi frequency $t_s\sim 1/(g\sqrt{N})$. The optimal Wineland squeezing parameter scales as $\xi_{\rm opt}^2\propto N^{-1/3}$, which is an outcome of the competition between the geometric focusing effects and the cavity-field vacuum fluctuations. The squeezing remains robust against realistic dissipation. In the end, an application example of $^{171}$Yb is briefly discussed to show the feasibility of our protocol.
Read the full abstract
Computation and Language
Language Models Can Control Their Own Attention
Language models spend most of their attention on a small fraction of context, yet they read the entire KV cache to find the few tokens that matter. If the user asks about a previous detail in a 1M-token conversation, global attention layers must scan the full context to generate each token of the reply. A prominent approach pre-selects relevant tokens via lightweight proxy scores, but this extrinsic scoring still incurs O(N) per step. We take an intrinsic approach motivated by the simple question: wouldn't the model already know which parts of the context are relevant? We introduce Declarative Attention (DA), a protocol that elicits the model to declare where it needs to attend within its chain-of-thought, partitioning generation into three modes: <global> (full context), <focus> (a specific region), and <local> (recent output only). The inference engine parses these declarations like tool calls and skips most of the KV cache read. Under zero-shot evaluation across 15 long-context tasks, DA on off-the-shelf models (Gemma-4-31B, Qwen-3.6-27B) reduces total attended tokens per response (45.8%, 19.6%), despite extra decoding steps, with modest accuracy drops (1.40pp, 2.10pp) that shrink with model scale. DA unlocks a new axis of sparse attention, with further potential under training-based methods. Code is available at https://github.com/itsnamgyu/declarative-attention.
Read the full abstract
Mesoscale and Nanoscale Physics
QArray+: A physics-informed GPU-accelerated simulator for quantum dot arrays
Semiconductor quantum-dot arrays are a compelling platform for scalable quantum technologies, yet their practical operation is hindered by the complexity of tuning large-scale devices. Existing automation tools rely on simplified physical models—such as constant-capacitance approximations and equilibrium Hubbard models—which assume instantaneous relaxation to a steady state. These frameworks fail in experimentally critical regimes where measurement rates exceed tunneling dynamics, necessitating more sophisticated non-equilibrium control strategies. To bridge this gap, we introduce QArray+, an extension of the QArray framework that incorporates gate-dependent tunnel coupling and a quantum open-system description of dissipative processes. This approach enables the unified simulation of coherent interdot charge-state hybridization and the non-equilibrium latching dynamics essential for training robust machine-learning models for automated device operation. Implemented in JAX with GPU acceleration, QArray+ scales across GPUs and multi-node systems. For example, a charge stability diagram for a 100X100 grid of gate voltages over 64 dots can be computed in $\sim0.17\,\mathrm{s}$ on multiple GPUs. Since interdot interactions are short-ranged and the corresponding tuning corrections are local, simulations at these scales capture the physics relevant to even larger devices. These capabilities support high-throughput dataset generation for automated device tuning.
Read the full abstract
Computation and Language
Choosing a PEFT Variant for Per-Patient Dysarthric ASR: A Single-Speaker Case Study on Two ASR Bases
Per-patient adapters are the preferred production architecture for dysarthric automatic speech recognition (ASR), yet parameter-efficient fine-tuning (PEFT) variants have not been compared in the speaker-dependent, per-patient regime. We present a single-speaker case study comparing seven LoRA-family methods (LoRA, QLoRA, AdaLoRA, DoRA, LoHA, VeRA, VB-LoRA) on two production bases (Whisper-large-v3 with Hungarian fine-tuning, and a multilingual Qwen3-ASR-1.7B checkpoint) for one post-stroke Hungarian male speaker (S1, 409 utterances; severe dysarthria on auditory-perceptual clinical assessment). Attention-projection adapters substantially improve CER on both bases. Across three seeds, a paired bootstrap detects no significant LoRA-DoRA difference (p>0.5; 13.86/13.90 % CER on Whisper, 28.10/28.33 % on Qwen3-ASR), so we adopt the simpler, cheaper LoRA. Real 4-bit (NF4) QLoRA is worse on every seed and both bases (14.56/30.09 % CER) with no memory saving at this scale, and LoHA, VeRA, VB-LoRA and AdaLoRA do not reach the LoRA family, though LoHA still gives an 18.6 % relative CER reduction on Whisper. On the same base, full fine-tuning is more accurate (11.43 % CER), but a 115 MB LoRA that also adapts the feed-forward blocks reaches within 0.66 pp of it at approximately 3.7 % of the per-patient storage. A 6-point enrollment grid shows about 5 min of patient audio captures 45.6 % of the zero-shot-to-30-min CER reduction, with further gains at 10 and 30 min (caveat: one speaker, one language, severe post-stroke dysarthria). Training scripts and recipes will be released, source-available under a research-use licence, on publication.
Read the full abstract
Machine Learning
LoRA-TSD: Tangent-Space Spectral Descent for LoRA via Muon-Style Updates
Low-rank adaptation (LoRA) is the standard way to fine-tune large models, yet when its two factors are trained independently, the update ignores the geometry of the low-rank weight change it induces. We introduce LoRA-TSD, an optimizer that treats every LoRA step as a tangent vector of the fixed-rank matrix manifold and takes the spectral-norm steepest-descent step of Muon inside that tangent space, mapping the result back to the factors through a retraction native to the LoRA parametrization. The step avoids expensive operations on full weight matrices, and its retraction is up to $2.8\times$ cheaper than the truncated-SVD retraction used by prior manifold methods. We prove that the Frobenius-norm version of our surrogate recovers LoRA-Pro, and we identify the tangent-projected gradient, the Riemannian gradient of the manifold, as the stationarity measure natural to LoRA training and computable from the factor gradients alone. Under this measure we give the first global convergence guarantees for both LoRA-Pro and LoRA-TSD, with rates that drive the factor-gradient norms to zero. Across six commonsense and natural-language-inference benchmarks with Llama-3.2-1B, Llama-3.1-8B and Qwen3-32B, LoRA-TSD outperforms every competing LoRA optimizer and stays robust to the adapter rank. Code is available at https://github.com/brain-lab-research/LoRA-TSD.
Read the full abstract
Optics
Optical Reservoir Computing with Structural Nonlinearity for Forecasting Chaotic Time Series
Optical computing offers a promising route to high-speed, energy-efficient information processing. However, implementing nonlinear operations typically requires strong light-matter interactions. Here, we introduce structural nonlinearity into an optical reservoir computing architecture, using only linear components. A digital micromirror device acts as a reconfigurable scattering potential, while the propagation of light through a scattering medium provides the high-dimensional mode mixing. By enforcing a second interaction with the wavefront shaper, we obtain a quadratic mapping of the input within the reservoir. A single experimental apparatus simultaneously implements conventional and structurally nonlinear reservoir architectures, enabling a direct comparison on forecasting of the chaotic Mackey-Glass time series. We demonstrate that the structurally nonlinear reservoir improves short-term prediction accuracy and better preserves the long-term dynamics of the chaotic system, demonstrating the potential of structural nonlinearity for low-power optical computing.
Read the full abstract
Analysis of PDEs
Conservation of mass for solutions of Leibenson's equation on Riemannian Manifolds
We consider on a Riemannian manifold $M$ the Leibenson equation \begin{equation*}\label{eqabs}\partial _{t}u=\Delta _{p}u^{q},\end{equation*} where $p>1$ and $q>0$. When $q(p-1)\geq 1$, we prove conservation of mass for solutions of Leibenson's equation assuming only the volume bound $V(x_0, r)\leq \exp\left(C r^{\frac{p}{p-1}}\right)$ for some $x_{0}\in M$ and all large enough $r>0$. When $q(p-1)< 1$, we prove this property assuming $V(x_0, r)\leq Cr^{N}$ and $p>N[1-q(p-1)]$, which matches the threshold in $\mathbb{R}^{n}$ with $N=n$. We also show that solutions on the hyperbolic space $\mathbb{H}^{n}$ have a finite extinction time in the case $q(p-1)< 1$, which implies that the conservation of mass property does not hold. Using the conservation of mass result in the case $q(p-1)=1$, we also prove a $L^{p-1}$- Liouville property, which partially answers a conjecture stated by I. Holopainen \cite{holopainen2000sharp}.
Read the full abstract
Computer Vision
RVSD: Retrieval Vision Sparse Decoding for Mitigating Visual Hallucinations in Large Vision-Language Models
Large vision-language models have achieved remarkable success in vision-language tasks. However, they remain prone to Visual Hallucinations (VHs), undermining their reliability in real-world applications. Existing solutions typically require curated datasets, additional training, or multi-round decoding, resulting in considerable computational overhead. In this paper, we propose RVSD (\underline{R}etrieval \underline{V}ision \underline{S}parse \underline{D}ecoding), a training-free and plug-and-play decoding framework that, for the first time, unifies token sparsification and Semantic-Space Visual Retrieval (SSVR) within a single decoding pass. Within RVSD, we introduce a semantics-directed token selection strategy that selectively sparsifies redundant tokens while preserving critical visual information. We further propose the SSVR mechanism, which reformulates visual compensation as an on-demand cross-modal retrieval process within a shared semantic space. Extensive experiments demonstrate that RVSD achieves state-of-the-art performance in mitigating VHs while maintaining robust suppression capabilities under long-context generation settings. Our code is available here.\footnote{https://github.com/canjie-liu/RVSD}
Read the full abstract
Computation and Language
CORAL: An LLM-Native Harness for Production Recommender Systems
Production recommender systems shape what billions of people see, and sustaining their performance requires continual optimization: as content, user behavior, and upstream models shift, the choices governing retrieval, ranking, and serving must be revisited. Traditionally, human engineers test such changes through online experiments–a slow, reactive process limited by engineering effort, leaving parts of the system unrevised as conditions change. Although large language models have been applied to ranking, user modeling, and offline model development, few systems place an agent in a continual closed loop that acts on a live recommender and learns from the measured effects of its decisions. We present CORAL (Constraint-Optimized Recommender via an Agentic Loop), an LLM-native harness that closes this loop: each cycle, the agent observes operating signals, reasons over a memory of past decisions and outcomes, and invokes tools–including a numerical optimizer that keeps changes within a fixed operating budget–to reconfigure the recommender, with measured outcomes informing the next cycle. We formulate this as a partially observed, non-stationary, constrained optimization problem in which the policy improves in context, without parameter updates, from its prior actions. Across two large-scale social platforms, evaluated with A/B experiments, the same harness improves engagement at no additional serving cost on one and reduces serving cost without degrading engagement on the other, spanning the engagement-efficiency frontier. Performance improves as the loop iterates, suggesting that a single agentic loop can automate continual optimization work traditionally performed by human algorithm engineers under explicit guardrails.
Read the full abstract
Human-Computer Interaction
BuildOcc: A Large Language Model Occupant Agent Platform for Building Energy Research
Occupants are a primary source of uncertainty in building energy consumption and management, yet existing occupant behavior models cannot capture adaptive and reasoning responses considering the occupant's personal history, current context, and the type of energy signal being delivered. This study presents BuildOcc, an open-source Python platform that grounds large language model agents in the American Time Use Survey (ATUS), a nationally representative diary dataset covering 16,684 respondents. Through BuildOcc, each simulated occupant agent can be instantiated with a demographic persona drawn from ATUS population statistics, a memory stream that accumulates and reflects on timestep-level observations, and an activity scheduler that samples empirically from ATUS time-at-activity distributions. The platform exposes a three-layer interface - Python library, REST API, and Model Context Protocol server - so that any building energy tool (EnergyPlus, Home Assistant) can integrate behavioral intelligence without bespoke coupling code. A plugin registry lets the community add new occupant strata, custom schedulers, and alternative memory backends as separate installable packages. Two validation tiers show that ATUS-grounded sampling reproduces empirically calibrated activity distributions and that demographic priors propagate into persona-consistent agent reasoning across timesteps, establishing internal consistency across strata. BuildOcc provides the building energy community with a reusable, openly available implementation of the occupant behavioral layer. BuildOcc is openly released at https://doi.org/10.5281/zenodo.21192895 under the Apache License 2.0 and installable via pip install buildocc.
Read the full abstract
Machine Learning (Statistics)
Momentum in large-batch training: Polyak enlarges the critical batch size, Nesterov improves data efficiency
We study when and how momentum improves large-batch training in the one-pass regime, using power-law kernel regression as a tractable setting. We first characterize stochastic stability through the critical learning rate, defined as the largest learning rate for stable training, and obtain $\eta_{\mathrm{SGD}}^{\mathrm{crit}}\eqsim 1$, $\eta_{\mathrm{Polyak}}^{\mathrm{crit}}\eqsim \min\{1,B(1-\beta)\}$, and $\eta_{\mathrm{Nesterov}}^{\mathrm{crit}}\eqsim \min\{1,B^q(1-\beta)\}$, where $B$ is the batch size, $\beta\in[0,1)$ is the momentum factor, and \(q>1\) is the capacity exponent governing spectral decay. Within the stable region, we derive scaling laws for the full risk dynamics, capturing the progression from an early transient through power-law decay to a noise floor. We then optimize the final-step risk under a fixed data budget, obtaining a three-regime phase diagram that reveals how the benefits of momentum change with batch size. Notably, Polyak enlarges the critical batch size, the largest batch size preserving the optimal small-batch data-scaling exponent, thereby allowing fewer sequential updates without sacrificing data efficiency. In contrast, Nesterov improves the optimal data-scaling exponent in the large-batch regime through look-ahead-induced suppression of noise amplification. Experiments on PLK regression quantitatively support the theoretical scaling predictions, while neural-network experiments provide qualitative evidence for the predicted batch-size dependence and stability behavior.
Read the full abstract
Numerical Analysis
Neural operators approximate strongly continuous convex monotone semigroups
We approximate strongly continuous convex monotone semigroups by learning their Chernoff-type one-step operators with neural operators. First, we introduce the general class of so-called Chernoff-neural operators and show in a universal approximation theorem that they can approximate the Chernoff one-step operators arbitrarily well. By using stability estimates between weighted Hölder spaces, the one-step approximation error can be propagated through the iterations which yields universal approximation of the corresponding semigroup. Second, we introduce the more specialized class of envelope-neural operators for envelope semigroups which allows us to derive quantitative approximation rates. Finally, we illustrate the effectiveness of these neural operators in several numerical examples arising from non-linear partial differential equations, stochastic optimal control and stochastic processes under model uncertainty.
Read the full abstract
Earth and Planetary Astrophysics
Revisiting orbital recurrence in the dimmings of Boyajian's star (KIC~8462852)
KIC~8462852 (Boyajian's star) shows deep, asymmetric, and highly irregular dimming events with no simple repetition pattern. A symmetric 1.1% transit detected by TESS in 2019 has been interpreted as a possible giant planet or brown dwarf, while the irregular dips are generally attributed to transiting dust from exocomets or planetesimal fragments. We test whether the broader dip families may nevertheless recur orbitally. Our scenario assumes a major disruption that produced fragments with slightly different periods, causing successive returns to spread progressively in orbital phase. We compare this recurrent model with a null hypothesis of chronologically independent family centers, accounting for the observing window, and marginalize over the recurrence period $P$ and dispersion scale $\sigma_P$. We obtain $B_{10}=44.9$ in favor of recurrence, corresponding to “very strong” evidence on the Jeffreys scale. The maximum-likelihood solution is $P_{\rm ML}=781.6$~d and $\sigma_{P,\rm ML}=80.0$~d, close to the 776.1-d timescale obtained if D800 and the 2019 TESS event are separated by four revolutions. Independently of the timing fit, this period predicts a transit duration of about 20.5~h for the published stellar and transit parameters, in close agreement with the observed $\sim21$~h. The result is conditional on the adopted topology, priors, family definition, and observing selection function, and does not prove orbital recurrence. Nevertheless, the family timings strongly favor this specific recurrent-fragment model over independent occurrence, making progressive orbital phase dispersion a viable and physically motivated interpretation.
Read the full abstract
Quantum Physics
Effective quasiparticle conserving Lindbladians in the thermodynamic limit
Open quantum many-body systems are commonly described by Lindblad master equations, yet the treatment of Lindbladian operators in the thermodynamic limit remains a major challenge. We develop a framework for constructing effective quasiparticle-conserving Lindbladian operators directly in the thermodynamic limit. Our approach extends continuous similarity transformations to non-Hermitian open quantum systems and enables the systematic block diagonalization of Lindbladians with respect to the quasiparticle (qp) number. We formulate two complementary methods. The first, projective continuous similarity transformations (pcst++), generalizes perturbative continuous unitary transformations to Lindblad operators so that a linked-cluster expansion allows us to obtain high-order series expansions of the infinite system. The second, deepCST, extends directly evaluated enhanced perturbative continuous unitary transformations by combining the same qp-conserving generator with a perturbative truncation scheme that yields non-perturbative effective Lindbladian operators directly in thermodynamic limit. We apply pcst++ and deepCST to the dissipative transverse-field Ising chain with local dissipation. We focus on the low-Ising regime. We purify the Lindbladian by splitting each spin into two sites. A spin flip then corresponds to two qps. We derive and analyze the effective, qp-conserving Lindbladian in the sectors with zero to two qps. We show how the Ising interaction renormalizes the decay rates of elementary spin-flip excitations and provides the microscopic mechanism for a competition between coherent interactions and dissipation. Our work establishes perturbative and non-perturbative continuous similarity transformations as a versatile tool for deriving effective qp pictures of open quantum many-body systems in the thermodynamic limit.
Read the full abstract
Game Theory
Almost Envy-Freeness for Additive Mixed Manna with Entitlements: Deterministic and Randomized Guarantees
We investigate the fair allocation of indivisible items among agents with asymmetric entitlements in mixed manna settings, where the items consist of both goods and chores. For additive valuations, we establish that weighted envy-free up to one item (WEF1) allocations always exist and can be computed in polynomial time. We also study fair and efficient allocation and show that weighted envy-freeness up to one transfer (WEF1T) is compatible with fractional Pareto optimality (fPO) for every mixed-manna instance. This relaxation from WEF1 to WEF1T is tight, as demonstrated by our impossibility result. We further show a best-of-both-worlds result via a finite lottery that guarantees weighted envy-freeness (WEF) in expectation, with every realized allocation satisfying WEF1T and achieving the tight characterization complemented by the existing impossibility result.
Read the full abstract
Mesoscale and Nanoscale Physics
Twist as a Mechanical Switch for Reconfigurable Stacking in h-BN
Twistronics of layered materials has emerged as a highly active field due to its profound implications for quantum electronics and materials engineering. However, the controllable manipulation of interlayer stacking remains a significant experimental challenge. Here, the homogeneous contact between hexagonal boron nitride layers is shown to be reproducibly switched between two distinct stable stacking configurations via an externally applied torque. Combining experiments and computational modelling, we identify the stacking order of these stable states as the commensurate AA and AB modes. These two states are associated with different rotational torque maxima, exhibiting an asymmetry ratio of 0.7, and distinct dynamics as a function of the twist angle. Moreover, the peak torque values scale linearly with contact area, highlighting the dominant role of edge elasticity in the twisting process. Given that the AA and AB stacking modes correspond to different out-of-plane electric polarization states, our findings offer a pathway for reconfigurable nano- and micro-electromechanical devices.
Read the full abstract
Optimization and Control
An Adaptive Projected-Gradient Algorithm for Sample-Average Approximations of Stochastic Multi-Objective Optimization
We consider stochastic multi-objective optimization over a nonempty closed convex set, where every objective is an expectation and only sample-gradient information is available. We develop a line-search-free and function-value-free adaptive projected-gradient algorithm for the sample-average approximation (SAA) problem. Each iteration computes a feasible regularized multi-gradient step and updates the regularization parameter from the projected step length. A normal-cone-based certificate yields descent estimates and an explicit complexity bound for the Pareto-stationarity residual of the SAA problem. The consistency of SAA gradients then transfers vanishing SAA residuals to Pareto stationarity for the population problem, while an additional concentration argument gives a finite-sample residual bound on compact sets. Experiments on synthetic problems, classification, portfolio selection, multi-task learning, and robot control illustrate the practical performance of our algorithm.
Read the full abstract
Numerical Analysis
A stable rank-adaptive step-and-truncate finite volume method for Vlasov transport on domains with piecewise linear boundaries
We consider the numerical solution of the linear Vlasov transport equation on bounded spatial domains with inflow boundary conditions based on low-rank approximation. We combine a finite volume discretization with a rank-adaptive step-and-truncate scheme for the resulting matrix ODE. The spatial and velocity meshes may be unstructured, while suitable numerical fluxes retain a separated space-velocity representation. For homogeneous inflow, we show that the low-rank scheme inherits the $L_2$ stability and CFL restriction of the underlying full finite volume forward Euler method. In addition, the low-rank approximation error is bounded explicitly in terms of the truncation tolerances, avoiding the modeling error associated with tangent-space projections in dynamical low-rank approximation. Numerical experiments in 1d1v and 2d2v confirm the predicted error behavior. In 2d2v, the method handles an unstructured spatial mesh with nonzero inflow and a full tensor-product discretization of approximately $5.8\cdot10^{10}$ phase-space cells while the numerical rank is at most twelve.
Read the full abstract
Statistical Mechanics
Optimal-work feedback on particles with activity — gliding on active fluctuations using positional information
We study the minimum-work feedback control of particles subject to active fluctuations. Considering an active Ornstein-Uhlenbeck particle confined by a moving harmonic trap, we derive exact optimal protocols following an initial position measurement. Our results show that nonequilibrium correlations between position and active fluctuations allow work extraction from the activity based on positional information only, i.e., without directly measuring the active degree of freedom, which was the focus of earlier literature. We find that depending on the persistence time, activity can either facilitate or impede transport relative to passive systems. Surprisingly, unlike feedback schemes based on direct measurements of the active fluctuations, positional feedback remains energetically advantageous even in the limit of infinitely persistent activity. Our results provide design principles for information engines and optimal control strategies operating in active environments.
Read the full abstract
Human-Computer Interaction
Large Language Model-Driven Context-Aware Eco-Feedback Generation and Evaluation
The objective of this study is to demonstrate the potential of generating context-aware eco-feedback - eco-feedback that reflects a household's contextual characteristics alongside its energy use patterns - through a large language model-integrated framework. Previous studies have introduced personalized eco-feedback, mostly relying on household energy use patterns; however, they frequently did not reflect distinct household characteristics, including their persona or non-negotiable routines, leaving eco-feedback ineffective and sometimes superficial. To address these limitations, we introduce a contextual engineering framework that generates eco-feedback using a self-consistency with chain-of-thought prompt, leveraging household energy analysis data, utility rate structures, and household characteristic information. We conducted a rigorous empirical validation and a combinatorial evaluation analysis to assess this framework systematically. The former tested the framework's ability to generate accurate and contextually grounded eco-feedback for three households by comparing its output against reference interventions independently derived from the same household data. The latter examined the framework's adaptability across 400 scenarios spanning 50 households, two utility rate structures, and four behavioral personas. Our framework generated eco-feedback that aligned with reference interventions at a mean rate of 92.0% and grounded its recommendations in the provided household data with 95.7% citation accuracy. It also proved highly adaptive, shifting both the appliances targeted and the energy-saving strategies recommended in response to rate structure and household context. Ultimately, this study contributes to realizing the next level of context-aware interactions between occupants and buildings which paves the way for higher occupant living quality and sustainability.
Read the full abstract
Algebraic Geometry
Smooth Counterexamples to the Eisenbud–Schreyer–Weyman Ulrich Existence Problem
In their 2003 article in the Journal of the American Mathematical Society, Eisenbud, Schreyer and Weyman asked whether every embedded projective variety carries an Ulrich sheaf. In 2017 Beauville proposed a numerical route toward a surface with no Ulrich bundles: in Picard rank one, existence forces $H^2\ge K_S^2-8\chi(\mathcal O_S)$, suggesting a search near the Bogomolov–Miyaoka–Yau boundary. We show that the Picard-rank-one hypothesis is not needed for the obstruction: a rank-independent Bogomolov–Hodge argument gives the same inequality on every smooth polarized surface. Using additional Neron–Severi directions on Hirzebruch–Kummer resolutions, the exponent-$3$ Hesse surface admits a very ample class $H=4A-E$ with $H^2=7\cdot3^9<16\cdot3^9=K_Y^2-8\chi(\mathcal O_Y)$, hence a smooth counterexample to the Eisenbud–Schreyer–Weyman problem in its standard formulation. Consequently its Chow form has no ESW-type linear determinantal representation arising from an Ulrich sheaf on the embedded surface. Moreover, for every $n\ge3$ the Hesse pair $(Y_n,4A_n-E_n)$ is a counterexample, the surfaces are pairwise non-isomorphic, and $H_n^2/\sigma(Y_n)=7/(3n^2-11)\to0$, while $K_{Y_n}^2/\chi(\mathcal O_{Y_n})\to60/7\approx8.5714$. A general arrangement-theoretic Rees-algebra/Segre mechanism yields further infinite families.
Read the full abstract
Computer Vision
MV-dVRK: A Multi-Viewpoint Benchmark for Spatial Surgical Perception
Large-scale training and refined optimization techniques have greatly improved sparse multi-view 3D reconstruction. Despite their relevance to surgery, such methods have never before been rigorously evaluated on real endoscopic images. Current clinical telerobots deploy a single stereo camera inside the patient, making multi-viewpoint data extremely rare. This paper presents MV-dVRK, the first ex-vivo surgical dataset to combine multiple exposure-synchronized stereo viewpoints with accurate surface geometry and camera poses. The static subset of the benchmark provides dense SfM reference geometry, validated against an industrial 3D scanner, together with ground-truth camera poses and sparse-view test sets. We use MV-dVRK to systematically compare zero-shot monocular, stereo, multi-stereo, and multi-view 3D reconstruction methods as the number of viewpoints increases. With two endoscopes, multi-stereo reconstruction achieves the highest coverage. With a third viewpoint, optimization-based multi-view methods perform best, covering 67% of ground-truth surface points within a 1 mm tolerance and recovering highly accurate relative camera poses. By contrast, feed-forward foundation models cover only 43% of the ground-truth surface in the same setting. MV-dVRK also includes ten dynamic sequences spanning multiple surgical tasks, with increasing kinematic complexity and tissue deformation, providing a basis for future research in multi-viewpoint surgical perception. The project is available at: https://mv-dvrk.is.mpg.de.
Read the full abstract
Cryptography and Security
Card-Based Computation in the Virtual Player Simulation Model
Player simulation has recently emerged as a new direction in card-based cryptography, with protocols developed for simulating virtual players in physical card games such as Old Maid, UNO, and President. Unlike conventional card-based secure computation, player simulation imposes additional constraints: the cards represent a persistent game state, the remaining cards in a virtual player's hand must be preserved after each action, and it is desirable to represent each card in the game by a single physical card. In this paper, we study generic card-based computation in the virtual player simulation model. We focus on games whose cards admit a publicly known ranking and propose two fundamental protocols. First, we present the Play-Minimum protocol, which securely selects and plays the minimum-value card from a virtual player's hand when all cards in the deck have distinct values. By symmetry, the protocol can also be used to play the maximum-value card. Second, we present the Sorting protocol, which securely arranges a virtual player's hand in nondecreasing order and remains applicable when multiple cards have the same value. These protocols provide generic computational primitives independent of any particular card game and constitute a step toward understanding the computational capabilities of the virtual player simulation model.
Read the full abstract
Quantum Physics
Quantum amplitude estimation beyond power-of-two schedules
Non-adaptive quantum amplitude estimation (QAE) fixes its Grover depths in advance, so every circuit can run in parallel, but it has so far needed more queries than the best adaptive methods. We show that most of this gap comes from two conventional choices: subspace-based post-processing and power-of-two depth ladders. We replace the first by the exact maximum-likelihood estimate, one matrix multiplication per batch of estimates, and the second by a geometric ladder with ratio $r \approx 1.45$. The result is a fully parallel, deterministic-schedule estimator with total query complexity $2.8$-$3.1/\varepsilon$ at 95% confidence for target errors from $3.5\times 10^{-3}$ to $10^{-6}$. This matches the average-case complexity of chebAE, the best benchmarked adaptive method, within statistical uncertainty (with the lower point estimate at every scale tested), beats its maximum-observed complexity by $1.6\times$, and needs a maximum sequential depth of only $0.21/\varepsilon$ against chebAE's $2.9/\varepsilon$. Relative to csAE, the best non-adaptive benchmark, the constants improve by 30-35% at 95% and $1.5$-$1.7\times$ at 99% confidence. The optimal ratio has a simple origin. Doubling is the fastest depth growth at which the data can still tell neighboring candidate values apart, so power-of-two ladders sit at the edge of confusion and must buy reliability with extra shots; a slightly denser ladder checks every scale redundantly. An error-probability analysis reproduces the measured failure rates and locates the optimum. The likelihood formulation extends directly to noise-aware estimation, and uniformly scaling the capped ladder covers the depth-limited regime, realizing the optimal trade-off $M N_{\mathrm{tot}} \approx (0.4$-$0.6)/\varepsilon^2$ within $\sim 1.1\times$ of the schedule's Cramér-Rao limit.
Read the full abstract
Functional Analysis
Mathematical Koans and Cartan Convexity: $\Gamma$-Convex Hulls and Butterfly Realizations
In homage to Hofstadter, we call a compact, citable specification of definitions, theorem, proof mechanism, and examples a mathematical koan when a full paper can be reconstructed and verified from it, including in reader-specific forms generated by AI. Every AI-solvable problem is itself a koan, although reconstruction need not be cheap; the present article is an expansion of one. We introduce Cartan convexity for self-adjoint free functions: locally, such a function agrees with a locally bounded matrix-convex free function. If the variable set has cardinality $\tau$ and $\lambda=\max\{\tau,\aleph_0\}$, every bounded real free set has a universal direct sum on a Hilbert space of dimension at most $2^\lambda$; every point of the set is a reducing summand of an amplification of this sum. Applying the noncommutative Kraus–butterfly theorem at that universal point yields an extension to an open matrix-convex neighborhood of the noncommutative convex hull. For normal affine pencils, the extension domain also contains the bounded strong closure of the hull. We prove an analogous theorem for a graph embedding $\Gamma$. A local convex lift through $\Gamma$ extends to a neighborhood of $\Gamma^{-1}(\operatorname{co}_{\mathrm{nc}}\Gamma(K))$ and admits a butterfly realization in the $\Gamma$-coordinates. We distinguish this lift condition from intrinsic $\Gamma$-convexity. For $\Gamma(x,y)=(x,y,y^2)$, the intrinsically $\Gamma$-affine polynomial $xy+yx$ has no convex lift germ at the origin, whereas a single quadratic coordinate suffices to lift every uniformly real analytic germ.
Read the full abstract
Number Theory
On divergence related to Riemann–von Mangoldt's explicit formula of the prime-counting function
An explicit formula for the prime-counting function $\pi(x)$, usually attributed to Riemann and von Mangoldt, is prominently stated as the equation $\pi(x)=R(x)-\sum_\rho R(x^\rho)$, where the sum runs over all zeros $\rho$ of the Riemann $\zeta$-function, the non-trivial ones being ordered by increasing absolute value of their imaginary parts and counted with multiplicity. This particularly entails the claim that the partial sums over the non-trivial zeros, $\Sigma R_T(x):=\sum_{0<|\Im m(\rho)|\le T} R(x^\rho)$ converge as ${T\to\infty}$. Writing $\Theta:=\sup\{\Re e(\rho):\ \zeta(\rho)=0,\ 0<\Re e(\rho)<1\},$ for what has recently been called “Riemann's constant”, we prove that, for every fixed $x>1$ and every $\theta<\Theta$, the sums $\Sigma R_T(x)$ are not $O(T^\theta)$. As a consequence, $\limsup_{{T\to\infty}}|\Sigma R_T(x)|=\infty$ and $\sum_\rho R(x^\rho)$ diverges. We conclude the paper by showing that an adapted, but simpler strategy also gives the analogous result – and hence divergence – for the contribution of the trivial zeros to $\sum_\rho R(x^\rho)$. In both cases the exponent $\Theta$ is proved to be sharp.
Read the full abstract
Optimization and Control
Insensitizing Control Problems for Coupled Stochastic Parabolic Systems with State and Gradient Observations
We study insensitizing control problems for a class of coupled linear stochastic parabolic systems. We establish the existence of controls such that a sentinel functional, involving localized observations of the state variables and their spatial gradients, is insensitive to small perturbations of the null initial data. We first reformulate the insensitizing control problem as a null controllability problem for a coupled forward–backward stochastic parabolic system, in which the observation terms induce both zeroth- and second-order coupling terms. By duality, the analysis is reduced to an observability inequality for the corresponding adjoint system. The main analytical contribution is the derivation of new global Carleman estimates for coupled stochastic parabolic systems with zeroth- and second-order coupling terms, under suitable geometric assumptions on the control and observation regions. These estimates yield the required observability inequalities and, consequently, the existence of insensitizing controls. Furthermore, depending on the value of a weighting parameter $\beta\in[0,1]$, which determines the relative contributions of the two state components to the sentinel functional, we consider two cases. If $\beta\in\{0,1\}$, the sentinel functional depends on only one state component, and a single localized control acting in the drift of the first equation is sufficient. In contrast, if $\beta\in(0,1)$, both state components contribute to the sentinel functional, and two localized controls acting in the drift terms of the two equations are sufficient. Moreover, the control strategy in this paper involves two additional controls acting throughout the diffusion terms.
Read the full abstract
Probability
Differentiability of the Leading Lyapunov Exponent of a linear differential equation with random coefficients Application to the Calculation of the Selection Gradient in Random Environments
The study of evolution in temporally fluctuating environments often relies on the analysis of Lyapunov exponents, which quantify the exponential growth of populations. However, when model parameters depend on a stochastic process, calculating the selection gradient, a key tool for predicting the evolution of phenotypic traits, becomes a mathematical challenge. While the periodic case has been resolved, a general approach for random environments remains to be developed. This article proposes a rigorous method to: Establish the differentiability of the leading Lyapunov exponent with respect to a parameter, providing an explicit integral formula for its derivative. Develop a numerical algorithm to approximate the derivative by solving an extended differential equation. Apply these results to the analysis of mutant invasion in a resident population at equilibrium, identifying the selection gradient as the derivative of the top Lyapunov exponent.