Papers

The operation of residential energy hubs with multiple energy carriers (electricity, heat, mobility) poses a significant challenge due to the energy storage differences in time-constants, round-trip efficiencies and self-discharge rates. Usually, thermal storage exhibits flexibility in yearly planning optimizations or long-term scenarios. However, as optimization horizons shrink (1-48hs) so does their supplied value due to the lower round-trip efficiencies. To avoid this early depletion during operation this paper proposes a data-driven highway to steer the short-term daily control towards long-term optimality. The proposed methodology also presents how to optimally size the thermal storage and avoid yearly simulations and how all of this is related to nonlinearities in the daily operation. The presented framework links seasonal and daily optimizations through dynamic terminal sets and value functions. The seasonally-aware nonlinear economic model predictive controller achieves the most balanced performance, with the second best mean grid cost of all MPCs at -\texteuro 209. It also achieves better battery degradation control than its linear counterparts (between 26-34%) and the best thermal comfort of the nonlinear benchmarks. Nevertheless, the data-driven seasonal highway restrains the ability to control battery degradation and slightly increases computational time.

Backchannel prediction has been studied almost entirely in dyadic conversation. We introduce a multi-party benchmark based on the AMI corpus, comprising 682 masked-listener views from 171 meetings, 190 speakers, and 18,697 backchannel events, with a person-disjoint held-out split. A state-of-the-art dyadic model applied zero-shot to meeting audio performs at chance (AUROC 0.499); nevertheless, its frozen acoustic features remain informative: a linear probe reaches 0.704, and retraining the predictor raises performance to 0.751. Retraining reveals a second limitation. Listener conditioning improves prediction for listeners seen during training but not for unseen listeners, and the gap remains under capacity reduction, listener-adversarial training, per-listener adaptation, and oracle lexical conditioning. Adversarial training removes only part of the speaker-identity information, while stronger removal hurts prediction, suggesting that identity is entangled with cues that are useful for backchanneling. A within-model control helps explain this pattern: with the same features and data splits, turn-onset prediction transfers to unseen listeners, while backchannel prediction does not. Backchannel rates also vary about twice as much across individuals as turn-onset rates. Since backchannels occupy only about 1% of frames, frame-level F1 is strongly affected by the base rate. We therefore report AUROC alongside event-F1 on listener-active regions. We release the benchmark and evaluation tools at https://github.com/HafsatiMohammed/bc_multiparty_release.

We study an incompressible visco-morphoelastic system in three dimensions in which the Eulerian effective strain is transported by the Zaremba-Jaumann rate and coupled to a Navier-Stokes momentum balance through a Kelvin-Voigt stress. The strain remains symmetric and splits into a scalar transport-relaxation equation for its trace and a deviatoric equation containing the nonlinear stretching and Jaumann terms. We regularize both equations by diffusion of order $\varepsilon$, which admits a mechanical interpretation as weakly nonlocal remodeling with a characteristic length scale. For fixed $\varepsilon>0$ we construct Leray-Hopf type weak solutions by a truncated Galerkin scheme and remove the truncation a posteriori using a maximum principle for the scalar variable. As $\varepsilon\to0$, the scalar variable converges strongly in $L^p((0,T)\times\Omega)$ for every $1\le p<\infty$, by a DiPerna-Lions commutator argument on bounded Lipschitz domains, allowing the stretching term to be identified in the limit. In contrast, the Jaumann commutator is a product of two weakly convergent sequences, and its limit cannot be identified from the available estimates. We therefore obtain a defect \[\boldsymbol{\mathcal{D}}\in L^2(0,T;(H^2(\Omega;\mathbb{S}_0))^\ast)\] in the limiting deviatoric equation. We give two sufficient conditions for $\boldsymbol{\mathcal{D}}=0$. The energy-level cancellation used in related corotational viscoelastic models does not close here because the coefficient of the stretching term is itself an unknown.

Most proposals for $Z_3$ parafermions in fractional quantum Hall-superconductor structures used the spin-unpolarized $\nu = 2/3$ Halperin $(1,1,2)$ state. The fractional quantum anomalous Hall (FQAH) states of twisted MoTe$_2$ and rhombohedral graphene are believed to be spin- and valley-polarized Jain states, with the same topological order but a different spin structure. We analyze a trench between two polarized $\nu = 2/3$ edges. Among the pairing and tunneling bilinears of the edge-mode electrons, exactly two of each have zero conformal spin. Pairing is compatible only with the neutral-mode superconductivity channel and tunneling only with neutral-mode backscattering, whereas in the unpolarized state singlet pairing and tunneling select the same channel. Since the two channels condense the same neutral anyon up to a local operator, the domain walls carry $Z_3$ parafermions, with 3 protected states per pair, whether or not the neutral channel changes between regions; fixing the fermion parity does not enlarge this count. Before the neutral gap forms, pairing is at best marginal for repulsive inter-edge coupling unless an attractive neutral coupling favors its channel; once the gap has formed, both charge terms are relevant for an inter-edge Coulomb coupling below 7/25 of the intra-edge one. The superconductor must supply even-frequency, equal-spin, intravalley pairing, which an s-wave singlet superconductor with only Ising spin-orbit coupling does not provide at leading order, whereas chiral $p+ip$ and $f+if$ states do. The $Z_3$ label is carried by a quasiparticle of the FQAH state, so no process confined to a junction whose tunneling segments are not FQAH can change it, and we expect a $6\pi$-periodic Josephson current with or without fixed fermion parity.

The plasma channel of the AWAKE plasma wakefield acceleration experiment is created by laser ionization of rubidium vapor in a 10~m long vapor source. Verifying the properties of the channel — the radius of its fully ionized core and the width of its boundary sheath — that are vital for ensuring good-quality wakefields is important as is validating numerical models of ionizing pulse propagation. Near-resonant schlieren imaging can detect the plasma channel in the low-density vapor, but there is no direct inversion to obtain channel parameters from the images. We use symbolic regression to find compact analytical formulas that connect measurable properties of the schlieren signal to plasma profile parameters. Using databases of simulated signals generated with three different plasma edge-profile families, we obtain formulas with one and two fitting constants to predict the equivalent channel radius with an accuracy below twice the camera pixel size and formulas to predict the sheath width. Re-evaluating previous measurements with these formulas, we confirm a theoretically predicted power-law relationship between the channel radius and the energy of the ionizing laser pulse. The formulas can be adapted to new experimental conditions by refitting the constants on a modest set of new samples, providing a lightweight, interpretable alternative to evaluation with deep neural networks.

Keyed modulation is often judged secure by its key-space size and a key-less receiver's error rate, evidence that fails when the key is reused and the waveform is harmonically coupled. For a phase-keyed Fourier-curve constellation, whose $k$ tones share one data parameter, integer relations among harmonic indices yield data-cancelling mixed moments that expose key characters. A modular relation lattice characterizes them; for consecutive harmonics, third-order moments recover the relative phases and a fourth-order moment completes the key up to cyclic relabeling whenever its coefficient is nonzero, as in all evaluated settings. A non-data-aided relation-moment estimator with code-aided resolution of the relabeling turns this leakage into an attack without key-space enumeration. With one key per 168-symbol codeword, the attack meets a predeclared $0.1$ compromise criterion at eleven of twelve operating points of a $(3,6)$ LDPC-coded link; at the middle noise level, the eavesdropper's block error rate is below $0.04$, and $0.03$ for consecutive harmonics without artificial noise when a key spans only $28$ symbols. Tangent artificial noise and a harmonic set without relations below order four raise her error rate at intermediate reuse lengths but do not remove the one-codeword vulnerability. On a $2^{128}$-key grid at the same noise level, the cheapest tested refresh schedule whose $95\%$ lower confidence bound on her block error rate reaches $0.9$ costs $3.54$ fresh key bits per information bit, and no tested schedule on either grid meets even a $0.5$ target with less key than a one-time pad.

Let $(\mathcal{C},\mathbb{E},\mathfrak s)$ be an extriangulated category equipped with a proper class $\xi$ of $\mathbb E$-triangles.We establish stability under iteration for the quasi-$\xi$-Gorenstein projective and injective subcategories and extend the relative Ext-vanishing criteria for their finite homological dimensions to larger coefficient subcategories. An example in a bounded homotopy category shows that $\mathcal{P}(\xi)\subsetneq\mathcal{QGP}(\xi)\subsetneq\mathcal{GP}(\xi)$. For module categories, every quasi-Gorenstein projective module over a left perfect ring or a commutative Noetherian ring of finite Krull dimension is projective. When a commutative ring admits a finite chain of trace and nilpotent quotients ending in an Artinian ring and the endomorphism rings at the trace steps have finite right global dimension, we construct finite families of quasi-projective test modules that detect prescribed upper bounds on projective dimension for modules of bounded cardinality through Ext vanishing in finitely many consecutive degrees.

Background: Rapid growth of high-throughput molecular data demands systems for efficient retrieval, integration, and scalability across omics layers. Traditional file-based workflows hinder cross-modal analysis and reproducibility because of fragmented storage and ad-hoc querying. Few existing tools for genomic variation data prioritize modular multi-omics integration. We developed vcf2db, a sample-centric relational framework modeling each omics modality as a distinct but linkable component centered on biological samples. This study evaluates whether this design delivers competitive genomic retrieval while enabling extension to additional molecular layers. Results: We implemented a proof-of-concept genomic schema and ingestion pipeline for annotated VCF data using the European subset of the 1000 Genomes Project (502 samples, 25 million variants). We benchmarked it against three established VCF-oriented tools on seven retrieval tasks: coordinate filtering, annotation-driven queries, genotype extraction, and aggregation. Under controlled conditions, vcf2db performed strongly on selective queries, often outperforming other systems for coordinate and annotation filters, and remained usable for genotype retrieval. Aggregation-heavy tasks were less efficient, indicating optimization targets. We also validated modular extensibility by adding a synthetic transcriptomic layer without modifying genomic tables, linking layers via shared sample identifiers. Conclusion: vcf2db supports cross-layer retrieval directly as SQL queries anchored on shared sample identifiers, enabling integrated multi-omics access that is difficult with file-based approaches.

We prove that the tangent sheaf of $\SU_C(n,L)$ is slope stable for every smooth complex projective curve $C$ of genus $g\geq2$, every $n\geq2$, and every determinant $L$. For $g\geq3$, we prove a more general stability theorem on moduli spaces of principal bundles for connected reductive groups, using cohomology sheaves associated with representations in place of tangent sheaves. The adjoint representation gives tangent sheaf stability for the case of almost simple groups. The moduli spaces may be singular, and stability holds with respect to every ample Cartier polarization.

With the continuous expansion of aquaculture, precise and efficient monitoring of fish behavior has become increasingly critical for improving farming efficiency and reducing economic losses. In particular, with the ongoing enhancement of computational capabilities in deep learning models, vision-based fish segmentation methods are garnering growing attention. By analyzing video segmentation results, fish behavior can be effectively tracked, thereby providing reliable data support for the precise regulation of aquaculture environments. However, existing deep learning-based video segmentation methods for aquaculture scenarios often overlook the dynamic correlations between video frames. In contrast, Interactive Video Object Segmentation (IVOS) employs an interaction-propagation scheme to achieve high-precision segmentation while minimizing user effort, thereby enhancing monitoring efficiency. Yet, IVOS applications in aquaculture remain limited due to data scarcity, and are susceptible to error accumulation and mask loss over long sequence propagation due to high intra-class similarity. In response, this paper proposes an improved interactive video object segmentation method (FiVOS) and constructs two fish-specific datasets. FiVOS utilizes a mask block filter to enable early detection and correction of erroneous propagated mask blocks, enhancing filtering accuracy through a rule-based thresholding approach. Additionally, it serializes noise filters to further eliminate erroneous mask noise, thereby improving model robustness. Experimental results demonstrate that FiVOS achieves state-of-the-art (SOTA) performance in fish video segmentation tasks, providing robust technical support for fish behavior research.

We consider the semiclassical magnetic Schrödinger operator in dimensions one and two. The first result is that an even one dimensional semiclassical potential is spectrally determined by the Schrödinger operator. The second result states that a two dimensional radial magnetic field is spectrally determined by a purely magnetic Schrödinger operator. This extends the earlier work of Guillemin-Wang and answers a conjecture of theirs therein.

Large language models (LLMs) are increasingly deployed as computational engines in autonomous decision-making and planning loops, yet their systems and control treatment remains hindered by architectural simplifications, index conflations, and informal descriptions of tool interactions. This paper presents a control-theoretic formulation of decoder-only language models as multi-index, multi-rate systems, and sets the stage for a stochastic hybrid systems framework to govern the multi-phase dynamics of agentic tool interaction. We formalize the architecture across three hierarchically coupled evolution indices: (i) an ultrafast feedforward cascade of transformer blocks across layer depth, where layer normalization is cast as a spherical projection and key–value caching is proven to be an exact internal state realization via causal prefix invariance; (ii) an uncontrolled stochastic difference recursion over token generation steps, where finite context truncation induces a time-homogeneous Markov chain; and (iii) an autonomous mode-switching mechanism governing transitions between token generation and tool execution regimes, where tool invocations are triggered upon trajectory arrival at switching manifolds, followed by exogenous state jump maps that augment the context string with external observations. By defining a prompt-dependent evaluator over successive evaluable claims, we obtain a task-level error process whose fault-free histories and expected error growth admit bounds under conditional fault-hazard and error-drift assumptions.

Tracking seasonal change in crops and forests requires observing the same plants repeatedly. Ground robots can do this at close range, and a manipulator gives their sensors more viewpoints. Yet the robots behind long-term field datasets are rarely released with their design files, and how a robot's own structure limits arm reach and occludes its sensors is seldom compared between builds. We present ALFRED, an open-source mobile manipulator built from commercially available components. It carries a six-degree-of-freedom arm, LiDAR, RGB-D cameras, RTK GNSS and an IMU on an Ackermann-steered base, all mounted on a reconfigurable aluminium strut frame, and runs containerised ROS software. It was developed through four builds against six requirements for repeated outdoor deployment: durability, modularity, repairability, sensing reach, endurance and reproducibility. Model-based analysis of the last three builds shows the usable share of the arm's reachable poses rising from 34.0% to 60.0% and then 66.1%, and ray casting shows that only the final build keeps the frame-mounted LiDAR's horizontal view clear both forwards and backwards. ALFRED completed a year of monthly forest surveys (528 traversals) without missing a scheduled collection. This was despite battery degradation, reconfiguration for another researcher's study, and the parallel development of ALFRED 2.0 for autonomous crop-row operation, with each switch between builds taking about six hours. The deployment also showed that mechanical modularity is only as dependable as the robot description that tracks it.

We study closed timelike curves (CTCs) in three rotating spacetimes arising in modified gravity: the tidally charged braneworld black hole, the Kerr-Newman black hole in Rastall gravity, and the charged Johannsen-Psaltis spacetime. Using the sign of $g_{\phi\phi}$ on the equatorial plane, we show that an additional parameter of each geometry controls the chronology-violating region, and we obtain critical parameter values analytically. In the braneworld case, the positive $r$ CTC region exists only for positive tidal charge and disappears completely for $q\leq 0$. In Rastall gravity, the extent of the CTC region depends jointly on the Rastall coupling and the surrounding matter. For each matter type we find a finite range of the coupling for which the CTC region is absent. In the charged Johannsen-Psaltis spacetime, the deformation parameter can either increase or decrease $g_{\phi\phi}$ depending on the radial position, and we derive a window of its values for which the CTC region is removed from the radial domain considered. Thus, in all three examples, the additional parameters can be chosen to control the extent of equatorial CTCs, and even to eliminate them entirely.

We consider two coupled linear heat equations on different spatial domains that interact through a lower dimensional interface. This models conjugate heat transfer. The problem is solved using Dirichlet–Neumann waveform relaxation. This allows the subproblems to be solved using separate codes, a so called partitioned approach. Our overall goal is to develop more efficient partitioned methods, and to this end, we want reliable error estimates. Here, we use an exponentially weighted Fourier technique to derive new error estimates in $l^2$ for finite time $T$ in the fully discrete setting. These describe both linear and superlinear behavior. We show that the fully discrete estimate is close to a previously obtained time discrete estimate and independent of $\Delta x_1$ and $\Delta x_2$ when the CFL number in the subsolvers is large. We also show that the convergence behaviour depends on the ratio $\Delta x_1/\Delta x_2$ when the CFL number is small. Our numerical experiments show that the fully discrete estimate is accurate across a wide range of grid sizes $\Delta x_1, \Delta x_2$, time step sizes $\Delta t$ and $T$.

Integrated Sensing and Communication (ISAC) is a foundation of 6G networks, demanding architectures that simultaneously enhance sensing accuracy and communication reliability. This paper presents a Reconfigurable Distributed Antenna and Reflecting Surface (RDARS) aided ISAC framework, where an RDARS overcomes the limitations of conventional passive Reconfigurable Intelligent Surfaces (RIS) and Distributed Antenna Systems (DAS). By enabling each element to dynamically operate in either reflection or connection mode, RDARS synergistically harnesses reflection gain, distribution gain, and an additional mode-selection gain. We investigate the joint optimization of transmit beamforming at the base station and dynamic mode selection at the RDARS to minimize the sensing performance metric, namely the Cramér-Rao Bound (CRB) for target localization, while guaranteeing a minimum required Signal-to-Interference-plus-Noise Ratio (SINR) for multiple communication users. To solve the resulting non-convex and mixed-integer problem, we develop an efficient iterative algorithm based on the Alternating Optimization (AO) framework, effectively leveraging Majorization-Minimization (MM) and Penalty methods. Comprehensive simulations validate the proposed design, demonstrating that the dynamic RDARS configuration achieves a superior trade-off between the Position Error Bound (PEB) and communication SINR, significantly outperforming benchmark passive RIS and DAS systems.

We numerically study exciton-photon coupling in a hybrid structure composed of a period-doubled Si3N4 photonic crystal slab and an hBN-encapsulated MoSe2 monolayer. Period doubling folds quasi-guided modes into the light cone and produces spectrally separated photonic branches whose radiative character is controlled by the symmetry of the doubled unit cell, resulting in one bright and one bound state in the continuum (BIC)-like mode at normal incidence. We then use the hBN thickness as an independent tuning parameter for the vertical field profile. Increasing the hBN thickness redistributes the resonant field toward the MoSe2 monolayer and enhances the Rabi splitting from about 3-4 meV to 8.5 meV, corresponding to an increase in the coupling strength from approximately 1.5-1.7 meV to 4.3-4.4 meV. Comparison between bright and BIC-like configurations shows that the quality factor primarily determines radiative visibility and linewidth, whereas the coupling strength is mainly governed by the field amplitude at the monolayer. We further show that the hBN layer controls the hybridization of folded TE-like modes at finite momentum, leading to an additional photonic gap whose size can be tuned by the hBN thickness and closes at an hBN thickness of approximately 113 nm. In the hybrid structure, we observe finite-momentum exciton-polariton states that, depending on the hBN thickness, form either through coupling to an isolated photonic branch or through coupling to two nearly degenerate photonic modes. These results identify period-doubled photonic crystal slabs as a flexible platform for controlling radiative coupling, exciton-photon overlap, and polariton dispersion in hybrid structures based on transition-metal dichalcogenide monolayers.

Comparing two high-dimensional discrete distributions has always been a challenging task due to the exponentially growing state space and complex changes in interactions. A recent work suggests comparing distributions through a vector field trained using flow matching between two continuous distributions. The resulting vector field at mid-point vanishes if and only if two distributions identical. However, such a flow-based criterion does not naturally apply to discrete distributions. We extend this principle to the discrete domain and introduce the Zero Flux criterion, a discrepancy based on local probability fluxes. Under independent coupling, we show that all local probability fluxes vanish at the midpoint if and only if two distributions are the same. This discrepancy decomposes the joint distributional difference into smaller, local contributions and can be efficiently estimated from samples. We establish finite sample error bounds for our estimator. Experiments on synthetic and real categorical data demonstrate reliable recovery of sparse dependence signals and stable tracking of distribution shifts in high dimensions.

Large language models now generate code, documentation, and analyses, and are increasingly used to review such output. We ask when a second review by a different model helps. Building on the author's earlier preprints, which varied context, repetition, and role structure within one model, we test model independence in a controlled experiment: 30 artifacts with 150 planted errors, 10 review conditions, and 900 review sessions with three reviewer models from two developers. In this experiment, (1) a top-tier cross-model reviewer is not significantly different in F1 from same-model review in a fresh session (CCR), which does not establish equivalence; (2) the two find partly different errors (Jaccard 41.2%); and (3) at two review calls, one CCR plus one cross-model review matches more planted errors than two CCR reviews (56.7% vs. 42.7%; Holm-adjusted p=.006), but not significantly more than two reviews by the top-tier cross-model reviewer, so model difference and reviewer capability are not separated. A lightweight cross-model reviewer scores no higher than same-model review. Withholding requirements from the reviewer raises F1 for the two lower tiers but not the top tier, in untested point estimates whose pattern depends on how failed sessions are scored. Before analysis we audited all session records, excluding one baseline run of uncertain provenance and 14 failed calls; results with all sessions are also reported. A partial check on public detector outputs from another benchmark neither replicates nor contradicts the main comparison. Records, artifacts, and scripts are available from the author on request.

Mechanical properties of rubble-pile asteroid regolith remain poorly constrained because contact responses depend on both material properties and local particle configuration. This study examines whether early-time high-speed penetrator dynamics can constrain particle stiffness, represented by the particle Young's modulus E, within a controlled discrete-element model. A bidirectionally coupled EDEM-Adams discrete-element-multibody model is used to screen six parameters: Young's modulus, Poisson's ratio, coefficient of restitution, static friction coefficient, rolling friction coefficient, and adhesion strength. Five Young's-modulus levels are then tested at the same 17 spatial locations in one settled granular bed, giving 85 simulations with location-matched comparisons across modulus levels. Within the 75-90 m s^-1 speed range, varying Young's modulus yields the clearest and most systematic differences in probe deceleration, whereas the other parameters have weaker effects over the tested ranges. Validation by withholding all modulus cases at one location in turn indicates that the early response is more informative for distinguishing broad modulus ranges than adjacent modulus levels. Mesoscale analysis links the response differences to concentrated interface load sharing and spatially extended three-dimensional high-capacity contact paths as Young's modulus increases. The results provide a numerical basis for narrowing the particle Young's-modulus range from high-speed penetration responses despite local granular variability. Quantitative application to real asteroid regolith requires further experimental validation.

Given a prime $p$, the Nottingham group $G(p)$ is the group of formal power series $x+a_2x^2+a_3x^3+\cdots$, $a_i\in{\bf Z}/p{\bf Z}$, under substitution. For $n\geq 1$, we write $G_n(p)$ for the quotient of $G(p)$ by its normal subgroup $K_n=\{x+a_{n+1}x^{n+1}+a_{n+2}x^{n+2}+\cdots\,|\, a_i\in {\bf Z}/p{\bf Z}\}$, namely the group of truncated polynomials $x+a_2x^2+\cdots+a_nx^n$, $a_i\in {\bf Z}/p{\bf Z}$, under substitution, which is a $p$-group of order $p^{n-1}$ generated by $x+x^2,\dots,x+x^n$. In this paper, we determine the power-commutator presentation of $G_{15}(p)$ relative to $x+x^2,\dots,x+x^{15}$. This, in turn, automatically gives the power-commutator presentation of $G_{n}(p)$ relative to $x+x^2,\dots,x+x^{n}$ for any $1\leq n<15$, as well as the initial segments of the power and commutator relations in $G_{n}(p)$ corresponding to $x+x^2,\dots,x+x^{15}$ for any $n>15$.

Low-cost NAS must both explore high-performing architectures and identify them reliably, yet reducing evaluation cost often weakens the fidelity of candidate comparisons. Training-free methods reduce evaluation cost by replacing learned task feedback with proxy signals measured at initialization. We introduce LESS (Lightweight Evolutionary Supernet Search), a data-driven method that combines a brief fair hard-path warm-up with discrete search under a single CMA-ES distribution. Each proposal is evaluated as its decoded hard genotype after six candidate-conditioned supernet updates. On NAS-Bench-201, LESS achieves \(93.189\pm0.467\%\) CIFAR-10 test accuracy in 409.1 seconds, coming within 0.04 percentage points of FairNAS using approximately \(1/24\) of its source-reported search time. Matched controls show that calibration improves selected validation accuracy by \(0.577\) percentage points while changing best-visited accuracy by only \(0.054\) points, indicating that its primary effect is to reduce selection regret. The frozen configuration transfers without tuning to CIFAR-100 and ImageNet16-120 with \(69.615\pm1.139\%\) and \(43.720\pm1.697\%\) accuracy. Applied without tuning to the larger DARTS space, LESS achieves \(96.95\pm0.14\%\) on CIFAR-10 and \(82.43\pm0.80\%\) on CIFAR-100, with each search completing in approximately 43.5 minutes on a single GPU. Together, these results show that short, balanced, data-dependent updates enable competitive neural architecture search across datasets and search spaces within minutes.

We study the genus-$19$ Albanese fibration of the Cartwright–Steger surface and its order-three symmetry. Passing to the orbifold quotient of the elliptic base gives a relation among the two handle monodromies and the three Dehn twists about the vanishing cycles. The two real vanishing paths give cycles whose real fixed points lie on different ovals of the invariant fiber. For a cyclic triple cover, we express the intersection of a curve with its image under the deck transformation as a signed crossing count on the quotient. We determine the two local branch values of a degree-$72$ bicanonical pencil near a node and study a symmetric bicanonical pencil. A general pencil also gives a finite degree-$72$ map from a blowup of the surface to the product of the elliptic curve and a projective line. The monodromy translates of the nodal vanishing classes span the first homology of a smooth fiber.

The Pauli twirl of amplitude damping (AD) is generalized amplitude damping at infinite temperature: it keeps the contraction of AD and removes its non-unital term, so comparing the two in variational circuits isolates the zero-temperature bias, which acts mainly through the scale of the features. For random parameters, features under AD settle on a floor, which at strong damping is set by the last layer, has a closed form, and at fixed $T_1$ falls with temperature as $\tanh(\hbar\omega/2k_BT)$; under the twirls they shrink by a constant factor per layer, up to eight qubits. A trainable output scale removes most of the resulting accuracy differences, leaving AD ahead of its twirls by at most about three percentage points in our simulations; what it removes reappears as a cost in measurement shots: trained and tested with $10^3$ shots per image, a four-qubit classifier under AD at $p=0.3$ stays within 1.5 points of noiseless accuracy, while the twirled classifiers lose up to 33. At weaker damping the separation depth grows roughly as $(np)^{-1}\ln(1/p)$. The damping direction is a gauge when the damping follows complete entangling layers and the circuit boundaries are trainable; in an eigensolver it becomes physical inside a decomposed two-qubit gate.

The increasing volatility of electricity prices driven by renewable energy integration, market shocks, and regulatory changes has reinforced the need for forecasting methods that go beyond point predictions and accurately describe the full conditional price distribution. This paper applies the Generalised Additive Models for Location, Scale and Shape (GAMLSS) framework to forecast Spanish day-ahead electricity prices using hourly data from 2020 to 2024. Alternative specifications based on Normal, Johnson's SU (JSU), and Sinh-Arcsinh (SHASH) distributions are considered, allowing the location, scale, and shape parameters to vary with market fundamentals, including electricity demand, renewable generation, seasonal effects, and regulatory and geopolitical risk factors. Forecasts are generated using a rolling-window approach and evaluated through the mean absolute error (MAE), pinball loss, and Diebold-Mariano tests. The results show that flexible distributional specifications improve forecasting performance relative to a naive benchmark and the standard normal specification. While SHASH and JSU specifications provide the lowest point forecasting errors, the hourly analysis reveals substantial intraday variation in relative performance across specifications. JSU specification with all four parameters driven by covariates achieves the best probabilistic forecasting performance, particularly in the tails of the distribution. Diebold-Mariano tests confirm the statistical significance of these improvements. These findings highlight the importance of modelling time-varying shape distributional parameters and demonstrate the value of GAMLSS models for forecasting and risk management in increasingly volatile electricity markets.

Selecting difference-in-differences controls from a candidate pool by pre-treatment fit can alter the estimator's distribution and cause undercoverage despite parallel trends for every candidate. We develop the first empirically implementable selective inference for the average treatment effect on the treated. We estimate the full covariance matrix from the same observations and condition inference on the full selection event. Joint Gaussian approximation of outcome means yields uniform asymptotic validity including perfect fit and population ties. Endpoints approach known-covariance counterparts in standard-error units; widths shrink with effective sample size despite vanishing regional sample shares. We cover staggered adoption, reused observations and finite fitting-rule searches favoring larger estimates. At fixed smoothing, skewness and mean–covariance corrections give randomized tests second-order accuracy for joint selection and rejection probabilities. The adjusted interval for police deployment's effect on car theft includes zero; intervals for minimum-wage employment effects admit larger losses at $\$5$ to below $\$14$ per hour.

Thematic corpora, such as social media communities, contain unstructured text describing data that could be made structured. These include, for example, personal attributes, behaviors, and experiences mentioned in social media data. Extracting structured data is challenging as relevant attributes are often implicit, domain-dependent, and unknown in advance. We propose a fully automated, iterative framework for discovering and extracting domain-specific attribute schemas without a predefined ontology. Using large language models (LLMs), the framework induces candidate attributes, sequentially consolidates semantically overlapping attributes, and assigns a structural type. These enable creating an ontology and populating it with values from the corpus. The framework also enables the use of smaller LLMs for value extraction with estimable accuracy loss compared to large LLMs. We evaluate the framework on 5 health-related Reddit communities. Discovered attributes achieved 61% agreement with human-identified attributes, close to the 62% agreement between independent annotators. In most cases, the algorithm converges to a stable attribute set in fewer than 10 iterations. Structural type assignment achieves 82% accuracy, and value extraction reaches an F1 score of 0.8 compared to human annotations. Across four LLM families, smaller instruction-tuned models show statistically significant improvements in extraction performance with model scale when evaluated against a high-capacity reference LLM, supporting informed accuracy-cost trade-offs. These results show that attributes comparable to those identified by humans can be discovered automatically, enabling the creation of high-quality structured datasets economically and at scale. By removing the need for predefined ontologies, iterative model-driven schema induction offers a practical and scalable foundation for mining thematic corpora.

Indefinite causal order is known to offer enhancements in quantum metrology, notably including an unbounded advantage in the measurement of a geometric phase of a harmonic oscillator. This advantage, however, is specific to the infinite dimensional setting, and its finite dimensional analogue remains elusive, with recent findings suggesting that advantages in finite dimensions may be fundamentally limited to bounded constant factors. Here we show that, in fact, arbitrarily large advantages arise for finite dimensional systems in the finite sample regime. Specifically, we establish an unbounded separation between definite and indefinite causal order in the estimation of a geometric phase associated to two sets of $N$ displacements generated by discrete position and momentum operators on a $d$-dimensional quantum system: for any given constant $R$, there exist values of $N$ and $d=\Omega(N^2)$ such that a strategy with indefinite order uses an initial probe with $R$ times less energy than the probe required by every strategy with definite order achieving the same mean squared error, whenever the number of measurement shots $\nu$ is bounded as $\nu=\mathcal{O}(\exp(\pi d/16)/\mathrm{poly}(d))$. In other words, indefinite order offers an energy saving that grows arbitrarily large with the parameters of the problem. To prove this result, we establish an approximate Weyl relation for discrete Gaussian wavepackets, which is of independent technical interest.

We often plan ambitiously yet act habitually and wonder, in retrospect, whether we would have planned differently had we known what we would actually do. Hindsight offers a valuable perspective on past decisions, although we often wish we could have simulated hindsight at the moment of choosing. If a system could generate plausible trajectories from one's personal history, such previews might help people formulate more realistic plans and make better informed decisions. We introduce NextMe-800, an approximately 800-hour first-person dataset from one volunteer over 126 days with 1 Hz images, gaze, and audio, captioned at five hierarchical abstraction levels from atomic actions to major activities. We formulate personalized action anticipation as open-vocabulary K-step sequence prediction and construct NextAct, a 1,500-point benchmark combining NextMe-800 with the multi-person EgoLife dataset. Using an embedding-based soft edit distance as the metric, we evaluate how well different models can anticipate personal behavior across abstraction levels and prediction horizons. NextMe-800 and NextAct provide a months-long resource and evaluation framework for studying how far ahead personal behavior can be anticipated from egocentric observation.

We show that free groups admit amenable actions on unital simple AF-algebras. This settles, for free groups, the last remaining case of the existence problem for amenable actions of non-amenable groups on classifiable simple C*-algebras. The proof relies on the Baire category theorem. This result was obtained with substantial assistance from GPT-6 Astra.