Papers

A constrained optimization problem may involve a parameter in its objective and active constraints, yet the final decision may remain insensitive to small changes in that parameter. This raises a fundamental question: which inputs does a decision making system truly depend on? Building on this question, we introduce DePICT, a procedure for constructing decision preserving interfaces by ranking context directions according to the optimizer's solution sensitivity and aggregating them across an operating regime. We study this problem in a high dimensional setting where primitive context parameterizes a constrained task and the downstream agent observes only a selected subset of context directions. For locally regular constrained programs, we derive a Karush Kuhn Tucker (KKT) based characterization of when a context direction is optimizer relevant. Our analysis shows that appearing in the active optimization problem does not necessarily imply that a variable affects the final decision. Some context directions can alter the KKT conditions while leaving the optimal solution unchanged because their effect is absorbed by the dual variables. DePICT is designed to remove exactly these directions. In a controlled diagnosis, it recovers the decision relevant interface exactly and reduces linear predictor regret to 0.009, compared with 0.475 for the strongest competing baseline.

Microvesicles (MVs), a type of cell-secreted extracellular vesicle (EV), are essential for various biological processes such as local and long-distance intercellular communication and the disposal of unwanted materials. Recent evidence suggests that overexpression of MUC1 glycoproteins facilitates MV formation on the cell membrane, but the mechanism underlying this nonequilibrium process remains unknown. Here, the kinetics of glycocalyx-driven MV formation is studied in the framework of Onsager's variational principle. Using this deterministic theory, we first show that the process of MV formation is accelerated for the denser and thicker glycocalyx-grafted cell membrane and can be improved by reducing the ionic strength or enhancing the degree of glycocalyx ionization. Intriguingly, we further discover that there exists an optimal glycocalyx coverage for the fastest formation of MVs. We then investigated the process of MV formation using a stochastic approach. Our results suggest that the transition rate is not constant, leading to the prediction of a shorter formation time compared to the deterministic approach. These findings provide valuable insights into the regulation of membrane morphology and MV secretion by the glycocalyx, with implications for both understanding and modulating these processes.

The Event Horizon Telescope array has measured the average polarimetric spiral morphology of black hole accretion disks using the so-called “$\beta_2$” coefficient. This measurement is dominated by the weakly lensed, direct ($n=0$, counting photon half-orbits) image of the disk, providing insight into magnetic field structure around the M87* black hole. The future space-based Black Hole Explorer mission is expected to extend these capabilities, allowing for $\beta_2$ measurement of the strongly lensed $n=1$ image. In this paper, we explore the constraints $\beta_2$ measurements would impose on spacetime and plasma geometries using a semi-analytic conical emission model. We show that magnetic field orientation with respect to the emission cone qualitatively changes the dependence of $\beta_2$ modes on model parameters. For cone-aligned magnetic fields, we find that the leading-order theoretical prediction from equatorial emitters of $\beta_{2,\, n=0} \approx\beta_{2,\, n=1}^*$ is the dominant trend even in off-equatorial models. Specifically, we show $\arg(\beta_{2,\, n=0}\,\beta_{2,\, n=1}^*) \approx 4\chi$, where $\chi$ is the angle determining in-cone magnetic field orientation. We also find that $\arg(\beta_{2,\, n=0}\beta_{2,\, n=1})$ minimizes the effect of magnetic field geometry while preserving a dependence on spin that may be used to measure spin direction. We find that observations of the photon ring's $\beta_2$ complement the constraint from direct image polarization, and can greatly constrain magnetic field structure even with an unknown external rotation measure. We also find that black hole spin and emission cone opening angle would not be tightly constrained by $\beta_2$ alone, unless strong priors are placed on the magnetic field geometry.

When using diffusion models to target image restoration problems, diffusion inversion is typically employed to retain relevant image information from the degraded images. Instead of inverting back to the initial time step (i.e., T), many methods invert to a pre-determined intermediate time step, in order to better preserve information from degraded source images. However, a pre-determined time step for inversion is not ideal for reconstruction, as a severely degraded image requires an earlier starting time step than a mildly degraded one. In addition, DDIM-based models corrupt the original signal by adding Gaussian noise, which can be mismatched to the nature of blur-like degradations, such as blur, haze, and low-light. To address these problems, we propose two solutions: (1) we adopt an alternative diffusion process, called the Inverse Heat Dissipation Model, that diffuses the input image by gradually blurring a data point (2) we propose to implement a time predictor to estimate the starting time step for the inversion, with the model learning to adapt to the degradation severity. Extensive experiments on standard benchmarks show that our method achieves state-of-the-art performance in both quantitative and qualitative evaluations, with excellent generalization to many restoration tasks.

In this paper, we study the ferromagnetic nearest-neighbor Ising model on rooted spherically symmetric trees with one or \(q\) children per vertex, according to a prescribed sequence of branching levels. Unbounded runs of one-child levels force uniqueness at every nonzero homogeneous field and every finite inverse temperature. We quantify this effect through the locations of sufficiently long nonbranching stretches. Conversely, uniqueness need not entail any uniform rate: for each prescribed vanishing sequence, we construct a density-one tree whose finite-volume boundary influence decays more slowly along a subsequence. The tree retains branching number \(q\), although its nonzero-field coexistence region disappears. For eventually bounded gaps, we prove an optimal uniform comparison with the periodic tree having the largest allowed gap and obtain an explicit interval of coexistence fields. Thus bounded gaps are equivalent, within this family, to coexistence at some finite temperature and nonzero field. For eventually periodic gaps, Gibbs uniqueness is equivalent to fixed-point uniqueness of the full-period return map. A negative Schwarzian derivative yields a complete classification: in the supercritical regime the map has three fixed points inside a closed coexistence interval, two at its endpoints, and one outside. The nonzero endpoints are automatically nondegenerate folds, with square-root splitting of the merging branches.

During fine-tuning, a language model can assign less probability to previously learned answers even when the current gradient acts to preserve that probability. With momentum, each update also carries gradients computed at earlier model states, and these stored contributions can push the model in the opposite direction. We investigate how this optimiser memory contributes to forgetting by separating old-task loss into confusion among its answers and leakage of probability outside the answer set. Across three language-model families, answer mass consistently declines while discrimination among old answers usually improves: the model becomes less likely to produce answers that it can still distinguish correctly. Decomposing Adam updates reveals opposing contributions to this loss of answer mass. Over training, accumulated history favours leakage, while the current gradient opposes it. Resolving history by age shows that the harmful contributions come mainly from older gradients of the new task, whereas recent gradients tend to protect the old answers. Changes in history's effect are dominated by its orientation relative to the old-task gradient. Interventions that reset momentum while matching the initial update norm establish that stored history affects retention, with state-dependent immediate effects and lower final old-task loss over longer Adam continuations, mainly through recovered answer mass. Finally, integration along finite updates shows that most sampled large loss increases are captured by local projections, while curvature along history amplifies some events. Together, these findings reveal how an optimiser's memory can erode learned behaviour even as its current gradient acts to preserve it.

Many prediction tasks evaluate a trained tree ensemble on groups of rows that share feature values: discrete-time survival models expand each patient into $G$ time steps, click-through-rate models score every item in a search session, and scenario analyses vary a few inputs while holding the rest fixed. Standard inference treats each row independently and ignores this input-side correlation. We present TreeWalker, which instantiates partial evaluation for grouped tree-ensemble inference: constant features are static, varying features dynamic. TreeWalker walks each tree once per group, partitions a row bitmask at varying splits, and skips empty subtrees via an unsplit shortcut. It reads standard LightGBM and XGBoost models; training is unchanged. We prove a structural work decomposition: per-tree work splits into the constant-projected subtree size $|T_c|$, $G$ leaf writes, and a predicate-mask provisioning cost $Q$. For the trace evaluator, per-row work approaches $(d_v+1)/(d+1)$ as $G \to \infty$, giving asymptotic speedup $1/(1-\bar\rho_G)$. On Intel, TreeWalker is ${\sim}3\times$ faster than a row-independent traversal at the reference model size ($T{=}500$, $L{=}8$; $G{=}16$ for SUPPORT/FLCHAIN and empirical groups for Expedia) and $7.7$–$9.1\times$ faster at $G{=}128$ on the survival datasets; on Arm, $2.5$–$2.9\times$ and $5.5$–$6.0\times$. On a scenario-analysis benchmark it is faster in all $16$ configurations on both machines. For f64 models, outputs match treelite GTIL up to tree-ordering roundoff; for f32 models, TreeWalker's f64 leaf accumulation is closer than native f32 to an exactly rounded reference on every audited row.

Agentic multimodal large language models (MLLMs) have recently pushed the frontier of visual reasoning by calling tools such as zooming and tagging. Despite the recent strong success of agentic MLLMs, this work uncovers a critical safety failure in the tool-use paradigm: agentic tool-using MLLMs become less capable of refusing harmful requests. Our experiments confirm that, across three popular safety benchmarks, all the top open- and closed-weight MLLMs we test exhibit significantly lower safety in tool-using settings than in non-tool settings, with a relative refusal failure rate increase of up to 68.7%. Based on analysis of 100,000+ responses, including extended experiments, we also propose two possible reasons for this safety degradation.

Aggregated relational data replaces a fully observed network with a compressed version, where survey respondents are asked questions such as “How many of your friends belong to group $G$?”. We consequently only observe a matrix $R = OA$ of the adjacency matrix $A$ with a fixed design $O$. In this paper we understand the effects of this compression on community detection under degree-corrected stochastic block models. First, we give guarantees for when it is possible to recover the full network parameters, and then only just the communities under conditions on a diagonal corrected version of $\mathbb{E}[R]$, which can be used to inform experimental design. Secondly, we show that spectral clustering with single linkage exactly recovers communities when $n\rho_n\gtrsim K^2+K\log n$ when there are $K$ balanced communities under “uniformly spread” or near-orthogonal designs, and characterize more generally how the design influences rates of convergence. We also introduce second-order “friends-of-friends” queries and develop the corresponding theory.

We study a dynamical model of repulsive needles with centers fixed on a square lattice. Needles that overlap are active and receive a repulsive rotation $ \theta \in [0,\epsilon)$; needles with no overlap do not move. This model exhibits an absorbing state transition. Nematic order was observed for long needles. For a range of needle lengths and $\epsilon$, the system self-organizes into a state with checkerboard symmetry. As $\epsilon$ is increased, this phase melts and becomes disordered. Remarkably, we find that this phase boundary exhibits critical exponents identical to those of the equilibrium 2D Ising model.

General decision models, such as Jev, have recently emerged as efficient alternatives to LLMs for structured judgment and selection. But what kinds of decisions can these models reliably make, and how does their behavior change when individual decisions are composed into larger systems? To study this, we introduce JEVal, a bilingual benchmark comprising 11,257 instances from 36 datasets across 10 application domains, and evaluate 25 model configurations spanning general decision models and generative LLMs. Our results show that (1) general decision models are most competitive when decisions can be resolved from available evidence, but weaken when they require specialist knowledge or faithful uncertainty estimation: they can often identify the most likely outcome while substantially overstating its probability. (2) In more dynamic and realistic systems involving long-horizon, multi-step interactions, the advantages of fast local decision making are offset by reliability failures at the system level. on $\tau$-bench, faster local decisions reduce median episode time but lower task success as decision errors accumulate over long trajectories. (3) In large-scale social simulation, decision models approach strong generative LLMs on individual response prediction at substantially lower inference cost, yet remain weaker in user profiling and exhibit larger aggregate estimation errors and systematic bias. Finally, we propose SII-Decision-4B and SII-Decision-27B, which internalize an open-weight LLM's own reasoning into a single-pass first-token decision through Reasoning-to-Readout Self-Distillation, with SII-Decision-27B performing on par with Jev on JEVal while answering a typical query in about 0.1 s.

Existing analog integrated circuit design benchmarks make two questions hard to answer: whether a model has learned transferable circuit skills rather than recalled familiar examples, and whether its output works under defined process and test conditions. We introduce a benchmark built from human-designed, open-source circuits associated with Tiny Tapeout manufacturing shuttles. The collection contains 273 topologically distinct top-level designs. Every source is retrieved at the revision recorded for its shuttle submission and processed in a fixed containerized environment. The pipeline exports each eligible schematic image and its SPICE netlist from the same source file, giving transcription an exact structural reference. Commit dates support model-specific training-cutoff analysis, while author testbenches provide the simulation context for sizing. The benchmark evaluates schematic-to-netlist transcription and device sizing. Across seven models on a fixed set of 66 transcription tasks, the strongest model reaches 56.1% exact graph isomorphism, and six of seven models drop sharply from the small to the medium tier. For one frontier model, removing author-chosen labels reduces exact matches while preserving aggregate structural F1, suggesting that labels can aid connectivity tracing. On the 17 sizing tasks, the leading model converges on all 17 proposals and reaches 91.2 out of 100 against the human reference, while the two newest Claude models refuse 4 and 11 of the same prompts they transcribe without objection; a proposal without sizes scores zero. The two tasks produce different model rankings, exposing distinct visual and design capabilities and, in one family, a policy rather than capability limit.

The transient electrothermal technique (TET) is widely used to measure the thermal properties of thin fibers and films, but its accuracy declines when radiative heat losses are significant and surface emissivity is unknown. These conditions are difficult to avoid at the microscale, while conventional analyses either neglect them or rely on separate measurements that introduce sequential errors. We present a TET implementation that determines multiple thermal properties and effective surface emissivity from only two transient measurements. An analytical model incorporating thermal radiation and variable heat generation supports sensitivity-guided, multiparameter fitting. Two forms of the temperature solution yield thermal conductivity and diffusivity, and thermal conductivity and volumetric heat capacity, respectively. An iterative fitting procedure then recovers intrinsic properties corrected for environmental effects while reducing the required sample count and measurement time. Measurements of amorphous silicon dioxide (SiO2) optical fiber and Hi-Nicalon Type S silicon carbide (SiC) fiber agreed with literature values to within an average error of approximately 10% across all samples. Experimental and numerical uncertainty analyses quantify measurement variability and relate it to sample geometry. This method simultaneously resolves multiple intrinsic thermal properties, corrects radiative-loss errors, and provides quantitative guidance for sample geometry selection.

Public sector agencies have adopted participatory design as a prominent procedural safeguard proposed against algorithmic harm. Across methods, institutional contexts, and well-intentioned implementations, the participation gap persists. Communities engage, input is documented, and systems are deployed largely unchanged. This paper locates the persistence of that gap in the mechanism institutions use to document participation itself. We develop the concept of the legibility trap, in which institutional mechanisms for documenting participation systematically select for forms of community input that fit institutional categories and filter out the forms that would most constrain state power, including foundational dissent, demands for non-deployment, and what Kelly Oliver calls witnessing to structural conditions of harm. Drawing on James Scott's theory of state simplification, we demonstrate the legibility trap through critical discourse analysis of twenty-two official, community-produced, and public-record documents across three public sector AI deployments: the Allegheny Family Screening Tool, Detroit Project Green Light, and the Community Control Over Police Surveillance (CCOPS) ordinances in Oakland. We identify the structural conditions under which participation partially escapes the trap, and we theorize that repeated non-response constitutes a harm to community subjectivity that AI ethics frameworks have not yet named.

This paper presents sparse calibration-based personalization for kernel-based gait phase and walking speed co-estimation from wearable inertial measurement units (IMUs). A 28-speed reference library was constructed from bilateral thigh and shank trajectories in a 20-subject locomotion dataset. Rather than directly applying population-average kernels, the method combines a user-specific baseline estimated from three calibration speeds with a principal-component (PC) model of baseline-centered kinematic deviations. Offline validation showed that personalized kernels reduced phase error relative to the population kernel. In a single-participant online pilot evaluation, the personalized kernels produced numerically lower mean phase and speed errors and ran in real time on embedded hardware. However, the speed effect was not significant, and pairwise phase differences did not remain significant after multiple-comparison correction. The pilot evaluation establishes wearable implementation feasibility while indicating reference-to-online dataset mismatch and the need for larger-cohort validation.

Cooperative multi-agent networks deployed for target localization and tracking remain critically vulnerable to malicious cyberattacks, since a single compromised agent can corrupt the centralized target belief and mislead the estimation process across the entire network. This paper presents a resilient multi-agent target localization and tracking framework under a limited agent sensing range and false-data injection (FDI) attacks. The proposed method combines information-based navigation for target search and posterior target state uncertainty reduction, Bayesian attack graph (BAG) based attack detection for probabilistic identification of the compromised agents, and reachable set-guided recovery that guides the uncompromised agents to relocalize the target when the sole localizing agent is attacked. The framework preserves the network localization capability after an attack by trust-gating compromised sensing information and guiding attack-immune agents toward a dynamically constructed recovery set representing the new target belief using a Linear Quadratic Regulator (LQR). Simulation validates that the proposed framework maintains a robust tracking performance under FDI attacks, demonstrating the potential for resilient autonomy in safety-critical applications.

Entanglement entropy plays a central role in holography. The quantum extremal surface (QES) formula is the standard tool in settings where both dynamics and bulk quantum corrections are important, such as in black hole evaporation. We derive and explore QES-equivalent prescriptions — quantum covariant bit threads — that formulate holographic entanglement entropy in terms of minimal or maximal bulk flows rather than extremal surfaces. Like classical covariant bit threads, our prescriptions come in two types: (typically) spacelike V-flows, and timelike U-flows. To account for bulk quantum corrections, the flow lines can start and end at points in the bulk, with the distribution of endpoints controlled by bulk entanglement entropies. In our bit thread language, the QES is a quantum bottleneck to the flow, with some threads passing through it (accounting for the area term in the generalized entropy) and others jumping across it (accounting for the entropy term). Our prescriptions provide a complementary perspective to the QES formula and offer new insights on important phenomena such as black hole evaporation, islands, and baby universes.

Screen localization and pointing estimation are key to low-cost interactive devices. Yet developing and evaluating these algorithms requires realistic data: synthetic captures cannot fully reproduce the optical distortion, rolling shutter, motion blur, and display processing of a physical acquisition, and most existing datasets provide static images rather than the video needed to assess continuous pointing. In this paper, we introduce the DABACO Dataset. Developed within the DABACO (Dispositivo Apuntador de BAjo COste, or Low-Cost Pointing Device) project, this dataset supports the development and evaluation of screen detection and camera-based pointing algorithms for embedded systems. It comprises video sequences captured with multiple low-cost camera sensors, including monochrome global-shutter and color rolling-shutter modules, on embedded platforms such as the Raspberry Pi 4B and ESP32-S3. We present a systematic annotation pipeline combining temporary visual watermarking, optical-flow tracking, manual verification, and marker removal. Both the original marked captures and the marker-free images, together with explicit corner annotations and modification masks, are released to support auditing and the study of potential reconstruction bias. In addition, we release an open-source evaluation toolkit with two reference baselines: a classical screen-detection pipeline based on edge and contour geometry, and a closed-vocabulary, off-the-shelf YOLOv8 segmentation model evaluated without dataset-specific training. Both are evaluated on a general-purpose computer using Intersection over Union, corner error, and pointing error, establishing initial reference results for future embedded implementations. Overall, DABACO addresses a gap in existing resources and is intended to support the development and evaluation of new low-cost screen-localization and pointing systems.

Identifying patient subgroups with heterogeneous treatment effects is central to precision medicine, yet existing approaches face a tension. Black-box methods estimate individualized effects well but yield no interpretable subgroups, while tree-based methods produce explicit partitions but are unstable and less accurate in moderate-sample trials. We propose CausalSynthTree, a co-supervised method in which black-box causal estimators guide construction of an interpretable tree. It partitions the covariate space into cells, fits cell-wise conditional average treatment effect models from both observed data and synthetic data labeled by black-box causal teachers, and synthesizes them into a tree through a treatment-effect disparity criterion. Each leaf carries a sparse linear model, so the tree path defines a subgroup and the leaf model shows which covariates modify the effect. Honest inference for leaf-wise effects remains valid despite co-supervised augmentation and data-driven subgroup selection. In simulations it closes much of the gap between the two families. From moderate sample sizes onward it is more accurate than the black-box learners whenever the effect is linear or has a single threshold, and trails them only on a finer partition than the tree it grows. It controls spurious splits and attains nominal coverage throughout. In the ACTG175 HIV trial, where standard interaction tests detect no effect modification, the method reports no subgroups while the competing tree methods report several. In an adjuvant colon cancer trial it reports a borderline covariate in part of the resamples, together with a within-region age gradient that is consistent with an independent test.

The hydrodynamical Hopf-Rinow problem asks whether a given reconfiguration of an ideal fluid can be attained by an Euler flow. On the flat two-dimensional torus, we give an equivalent criterion for the strong version of this problem (which requires that the fluid motion minimizes the total kinetic energy required) in terms of metric betweenness. Toward this, we prove that sufficiently small normal neighborhoods in the volume-preserving diffeomorphism group of the flat $n$-dimensional torus, $n \geq 2$, equipped with the kinetic energy metric, are geodesically convex; resolving the strong problem locally in the $H^s$ topology. We further establish local bi-Lipschitz equivalence of the intrinsic geodesic distance and the extrinsic $L^2$ distance. Finally, we prove that the squared geodesic distance from the identity admits a $C^{1,1}$ extension from a sufficiently small normal neighborhood to the ambient $L^2$ space.

We compare the integrated kinematic properties of the Narrow-Line Region (NLR) of ~1500 type 1 and ~700 type 2 QSOs at 0.4 < z < 0.5 using emission-line profiles from SDSS-IV spectra. The NLR [OIII] 5007 and Hbeta emission lines are fitted using a two-component model comprising a narrow (v_n, sigma_n) and a broad component (v_b, sigma_b), while fainter lines are fitted with a single Gaussian component. Both QSO types exhibit similar median narrow-component velocities (v_n ~ 0) and velocity dispersions (sigma < 200 km s^-1), consistent with motion driven by the host galaxy's gravitational potential. However, parameters associated with the broad, high-velocity gas component (v_b, v_05, v_10, v_[NeV], sigma_b, W80 and W90) differ: QSOs 1 display larger blueshifts and broader emission lines than QSOs 2. This suggests that these components trace Active Galactic Nuclei (AGN) outflows, that appear stronger in QSOs 1, leading to higher estimated mass-outflow rates (\dot{M}_out) and kinetic powers (\dot{{E}_out). Rather than offering absolute determinations of \dot{M}_out and \dot{E}_out – which remain subject to physical uncertainties – this statistical study supports the higher NLR excitation found for QSOs 1 in Paper I, as an orientation effect: in QSOs 1, we observe gas closer to the AGN core, where velocities and velocity dispersions are intrinsically higher. Nevertheless, the trend of QSOs 2 exhibiting higher average [OIII] luminosities than QSOs 1 suggests an evolutionary contribution, where QSOs 2 represent an early obscured AGN phase followed by a blowout phase that reveals the central core as a QSO 1.

This paper addresses autonomous landing of unmanned aerial vehicle (UAV) rotorcrafts on a heaving ship deck using a recurrent policy trained via asymmetric actor-critic reinforcement learning: the critic sees 6 s of future deck height during training, while the actor sees only what the onboard sensors provide in flight, a 17-dimensional state made of the vehicle's position, velocity and attitude relative to the deck and outputs world-frame velocity and yaw-rate commands. Training is performed across 4096 parallel simulated environments, followed by fine-tuning in 16 environments with an onboard vision pipeline in the loop. A control-barrier-function (CBF) stopping margin on the deck-relative vertical state is incorporated at two stages: as a reward term during training, where it halves the median simulated contact speed relative to a policy trained without it, and as a runtime safety filter at deployment, evaluated at every control step to abort and retry the descent when the margin is violated. Because the autopilot's disarm logic cannot detect the vehicle resting on a moving deck, proximity-based thrust cutoff at touchdown is commanded directly in the landing pipeline. The proposed approach is validated using a parallel-manipulator-platform-based deck emulator that reproduces the heaving motion of the ship deck (scaled to 0.70 m peak-to-peak, 7.5 s mean period) and a quadcopter UAV with an onboard camera. Across 31 motion-capture and 20 vision-based trials, the UAV landed every time, with median times to contact of 6.5 and 7.3 s; 77% and 50% landed on the first attempt, with mean deck-relative speeds of 0.29 and 0.27 m/s, respectively, at the instant of thrust cutoff, which is the last speed under the policy's control. Supplementary video: https://youtu.be/S5hDkrSZJt4

LLM-driven sequential recommendation formulates next-item prediction as autoregressive generation conditioned on natural-language prompts. However, minor wording changes in semantically equivalent prompts can cause substantial performance fluctuations, undermining robustness and requiring costly manual prompt engineering. Continuous prompt learning reduces template dependence but faces two interacting challenges: shared task-level instructions lack user-specific reasoning guidance, while gradient updates can push continuous prompts outside the LLM's effective semantic space. Injecting personalized signals can further amplify this semantic drift. To address these challenges, we propose LRPRec, a learnable prompting framework that initializes continuous instruction prompts from discrete templates and introduces two complementary mechanisms. Personalized prompt injection encodes user behavior into a preference embedding and additively injects it into shared prompts, enabling parameter-efficient user-level adaptation. A semantic drift constraint regularizes the shared prompts within a trust region around their initialization anchors to preserve semantic validity during optimization. By constraining the shared component while allowing additive personalization, LRPRec decouples stability from expressiveness. Extensive experiments on three benchmark datasets demonstrate consistent improvements over strong baselines while eliminating the need for manual tuning of background and task inference templates.

Flashblocks divide a block's priority gas auction into shorter sequential auctions that commit transaction order before the block is complete. We ask how this auction cadence affects bidding and competition among automated arbitrageurs, or searchers. In July 2025 Base replaced a single 2 s auction with ten 200 ms auctions. We use this change to estimate the searcher response from on-chain data alone. On an address-day panel of 3,032 searchers and 8,053 activity-matched controls, a difference-in-differences design estimates a 0.187 gwei fall in the effective priority fee (59% of the searcher pre-period mean). The share of searcher priority-fee value paid in the first tenth of block gas falls from 0.98 to 0.28. This fee compression is the identified effect. The revert-rate response is not (causally) identified and is reported descriptively. At the level of a single opportunity, the winning fee splits into an approximately invariant floor and a competitive premium. The premium scales about linearly with auction duration at high contention and less steeply at low contention. A first-price auction model with uncertain arrival and partial payments by losing searchers accounts for this pattern. We match attempts to the pool they contest. A higher-fee transaction reverts behind a lower-fee winner that landed earlier in 1.9 to 2.7% of contested opportunities by value, an upper bound. Latency therefore decides part of the ordering. A micro-auction model projects diminishing returns from shortening the window below 200 ms, as bids approach a nonzero floor. Rollups such as Arbitrum One now make auction cadence a governance tunable parameter, so these estimates inform how it is set.

Recent experiments in monolayer and trilayer graphene have observed incompressible even-denominator fractional quantum Hall states within narrow regions of parameter space associated with crossings of zeroth Landau levels. This is surprising because at half filling the Coulomb interaction in the zeroth Landau level is expected to favour a compressible composite Fermi liquid. Here we provide the first quantitative energetic study of the candidate incompressible phases and of the microscopic mechanisms that can stabilise them near such Landau-level crossings. Using exact diagonalization and infinite density-matrix renormalisation group, we first study ultra-short-range corrections to the Coulomb interaction arising from finite layer separation, lattice-scale effects, and electron-phonon coupling. We find that a sufficiently strong attractive short-range interaction drives the composite Fermi liquid first into a weakly paired \(d\)-wave pseudospin-singlet state and subsequently into a strongly paired state. However, microscopic estimates of these short-range corrections are far too small to stabilise either incompressible phase. We then incorporate Landau-level mixing through screening within the random-phase approximation. Screening substantially reshapes the effective interaction, suppresses the weak-pairing regime, and strongly enhances the tendency toward strong pairing. For a realistic range of microscopic parameters, the strongly paired phase is stabilised over a substantial fraction of that range. Our results identify the interplay of Landau-level mixing and ultra-short-range interactions as a plausible microscopic mechanism for the experimentally observed incompressible states in monolayer and trilayer graphene and point toward a strongly paired composite-fermion phase as their candidate ground state.

Non-Clifford operations required for universal quantum computation can be implemented by measuring logical Clifford operators. We present a generalized framework to fault-tolerantly measure transversal Clifford logical operators in quantum low-density parity check (QLDPC) codes. Our protocol takes any QLDPC code and deforms the code across additional ancillary qubits such that a product of stabilizers measures the desired logical Clifford operator. We prove that the deformed, Clifford-stabilized code remains LDPC and that its distance is preserved up to a constant factor independent of the number of physical qubits. We further prove that the Clifford measurement has a fault distance against data qubit and measurement errors that grows linearly with the distance of the original code. We apply our methods to implement logical non-Clifford gates in algorithmically relevant subroutines and to prepare magic states in quantum stabilizer codes.

We provide a modern overview of several forcing methods in bounded arithmetic and compare them with forcing in set theory. We specifically discuss the countable forcing framework of Atserias and Muller, the restricted ultrapowers of Bydzovsky and Muller, and forcing with random variables developed by Krajicek. We illustrate these forcing arguments by reviewing the Paris–Wilkie forcing and by giving a self-contained construction of the shallow PHP model constructed by Krajicek. We observe that, from a formal point of view, Krajicek's forcing with uncountable structures can be interpreted in the framework of Atserias and Muller after a countable collapse using an LS argument. We further observe that the countable restricted ultrapowers of Bydzovsky and Muller, used to obtain models of the theory $T_{PV}$, can also be interpreted in this framework. This part serves as a primer on modern forcing methods in bounded arithmetic. In the next part, we discuss the analogies and differences between forcing in arithmetic and set theory. We embed Krajicek's Boolean-valued models into generic extensions of models of set theory, observing that the set-theoretic forcing framework can be used to interpret forcing in arithmetics. In our interpretation, set theory offers a convenient method for formalizing forcing arguments with uncountable structures. We end with a question whether analogies with set theory can facilitate the discovery of new uncountable partial orders for arithmetic.

We prove the Aleksandrov-Bakelman-Pucci maximum principle for $L^n$-viscosity sub/super-solutions of fully nonlinear uniformly elliptic equations in $\Omega\subset\mathbb R^n$ with the drift coefficient $\gamma$ such that $\gamma(\ln(1+\gamma))^{\frac{n-1}{n}}\in L^n(\Omega)$. The approach is based on convex duality and Trudinger-Moser type inequalities.

Motility-induced phase separation enables swarm aggregation to be regulated without prescribing a target spatial distribution. This paper develops a partial differential equation-constrained model predictive control framework for a vibration-driven robotic swarm. Three continuum models retaining progressively less orientation information are considered to study the trade-offs among model fidelity, information requirements, and computational cost. The framework is evaluated under centralized and decentralized settings, with focus on how limited sensing and local-to-global phase-separation estimation affect the performance of predictive control.

Partially observed multivariate time series pose two coupled challenges: recovering missing measurements and forecasting future trajectories when observation patterns are informative. Existing methods often assume ignorable missingness or use the mask only as an auxiliary input. We propose LIFTS (Latent-driven Imputation and Forecasting for partly observed Time Series), a probabilistic framework in which a latent process jointly governs the complete measurement trajectory and observation mask, allowing observation patterns to inform latent-state inference for both forecasting and imputation. LIFTS combines flexible neural-network parameterizations with a forward-backward architecture: the forward pass performs filtering and autoregressive forecasting, while the backward pass performs smoothing and conditional multiple imputation. The model and inference components are trained jointly using a masked energy score and self-masking. We establish forward-backward distributional representations for the general framework and, for an explicit structural subclass, provide full-law identification and consistency guarantees. Simulations and an application to PhysioNet 2019 show that LIFTS provides accurate point predictions, improved distributional accuracy, and well-calibrated predictive intervals; it outperforms competing methods particularly when training data are limited and has substantially faster inference than the diffusion-based baseline.