Papers

Long-context LLM training on long-tailed corpora faces a central communication–balance tradeoff. Such sequence-length heterogeneity makes any single sequence-parallelism (SP) degree a poor fit for the workload: a small degree leaves the few long sequences badly imbalanced, while a large degree forces the many short sequences that dominate the workload to pay excessive communication. Existing dynamic-SP systems mix SP degrees within a batch, but to run several groups at once they partition the GPUs into disjoint groups, which reintroduces imbalance across groups and forces costly micro-batch workarounds. We present NSP, a sequence-parallel training system that resolves this tradeoff by nesting differently sized SP groups on shared GPUs within a single training iteration. This lets long sequences use larger SP groups while keeping short sequences on smaller ones, so communication is incurred only where needed and load is balanced per GPU rather than per group. NSP realizes this idea with a tree-structured routing planner that assigns sequences under memory constraints and an executor that exploits the resulting hierarchy through inter-level phase streaming and tree-level recomputation. NSP supports common SP backends and requires no model changes. We evaluate NSP on Qwen3-MoE workloads with up to 384K-token contexts across multiple long-tail datasets on an internal production GPU cluster. Across these settings, NSP consistently improves end-to-end training throughput, outperforming Static SP by up to 1.48x and FlexSP by up to 1.16x.

Let $q$ be a prime power. We give a construction of Butson Hadamard matrices $BH((F_q\times F_q,+),lcm(6,d))$ for any divisor $d>1$ of $q-1$. Most character values of the group ring elements corresponding to these matrices involve products of two Gauss sums over $F_q$ (with non-quadratic Gauss sums occurring) and this property distinguishes our result from all previously known constructions of group invariant Butson Hadamard matrices. In fact, our $BH((F_q\times F_q,+),lcm(6,d))$ matrices arise as a special case of a more general construction based on what we call “Butson-Jacobsthal functions” that resembles the construction of Hadamard matrices from Jacobsthal matrices.

Natural-language service requests can require a language-model decision before execution starts, consuming part of the request's latency budget. We integrate Jev's decision-oriented application programming interface (API) into edge service orchestration to reduce this overhead while retaining service completion. The integration extracts four to eight bounded intent fields and applies a shared validator, admission policy, and scheduler, accounting for decision waiting throughout the request timeline. We compare Jev, two self-hosted decision models, and three hosted large language models (LLMs) on 8,280 verified requests and on a live admission path with modeled execution and a real optical character recognition service. Across 33 test conditions, Jev reduces median decision latency by 22.7-64.5% relative to the fastest LLM. This latency barely moves with input size, contract width, or catalog size. On four-field contracts, Jev's API fees per correct decision are 59.7-80.9% lower at a cost of a few exact-match points, while wide contracts mark the limit of the substitution. Receiving the service catalog with each request, Jev names unseen services as accurately as known ones. On the live admission path, Jev keeps 0.91-0.95 of requests exact and on time at loads where the LLMs fall below 0.1. Since caching repeated descriptions gives the interpreters nearly the same latency, Jev's gain lies in fresh decisions. These results support decision-model substitution for latency-bound admission on bounded contracts.

Diffusion models have demonstrated strong power in generative modeling tasks across multiple domains, exhibiting a remarkable capability of learning complex distributions from samples. In this paper, we leverage such capability to design an efficient universal diffusion-based solver for parameterized black-box optimizations (BBO), where the optimizer has only black-box access to queries of the objective function at the learning stage, yet is able to reduce the additional computational cost at the inference stage for each BBO instance while also capturing the potential multi-modal landscape of non-convex objectives. To cast our formulation as a compatible generative modeling task, we introduce the notion of minimization policy as a new solution concept, which defines a sampling distribution over the solutions that should concentrate around the minimizer set for each BBO instance. We then propose Diffusion-based Iterative Minimization Policy Learning (D-IMPL), a practical generative-model-based solver for solving parameterized BBOs that employs diffusion models to learn a minimization policy, whose density is proportional to the exponential of the negated objective values, thereby amortizing the computational costs across different BBO instances. Furthermore, we demonstrate the performance of our D-IMPL algorithm by establishing a sample complexity guarantee showing that a $\delta$-approximate minimization policy can be effectively learned within $O(\log(1/\delta))$ iterations, and by extensive empirical evaluations over a range of constrained and unconstrained BBO tasks.

Studies of dust growth in molecular clouds often prescribe the collapse history, leaving unclear how magnetic regulation of core formation shapes dust evolution and its observable and non-ideal MHD signatures. We address this problem by performing one-zone dust evolution and ionization calculations along time-dependent density and magnetic-field histories extracted from three-dimensional non-ideal MHD simulations of core formation through filament fragmentation. A stronger initial magnetic field delays contraction and thereby gives dust grains more time to grow at a given density. In our models, this delayed contraction leads to mass-weighted mean dust sizes of $1.2\,\mu\mathrm{m}$ and $3.1\,\mu\mathrm{m}$ at $n_{\mathrm{H}}=10^6\,\mathrm{cm^{-3}}$ for $B_{\mathrm{ini}}=20$ and $50\,\mu\mathrm{G}$, respectively. The associated depletion of very small dust grains reduces the adsorption of charged particles onto dust grain surfaces and lowers the conductivity: relative to models without dust growth, the ionization fraction increases from $\sim10^{-9}$ to $\sim10^{-8}$, the ambipolar resistivity increases, and the ion–neutral drift velocity rises to several $\mathrm{m\,s^{-1}}$ in the weak-magnetic-field model and approximately $10\,\mathrm{m\,s^{-1}}$ in the strong-magnetic-field model. These drift velocities remain below the observationally suggested range of $30$–$100\,\mathrm{m\,s^{-1}}$, indicating that still stronger magnetic fields may be required. The evolved dust populations also attain single-scattering albedos of order $\omega_\lambda\sim0.8$ at $3.6$–$4.5\,\mu\mathrm{m}$, with the stronger field shifting the onset of efficient infrared scattering toward lower densities. These results demonstrate that the magnetically controlled collapse timescale, rather than density alone, links dust growth to ionization, ambipolar diffusion, and infrared scattering in prestellar cores.

Manual attendance methods, such as paper or register-based systems, take a lot of time, can lead to errors, and are easy to falsify. Face recognition is more reliable, but it frequently struggles in classrooms because lighting and other conditions can vary. Face recognition datasets are designed for regulated environments and do not capture the actual challenges found in classrooms. To address this, a new face detection and recognition dataset, the Visage Face dataset, comprising 16,234 face samples, is proposed for the task of face detection and recognition. The photos are taken from different angles and under varying lighting conditions, with students showing a range of expressions, and some faces partly covered to reflect real-life situations. A YOLO-based system is used to detect faces and tested seven advanced face recognition models with thirteen configurations: LVFace, QCFace, FaceLiVTv2, TopoFR, EdgeFace, TransFace, and GhostFaceNets. Of these, FaceLiVTv2-M performed best, with 99.75% Top-1/Top-5 accuracy and an inference time of 6.459 ms. These results show that the Visage Face Dataset is a realistic and challenging benchmark for face recognition in classroom attendance.

This paper investigates the continuous dependence (stability) of solutions to the barotropic compressible Navier–Stokes equations in critical Besov spaces. Using the Lagrangian approach developed in \cite{Danchin2014}, it was shown that, for \(1<p<2d\), the flow map $(a_0,u_0)\mapsto (\bar a,\bar u)=(a\circ X,u\circ X)$ is Lipschitz continuous from $\dot B_{p,1}^{d/p}\times \dot B_{p,1}^{d/p-1}$ into $\mathcal{C}([0,T];\dot B_{p,1}^{d/p})\times E_p(T)$. However, this result does not directly imply the corresponding continuous dependence in the original Eulerian coordinates because of the low regularity of the critical initial data. Previously, only for $p<d$, continuous dependence was known only in certain lower-regularity spaces with a loss of one derivative relative to the natural solution spaces arising essentially as a by-product of the uniqueness argument. We close this gap and prove that for $1<p<2d$ the flow map $(a_0,u_0)\mapsto(a,u)$ is continuous (not Lipschitz continuous) from $\dot B_{p,1}^{d/p}\times\dot B_{p,1}^{d/p-1}$ to $\mathcal{C}([0,T];\dot B_{p,1}^{d/p})\times E_p(T)$ in Eulerian coordinates without any loss of regularity, which together with the known existence and uniqueness theory \cite{Danchin2014} completes Hadamard well-posedness in the critical spaces.

On a seven-compound drug-response model, warm-start quantum approximate optimization (QAOA) on IonQ Forte-1 returned valid assignments more often than random bitstrings, but this alone did not show effective optimization. Ideal QAOA raised optimum probability above uniform feasible sampling in only four of twelve reference circuits. Hardware often fell below its own noiseless circuits, while greedy search solved all hardware models within 200 objective evaluations. Expanded simulations showed a gain over feasible sampling in 28 of 35 CAMA-1 panels and none of four 647-V panels. Annealing solved all these panels in every seed. A Grover mixer preserved feasibility and improved optimum probability over feasible sampling in all ten tested models. We analyzed 25 completed tasks containing 5,300 shots from 18 circuits and 14 instances. The encodings use 6-35 qubits and at most 4,900 feasible assignments, which we enumerated to establish exact optima. Noiseless references now cover both the original six circuits and six wider circuits. At 35 qubits, with 4,900 feasible assignments, ideal feasibility was 35.85%, compared with 7 of 200 valid hardware outputs. Its ideal optimum probability was below both sampling controls. The circuits sample assignments in a model built from measured single-agent and pairwise responses.

Understanding the performance of large-scale recommender systems remains an underexplored challenge, especially for content creators and model developers. The raw engagement signals available to them, such as views and clicks, conflate content quality, model behavior, presentation bias, and audience reach, making it hard to attribute outcomes to the right cause. In this work, we present a general evaluation framework that enhances observability across multiple recommender systems at Netflix and demonstrate its effectiveness through several production deployments. The framework treats recommender-system observability as a counterfactual measurement problem: estimating what the recommender would have done, and what engagement would have followed, in the absence of a specific content item or model decision. We articulate three stakeholder-centered observability principles for content creators and model developers, and propose measurement methodologies covering bias reduction, relativity, and incrementality, applicable to both single-stage and cascading recommender systems and serving both audiences from a single measurement foundation.

Large Language Models (LLMs) have recently been introduced into traffic signal control (TSC) as decision agents due to their strengths in human-readable reasoning generation. Yet, existing LLM TSC methods optimize only from final outcomes and fail to distinguish valid from flawed reasoning steps, causing useful or misleading steps to be jointly updated and thus impairing the model's learning of effective reasoning. To bridge this gap, we propose an LLM-based framework ProcessLight to decompose signal decisions into verifiable semantic steps. Building on ProcessLight, we further develop Step-wise Traffic Process Policy Optimization (STeP-PO), a novel reinforcement learning framework that optimizes structured reasoning processes through step-level credit assignment. Specifically, STeP-PO uses step quality scores to evaluate local reasoning quality and step importance to measure each step's influence on the final action, and then assigns step-level advantages over a semantic step tree structure. The resulting step-level advantages are propagated to reasoning tokens, enabling fine-grained policy optimization beyond outcome-only rewards. Extensive experiments over multiple real-world datasets demonstrate the superiority of our methods. Our code is available at https://github.com/wenzhaoabc/processlight.

Structural nonlinearity enables nonlinear input–output mappings to emerge from otherwise linear wave dynamics. Repeated interactions with an input-encoded structure can enhance such mappings, but finite-depth implementations restrict the accessible functional order. Here we show that recurrent scattering in a resonant cavity provides a distinct regime for structural nonlinearity. Repeated interactions are coherently accumulated into a non-polynomial response to structural perturbations. The resulting mapping naturally takes a Kolmogorov–Arnold form: perturbation-induced resonance shifts realize the inner univariate mappings, while the resonant spectral response provides the outer mapping. We identify two complementary physical controls of representation capacity: the resonance linewidth governs the functional richness within each branch, whereas combining multiple branches expands the accessible function space. Microwave-cavity measurements validate the two-stage mapping and demonstrate a two-branch nonlinear computation through XOR classification. Our results connect recurrent resonant scattering with controllable non-polynomial computation in linear wave systems.

For a finite abelian group $G$, let $\mu_2(G)$ be the minimum number of positive eigenvalues of an entrywise positive real symmetric group-circulant matrix whose entrywise square is positive definite. We prove that $\mu_2(G)\le5$ precisely when $|G|\le15$, and that $\mu_2(G)=6$ precisely when $|G|=16$ or $G$ is cyclic of order $18$, $20$ or $22$. For the noncyclic groups of order at most fifteen, the minimum is three for $C_2^2$ and five otherwise. If the positive Fourier support of such a matrix is $\{0,\pm r_1,\ldots,\pm r_d\}$, where $d\ge1$ and all $2d+1$ displayed elements are distinct, then there are integers $\alpha_1,\ldots,\alpha_d$, not all zero, such that $\sum_i\alpha_i r_i=0$ and $\sum_i|\alpha_i|\le\max\{5,d+2\}$. The bound five is sharp for $d=2$ and $d=3$.

Weak-memory processors rely on ordering instructions for correctness, yet conventional implementations often enforce them more conservatively than the memory model requires. This over-enforcement manifests as drain-induced retirement stalls at ordering instructions and conservative squash/replay of speculative loads, suppressing legal executions and reducing throughput. We present TEMPO, a tag-based framework for precise micro-architectural implementation of ordering instructions. TEMPO assigns lightweight ordering tags to instructions and decomposes enforcement across retirement-time predicates and completion-time store ordering, allowing the core to enforce required ordering without conservative retirement serialization. TEMPO eliminates unnecessary retirement serialization at ordering instructions and speculative-load squash/replay. In our evaluation, TEMPO reduces geometric-mean normalized execution cycles by 7.9% on native four-thread workloads and improves geometric-mean IPC by 15.9% on an instrumented SPEC2017 dynamic binary translation (DBT) proxy for cross- ISA execution (e.g., x86-on-Arm), while adding only 262 bytes per core.

Recovering missing regions in 2-D LiDAR scans is crucial for maintaining geometric awareness in environmental sensing. LiDAR-only reconstruction relies on measurements surrounding the missing region and can degrade as the unobserved sector grows or the surface geometry becomes more complex. This paper presents a distributed integrated sensing and communication (ISAC) framework that uses directional 60-GHz beam-training measurements from multiple receivers to assist in the recovery of missing regions in 2-D LiDAR scans. Transmitter and receiver positions, beam directions, and relative beam-power thresholding are used to construct an RF-derived surface prior without requiring RF time-of-flight measurements. The resulting RF-derived range estimates are then fused with LiDAR measurements to estimate the missing geometry. The proposed framework is evaluated in an indoor environment using LiDAR and 60-GHz phased-array measurements from 55 receiver viewpoints comprising 37,620 beam-pair power measurements across 72 controlled missing-sector cases. The results show that RF-assisted reconstruction provides larger accuracy gains as the missing-sector width increases. For the 80-degree missing-sector case investigated in this paper, RF+polar fusion is shown to reduce the mean absolute error (MAE) from 2.271 m for polar LiDAR interpolation to 0.617 m, corresponding to a 72.8% reduction. The results further show that useful RF-assisted recovery can be maintained even with fewer beam-pair measurements under limited beam training. These findings demonstrate that directional communication measurements can support environmental sensing by providing complementary geometric information for missing LiDAR regions.

The cube polynomial $C_G(x)$ generates the number of $k$-cubes on a graph $G$. As a subclass of partial cubes, the tope graphs of lopsided sets (LOPs) generalize daisy cubes and median graphs. In this paper, we prove that every tope graph of a LOP shares its cube polynomial with some daisy cube, thereby answering affirmatively a problem posed earlier by the authors. Furthermore, we present explicit expressions for the cube polynomials of tope graphs of LOPs: $C_G(x)=f_\mathcal{K}(x+1)$, where $f_\mathcal{K}(x)$ is the $f$-polynomial of the cubical complex $\mathcal{K}$ of $G$. Finally, we show that a class $\mathcal{G}$ of partial cubes is the class of tope graphs of LOPs if and only if the following equivalent conditions hold: (a) $\mathcal{G}$ is the maximal pc-minor-closed class such that every graph in $\mathcal{G}$ has the same cube polynomial as some daisy cube; (b) for every $G\in\mathcal{G}$, each antipodal subgraph of $G$ has the same cube polynomial as some daisy cube.

We prove that Sierpiński carpet parameters are dense in the bifurcation loci of some natural families of finite type transcendental entire functions having exactly one active critical orbit. Moreover, some specific transcendental entire functions with Sierpiński carpet Julia sets are constructed such that they contain either an attracting basin, a parabolic basin, a Siegel disk or wandering domains. The proofs are mainly based on applying the polynomial-like renormalization and some results on the local connectivity of transcendental Julia sets obtained before.

We study a five-dimensional realization of endothermic dark matter motivated by the recent high-recoil LUX-ZEPLIN event. A pseudo-Dirac Standard-Model singlet fermion is localized in the extra dimension and couples to a bulk $U(1)_D$ gauge field. The same localization that suppresses a distant-brane Majorana operator and generates the inelastic splitting also controls the couplings to the Kaluza–Klein (KK) dark-photon tower, producing a localization-filtered mediator propagator. We organize the phenomenology around two representative mass–splitting points: an LZ benchmark, $m_\chi=1.10~\TeV$ and $\delta=365~\keV$, and an IceCube-motivated comparison point, $m_\chi=1.08~\TeV$ and $\delta=566~\keV$. The latter is used only to compare terrestrial and solar kinematics; the published IceCube bound is Higgsino-specific and is not directly imposed on our singlet KK model. At the LZ benchmark with $R^{-1}=0.30~\GeV$ and $m_0=0.10~\GeV$, the KK tower enhances the high-recoil xenon rate by a factor of about four relative to a single dark photon at fixed microscopic coupling. The LZ point lies just below the terrestrial halo speed ceiling, whereas the $566~\keV$ comparison point is inaccessible to ordinary xenon scattering and instead lies in a kinematic regime accessible to solar capture; any neutrino constraint on the present singlet model remains dependent on its capture and annihilation dynamics. The characteristic prediction of the extra-dimensional construction is therefore a correlated interplay of localization, KK spectral distortion, and inelastic kinematics.

We study the diffusion limit of the classical solution to the relativistic Boltzmann equation with the initial data near a global relativistic Maxwellian. By using spectral analysis, we establish the convergence of the classical solution to the relativistic Boltzmann equation to that of the incompressible Navier-Stokes system, and give for the first time the optimal convergence rate and a precise estimate of the initial layer. Moreover, through a refined analysis of the linear collision operator $L$, combined with Duhamel's principle, we obtain the existence and uniqueness of a global strong solution to the relativistic Boltzmann equation, as well as a convergence rate independent of the speed of light $\mathbf{c}.$

Let $U$ be an affine log Calabi–Yau variety. Although special and finitely generated valuations are defined using a log CY-Fano compactification of $U$, we prove that the corresponding skeleta in the dual complex are independent of this choice, and hence invariant under $\mathrm{Aut}(U)$. We also show that a finitely generated valuation is special if and only if it is maximal with respect to a natural partial order induced by regular functions on $U$. We then consider the affine log Calabi–Yau threefold obtained as the complement of the Markov cubic surface in $\mathbb{A}^3$. We describe the action of the three Vieta involutions on the special skeleton and relate it to the $(\infty,\infty,\infty)$-triangle reflection group on the hyperbolic plane. We also show that, for every finite triangulation of the dual complex, the special skeleton fails to be locally closed on some simplex. Via the cone construction, this provides a counterexample to a conjecture of the author and Xu.

This paper establishes the global existence of classical solutions to the initial-boundary value problem for the planar four-velocity Broadwell model in a multidimensional setting. While uniqueness was already established in our previous work (which addressed existence under the assumption of sufficiently small, bounded, and $C^{1}$-regular initial and boundary data), the objective of the present study is to completely lift this size restriction. By extending the previously developed fixed-point framework, we prove that the methodology remains highly effective for arbitrary initial and boundary data. Within this generalized setting, both the solutions and their first-order partial derivatives are shown to be well-defined and to exist globally for all times.

Process data from computer-based assessments record the sequence and timing of actions through which respondents solve a task, providing information about both the pace and structure of problem-solving behavior. Modeling such processes across countries is challenging because country-by-response-group cells are often small and unbalanced and the observed transition supports can differ substantially across countries. We propose a hierarchical framework that integrates a Bayesian multi-state survival model with a network-based representation of transition structure. Partial pooling across countries yields country-specific covariate and key-action effects, transition speed, and estimates of between-country heterogeneity. Posterior transition probability networks are embedded in a common latent space using a directed graph auto-encoder adapted to heterogeneous supports, and 1-Wasserstein distances between node-role distributions are evaluated across posterior draws to characterize global network structure while propagating estimation uncertainty. We apply the framework to two problem-solving items from the Programme for the International Assessment of Adult Competencies across 14 countries. The results reveal cross-country heterogeneity in transition speed, systematic response-group differences in global network organization, item-dependent variation in within-group dispersion across countries, and local differences in routing around shared intermediate actions.

Clinical domain classification plays an important role in organizing and analyzing large volumes of unstructured medical text. However, medical transcription datasets are often highly imbalanced, which can substantially degrade classification performance, particularly for underrepresented clinical specialties. In this work, we present a comparative study of machine learning and transformer-based approaches for clinical domain classification from medical transcriptions. We evaluate six traditional machine learning classifiers—Naive Bayes, Support Vector Machine (SVM), Decision Tree, Random Forest, K-Nearest Neighbors (KNN), and XGBoost—along with two pretrained transformer models, BERT and XLNet, and a few-shot large language model prompting approach. Experiments are conducted on medical transcription data collected from MTSamples, comprising 5,013 samples across 40 clinical specialties. To address severe class imbalance, we investigate two balancing strategies: text augmentation using NLP-based synonym replacement and Synthetic Minority Over-sampling Technique (SMOTE). Experimental results demonstrate that data balancing substantially improves classification performance across the evaluated models. In particular, BERT achieves the highest F1-score of 0.996 on the SMOTE-balanced dataset while requiring lower training time than XLNet. The results highlight the effectiveness of transformer-based representations combined with appropriate data balancing strategies for clinical domain classification and provide a systematic comparison of classical machine learning, transformer models, and few-shot prompting for medical transcription analysis.

This paper develops the Rapidly-iterating kinoDynamic Grid (RDG) algorithm, an asymptotically near-optimal kinodynamic motion planning algorithm that produces high quality solutions through rapid iteration. The algorithm leverages a state space grid decomposition to perform node selection, dynamics propagation, and graph revision in constant time complexity with respect to the number of nodes in the trajectory tree. Through a covering ball sequence induction proof, the algorithm is shown to be asymptotically near-optimal and probabilistically complete. Different subsystems of the algorithm are evaluated against common nearest-neighbor search-based methods at generating exploration bias. The RDG algorithm is evaluated through simulated trials in complex, kinodynamic motion planning problem environments up to 10 DOF relative to similar sparse, kinodynamic planning algorithms with optimality guarantees, SST and DIRT. The RDG algorithm outperforms both SST and DIRT in mean final solution quality by up to 104% and 40% respectively. Additionally, RDG maintained a 100% success rate, even on a 10-DOF test case where both SST and DIRT did not.

We introduce and study a new model of decision trees that lies at the frontier of provable circuit lower bounds. We study $\ell$-local decision trees, in which each internal node queries an $\ell$-local function of the input. We show that sufficiently strong lower bounds for $\ell$-local decision trees would imply several breakthrough circuit lower bounds, including super-linear size lower bounds for log-depth circuits and improved bounds for unrestricted-depth circuits. Previously, such consequences were known to follow from strong lower bounds for depth-$3$ circuits, which has been the main prior approach in attempting to prove these results. Since local decision trees are strictly weaker than depth-$3$ circuits, this provides a formally easier route to the same circuit lower-bound consequences. We prove essentially optimal lower bounds for a weaker variant that we call oblivious $\ell$-local decision trees, where all nodes at the same depth query the same function. Our lower bounds follow from a new technique that uncovers sumset structure in local maps, and our hard functions are sumset dispersers, sumset condensers, and directional affine dispersers. Along the way, we give an explicit construction of a sumset condenser with small entropy loss. As an additional contribution, we initiate the study of degree-$2$ decision trees, in which each query is a quadratic polynomial of the input. We show that proving lower bounds in this model is a natural stepping stone toward constructing dispersers for degree-$2$ variety sources, which imply improved circuit lower bounds for unrestricted depth circuits. We prove nearly maximal lower bounds for the oblivious variant of the model.

In this paper, we propose a high-precision yet low-complexity framework, termed Neural Pursuit for Line Spectral Estimation (NeuPLSE), for off-grid line spectral estimation from a single snapshot. The proposed framework systematically integrates model-driven signal structures with data-driven adaptability, enabling continuous-domain spectral representation together with adaptive parameter learning. Within this framework, model order detection and spectral parameter estimation are jointly achieved through residual analysis, with coordinated use of global and generalized component-wise residual information to ensure robust and accurate estimation performance. Moreover, NeuPLSE supports a unified extension from one-dimensional (1D) to multi-dimensional (MD) scenarios while maintaining low computational complexity. Extensive simulations in representative integrated sensing and communication (ISAC) scenarios demonstrate that NeuPLSE achieves accurate model order detection and high-precision spectral estimation. In sparse-path settings, it approaches the Cramér-Rao lower bound, while in dense-path settings, it attains competitive reconstruction performance with significantly lower computational complexity than state-of-the-art methods.

Robot manipulation tasks often involve hidden state information that cannot be directly observed and must be inferred through sequential physical interactions. In such partially observable settings, conditioning an imitation learning policy directly on the recent raw observation history leads to poor performance. This is due to state aliasing, wherein identical observations may arise from different hidden states, and the policy receives conflicting action labels for the same input. To enable history-aware disambiguation capability, we propose conditioning a diffusion policy on a structured representation of the hidden states using Bayesian belief that exposes both the current most likely state estimate and the remaining uncertainty. This representation replaces raw history with a structured, compact input, enabling the policy to implicitly modulate between exploratory and exploitative behaviors based on belief uncertainty, without explicit mode switching or reward shaping. We evaluate across two domains with qualitatively different belief representations: a continuous belief for cornstalk gripper alignment via tactile sensing, and a discrete categorical distribution for latched door opening. In both domains, the belief-conditioned policy substantially outperforms the observation-only baseline and approaches privileged ground-truth performance, with ablations illustrating that the policy adapts its exploration behavior depending on the belief uncertainty at inference time.

Recent advances in the variational quantum eigensolver (VQE) have strengthened its promise for molecular electronic-structure calculations on near-term noisy intermediate-scale quantum (NISQ) devices. However, barren plateaus (BPs) remain a major obstacle to scaling VQE, as vanishing gradients can severely obstruct optimization at large scale. While most BP studies focus on randomly initialized circuits and global gradient behavior, practical quantum chemistry VQE typically uses chemically motivated initializations and local optimization around HF- or MP2-derived states. Although chemically informed initializations are widely expected to improve VQE trainability, how they mitigate local barren plateaus remains unclear. We therefore study the occurrence of local barren plateaus in molecular UCCSD-VQE, considering both chemically motivated initialization points and iterates along the corresponding optimization trajectories. In the benchmarks considered here, chemically initialized starting points and variational trajectories show polynomial decay of local gradient variance with system size instead of the exponential BP-like decay observed for random controls, indicating that chemical initialization mitigates barren plateaus in the local optimization regions and highlights the trainability advantage of chemically motivated VQE strategies.

Residual-error preservation (REP) provides a sufficient rule for constructing concatenated composite pulses that compensate multiple systematic errors. We show that local REP is not necessary: deviations from REP can cancel across the full sequence. Using first-order error generators, we derive a necessary and sufficient condition for concatenation to retain the robustness of an outer sequence. A simple sufficient condition is that all inner pulses rescale the corresponding elementary error generators by a common real factor, not necessarily unity. We illustrate this principle using pulse-length and off-resonance errors, two systematic error models commonly considered in magnetic resonance. Short CORPSE and SCROFULOUS each compensate one error without preserving the other locally. Combining these inner pulses with suitable equal-rotation-angle outer sequences yields simultaneous first-order compensation of both errors, extending concatenated composite-pulse design beyond local REP.

Ideality factor of a diode will depend on the location of the plane of maximum recombination within the diode's depletion region. It will have a minimum of one, but can range up to infinity depending on placement of the plane. Case studies analyzed in this work demonstrate better agreement with experimental data from GaN-based p-i-n diodes.

Efficient coordination under limited communication remains a key challenge in decentralized multi-robot exploration. While centralized approaches benefit from global information sharing, they are often impractical in large-scale or communication-constrained environments. Existing Monte Carlo Tree Search (MCTS)-based approaches, such as Decentralized Monte Carlo Exploration (DMCE), enable decentralized planning by taking peer intent into account. This peer intent is obtained by communicating sequences of planned waypoints with robots within direct communication range. In this work, we extend this idea by introducing Probabilistic Peer Intent (PPI), which converts peer trajectories into a continuous spatial representation of predicted intent and incorporates it into local MCTS action evaluation. We additionally study the effects of sharing peer intent beyond direct communication range by propagating plans over multiple hops. Experiments across multiple simulated environments and team sizes show that PPI and Multi-hop propagation can each improve decentralized exploration, with their relative benefits depending on environment structure and team size. We also demonstrate the real-world deployment of our method on three robots operating in different environment types.