We present PHLEGETHON, a fully compressible, Eulerian magnetohydrodynamic (MHD) code designed for multidimensional simulations in stellar astrophysics. The code uses a time-explicit, second-order, finite-volume method optimized to model a wide range of dynamical processes in stars, from very low-Mach-number turbulent convection in the cores of massive stars to supersonic flows in subsurface convection zones. PHLEGETHON employs low-dissipation Riemann solvers and a well-balanced method to accurately capture slow flows arising from strongly stratified media. The induction equation is solved using a staggered constrained-transport method to ensure divergence-free evolution of the magnetic field. The MHD equations are coupled to arbitrary nuclear reaction networks solved in a time-implicit approach, together with super-time-stepping for efficient treatment of thermal diffusion. Equations of state appropriate for stellar plasmas are available, accounting for partial ionization, electron degeneracy, and electron-positron pair production. The code is implemented in a compact and user-friendly manner, and it scales to tens of thousands of CPU cores using MPI-based domain decomposition. We perform several verification tests to demonstrate the accuracy and versatility of the code, and present simulations of magnetoconvection in a core-collapse supernova progenitor star. The rich variety of physical effects and numerical methods implemented in PHLEGETHON enables the code to model diverse multidimensional processes that play a crucial role in stellar-interior dynamics, such as reactive convection, convective boundary mixing, internal-wave excitation, and magnetic-field amplification mechanisms. Within a single framework, these phenomena can be investigated across a wide range of stellar evolutionary stages, from main-sequence stars to supernova progenitors. PHLEGETHON is publicly accessible online.

Objective: This study aimed to assess how wearable physiological signals, alone and combined with salivary cortisol, distinguish physical and psychological stress and their recovery states. Methods: Six healthy adults completed three laboratory sessions on separate days: rest, physical stress (high-intensity cycling), or psychological stress (modified Trier Social Stress Test). Heart rate, heart rate variability, electrodermal activity, and wrist accelerometry were recorded continuously, and salivary cortisol was sampled at five time points. Features were extracted in non-overlapping 10-minute windows and labelled as rest, physical stress, physical recovery, psychological stress, or psychological recovery. A gradient boosting classifier was trained using wearable features alone and with five additional cortisol features per window. Performance was evaluated using leave-one-participant-out cross-validation. Results: Wearable-only classification achieved 77.8% overall accuracy, with high accuracy for physical stress and recovery but frequent misclassification of psychological stress and recovery (recall 50.0% and 54.2%). Including cortisol improved overall accuracy (94.4%), particularly for psychological states, increasing recall to 83.3% and 87.5%. Cortisol also reduced misclassification between psychological stress and rest. Conclusion: Wearable signals alone were insufficient to reliably distinguish psychological stress from rest and recovery. Integrating salivary cortisol improved classification of psychological stress and recovery and reduced confusion with rest, highlighting the value of endocrine context alongside wearable physiology. Significance: These findings support multimodal stress monitoring and motivate larger, ecologically valid studies and scalable alternatives to repeated cortisol sampling.

The response of Ni-Cr alloys to exposure to molten chloride and fluoride salts is typically characterized by Cr dealloying with the formation of a Cr-depleted bi-continuous porous subsurface layer. The exact mechanism behind the loss of Cr over distances unattainable by lattice diffusion alone is still debated. To address this question, two different surface finishes, namely electropolished and sanded, of a Ni-30Cr alloy were exposed to LiCl-KCl-2wt% EuCl3 eutectic salt at 500 °C for 96 hours. In the absence of fast diffusion pathways, dissolution occurred layer by layer and was kinetically controlled by Ni dissolution, as observed over the grain interiors of the electropolished sample. Grain boundaries were subject to diffusion-induced grain boundary migration (DIGM), leading to the formation of pure Ni islands above grain boundaries. This overall behavior contrasted with the sanded surface response that was characterized by several micrometer deep interconnected porosity and complete Cr depletion. DIGM of the dense grain boundaries created by recrystallization of the sanded surface was responsible for the observed sub-surface microstructure. This work unequivocally establishes DIGM as a key mechanism in alloy molten salt corrosion, and microstructure as a decisive contributor to an alloy's corrosion response.

We investigate the optical response of a conduction electron in a helically twisted mesoscopic medium containing a screw dislocation and a uniform torsional background, in the presence of an axial magnetic field and an Aharonov--Bohm flux. We show that the coupling between longitudinal motion and the geometric background produces an effective in-plane confinement, allowing bound states to emerge without the need for an external radial potential. Exact analytical solutions are obtained for the energy spectrum and radial wave functions, and these results are used to evaluate linear and third-order nonlinear absorption, changes in the refractive index, the photoionization cross section, and the oscillator strength. The combined action of torsion, magnetic field, and topological defect increases the interlevel spacing, compresses the radial electronic distribution, and breaks the dynamical symmetry between opposite angular-momentum channels, leading to strongly asymmetric and state-resolved optical spectra. Under intense optical excitation, the nonlinear contribution can overcome linear absorption, driving the system into a negative-absorption regime and enabling geometry-controlled optical gain. These results establish torsion and defect engineering as effective tools for tuning confinement, resonant energies, and selective amplification in mesoscopic nanophotonic platforms operating in the mid-infrared and terahertz ranges.

There is a growing recognition of the importance to involve patients in every stage of drug development. This shift acknowledges that patients' perspectives, experiences, and preferences are essential for ensuring that treatments meet real-world needs. In this context, a new body of statistical literature has emerged, focusing not only on the simultaneous consideration of multiple outcomes that reflect patients' overall experiences, but also on their structured prioritization. We refer to this class of approaches as hierarchical multi-component statistical methods. Among these, two influential frameworks - generalized pairwise comparisons (GPC) and desirability of outcome ranking (DOOR) - have emerged in the last decade, each aiming to offer a comprehensive approach to evaluating treatment effects. A new methodology, referred to here as the Markov ordinal state transition model (MOST), has recently been introduced without focusing on an explicit link with GPC nor DOOR. This paper seeks to fill this gap by offering a comprehensive and comparative analysis of the three approaches. Through examples and an exploration of the structural and philosophical differences between the methods, our aim is to provide guidance and encourage lines of research in the rapidly-evolving landscape of hierarchical multi-component statistical methodologies.

Quantum states of light are central resources for quantum communication, networking, and photonic information processing. In many quantum emitters, coherent internal dynamics arising from intrinsic or field-induced level splittings imprint a deterministic, time-dependent phase on the emitted light. When emission times are stochastic and detector timing resolution is finite, this phase evolution becomes effectively unresolved, suppressing observable entanglement. Here, we demonstrate a photonic-compensation protocol that removes this emitter-induced phase evolution directly in the photonic domain. Rather than modifying the emitter, we apply synchronized, time-dependent coherent operations to the emitted photons that reverse the accumulated phase independently of the emission time. Using exciton fine-structure splitting in a semiconductor quantum dot as a model system, we implement dynamic phase modulation and perform time-resolved two-photon polarization tomography. We show that this restores a stationary two-photon polarization state and recovers polarization entanglement without temporal post-selection and independently of detector timing resolution. Our approach provides a scalable route to robust solid-state entangled-photon sources and, more broadly, establishes a strategy for removing the imprint of coherent emitter dynamics on photonic entanglement in integrated platforms.

Active Inference is an emerging framework providing a quantitative account of behavioral processes in neuroscience and a principled approach to decision-making under uncertainty. Its application to agency problems is natural, offering an autopoietic interpretation of action while addressing classical challenges such as the exploration-exploitation trade-off. Recently, Active Inference has been applied to digital twin scenarios for adaptive and predictive modeling of complex systems. In this work, we extend Active Inference to multi-agent digital twins in which agents interact within a shared environment while maintaining decentralized generative models. Our multi-agent framework features two innovations: (i) contextual inference to improve adaptability in dynamic environments, and (ii) the integration of streaming machine learning within agents' generative structures, enabling tunable goal-oriented behavior while preserving efficiency and scalability. The framework is illustrated through a Cournot competition example, providing a digital twin representation of a socio-economic system and highlighting its potential for coordinated decision-making in multi-agent contexts.

In 1972, Fredman proposes the problem of sorting under partial information: preprocess a directed acyclic graph $G$ with vertex set $X$ so that you can sort $X$ in $O(\log e(G))$ time, where $e(G)$ is the number of sorted orders compatible with $G$. Cardinal, Fiorini, Joret, Jungers and Munro [STOC'10] show that you can preprocess $G$ in $O(n^{2.5})$ time and then sort $X$ in $O(\log e(G) + n)$ time and $O(\log e(G))$ comparisons. Recent work of van der Hoog and Rutschmann [FOCS'24] implies an algorithm with $O(n^ω)$ preprocessing time where $ω< 2.372$ and $O(\log e(G))$ sorting time. Haeupler, Hladík, Iacono, Rozhoň, Tarjan and Tětek [SODA'25] achieve an overall running time of $O(\log e(G) + m)$. In this paper, we achieve tight bounds for this problem: $O(m)$ preprocessing time and $O(\log e(G))$ sorting time. As a key ingredient, we design a new fast heap data structure that might be of independent theoretical interest.

We demonstrate that quantum light statistics can be used to control strong-field ionization at the tunneling step. Using a bichromatic linearly polarized field composed of a strong coherent driver and a weak bright squeezed vacuum (BSV), we show through simulation that photoelectron momentum distributions (PMDs) exhibit asymmetries that exceed those obtained with classical fields of comparable intensity by orders of magnitude. This enhancement is uniquely linked to the nonclassical statistics of the BSV field. A semiclassical analysis based on the strong-field approximation (SFA) reveals that the effect originates from fluctuations in the instantaneous field amplitude, which strongly modify the tunneling ionization probability while leaving the electron's continuum dynamics essentially unchanged. This selective control enables reconstruction of ionization pathways and provides a robust route to extract sub-cycle dynamics from strong-field observables.

Controlling the statistical properties of light, namely the fluctuations in photon arrival, entropy and number, is essential for both classical and quantum photonics. While integrated systems provide tunable control over amplitude, phase, and wavelength, real-time modulation of photon statistics has remained a long-standing challenge. Herein, we introduce the concept and experimental realization of a photon statistics transducer: a high-extinction, broadband electro-optic device capable of deterministically shaping photon-number distributions at nanosecond timescales. Our approach employs a cascaded thin-film lithium niobate (TFLN) Mach-Zehnder amplitude modulator delivering more than 50 dB extinction, enabling precise suppression and release of coherent seed light from an integrated InP laser. By exploiting the interplay between seed suppression and erbium-doped fiber amplifier dynamics, we demonstrate smooth, voltage-controlled switching between Poissonian and super-Poissonian photon statistics, with second-order coherence g2(0) tunable from 1.0 to 1.7. Complementary measurements with superconducting nanowire single-photon detectors further show photon-flux control down to sub-photon levels, highlighting the potential for future operation with non-classical sources. The photon statistics transducer thus establishes statistical modulation as a new functional primitive in integrated photonics. Applications range from entropy generation and secure communication to neuromorphic and hybrid quantum-classical processing, where controlled randomness and entropy are essential resources. By enabling programmable transitions between statistical regimes using only electronic drive signals, our work lays the foundation for adaptive, entropy-aware photonic systems that bridge classical and quantum domains.

Deploying applications across the computing continuum requires selecting infrastructure nodes from geographically distributed and heterogeneous environments while satisfying constraints (e.g., performance, location). This decision problem is an important facet of resource allocation. As infrastructures grow in scale and heterogeneity, the resulting decision space becomes inherently combinatorial. Existing approaches typically formulate this problem as a constrained optimization task using ad-hoc representations of infrastructure topologies and demand, which hinders generalization across solutions. In contrast, Software as a Service ecosystems address a structurally similar configuration problem through pricings -structures whose plans and add-ons implicitly define the configuration space of possible subscriptions. Building on this observation, this work explores the potential of pricings as general-purpose representations of configuration spaces, positioning them as a promising alternative for addressing configuration problems, such as resource allocation, across the computing continuum. To this end, the paper presents the following contributions: i) a pricing-based formulation of the resource allocation problem in the computing continuum, enabling infrastructure configuration spaces to be represented using pricings; ii) a workflow that leverages PRIME, a pricing analysis engine, to explore these spaces and compute cost-optimal deployments satisfying functional and non-functional constraints; iii) generation processes for synthetic infrastructure topologies and workload demands; and iv) a dataset comprising 9,600 precomputed resource allocation scenarios to support benchmarking.

Set $[n]=\{1, 2, \ldots , n\}$. The hypergrid $[t]^n$ is the collection of functions $f: \ [n]\rightarrow [t]$. We equip it with the natural partial order by letting $f\leq g$ whenever $f(x)\leq g(x)$ holds for all $x\in [n]$. Given a poset $P$ which can be embedded as an induced subposet of $[t]^n$, the induced poset saturation function $\mathrm{sat}^{\star}([t]^n, P)$ denotes the minimum size of a subset of $[t]^n$ that is both induced $P$-free and induced $P$-saturated. We show that for all $t\geq 2$, $\mathrm{sat}^{\star}([t]^n, P)$ satisfies a dichotomy: for every poset $P$, either there exists a constant $C_P$ such that $\mathrm{sat}^{\star}([t]^n, P)=C_P$ for all $n$ sufficiently large, or $\mathrm{sat}^{\star}([t]^n, P)=Ω(\sqrt{n})$. We also show chains fall in the former part of the dichotomy, while posets with the unique twin cover property fall in the latter part. These contributions generalize a number of results obtained by various authors in the hypercube ($t=2$) setting; the transition to the hypergrid setting provides novel challenges, however, and requires some new ideas.

Fairness is a critical requirement for human-related, high-stakes software systems, motivating extensive research on bias mitigation. Prior work has largely focused on tabular data settings using traditional Machine Learning (ML) methods. With the rapid rise of Large Language Models (LLMs), recent studies have begun to explore their use for bias mitigation in the same setting. However, it remains unclear whether LLM-based methods offer advantages over traditional ML methods, leaving software engineers without clear guidance for practical adoption. To address this gap, we present a large-scale study comparing state-of-the-art ML- and LLM-based bias mitigation methods. We find that ML-based methods consistently outperform LLM-based methods in both fairness and predictive performance, with even strong LLMs failing to surpass established ML baselines. To understand why prior LLM-based studies report favorable results, we analyze their evaluation settings and show that these gains are largely driven by artificially balanced test data rather than realistic imbalanced distributions. We further observe that existing LLM-based methods primarily rely on in-context learning and thus fail to leverage all available training data. Motivated by this, we explore supervised fine-tuning on the full training set and find that, while it achieves competitive results, its advantages over traditional ML methods remain limited. These findings suggest that LLMs are not a silver bullet for software fairness.

A path-integral hybrid Monte Carlo approach with enveloping bridging potentials (PIHMC-EBP) is proposed for calculating numerically exact tunneling splittings in molecular systems. The central idea is to construct an approximately barrierless bridging potential that smoothly connects symmetry-related regions of ring-polymer phase space, enabling direct sampling of the free-energy profile from which the relevant splittings are obtained. Two tailored nonlocal updates are designed to enhance the sampling of slow collective motions. Compared with path-integral molecular dynamics using thermodynamic integration, PIHMC-EBP requires neither quadrature nor time-step convergence checks, thereby substantially reducing the manual effort required to analyze the results. Applications to malonaldehyde (and its deuterated isotopologue) and the HCl dimer using state-of-the-art potential energy surfaces provide the most precise tunneling splittings reported to date for both systems, while simultaneously reducing the overall computational cost by several times and three orders of magnitude, respectively. Finally, application to the water dimer yields the first numerically exact path-integral calculations of the ground-state tunneling splittings on three different potential energy surfaces, all obtained simultaneously by reweighting a single set of trajectories.

We extend our previous symmetrized path-integral molecular dynamics approach to calculate tunneling splittings of molecules in rotationally excited states. In this new formalism, the system is rigorously projected onto selected rotational manifolds and states of a chosen symmetry through an Eckart spring, which connects the two end beads of the ring polymer via a permutation--inversion--rotation operation. This method is numerically exact within statistical uncertainty once convergence with respect to all simulation parameters has been achieved. Importantly, it enables the simultaneous extraction of tunneling splittings for multiple total angular-momentum quantum numbers $J$ from a single set of simulations, without additional computational cost relative to the original approach. After validating the formalism by computing the rotational levels of water (beyond the rigid-rotor approximation), we apply it to ammonia and obtain rotationally resolved tunneling splittings in excellent agreement with exact variational benchmarks. Except for small errors due to the underlying potential energy surface, the results capture the experimentally observed trend that the tunneling splitting decreases with $J$.

We study spectral properties of Schrödinger operators associated with substitution dynamical systems in higher dimensions. Focusing on periodic approximations generated by iterating substitutions on initial configurations, we analyze how structural defects influence the limiting spectral behavior. In contrast to the one-dimensional setting, we show that such approximations may exhibit significant spectral pollution, including changes in the essential spectrum and the Lebesgue measure.

An anomalous Hall effect (AHE) in antiferromagnetic (AF) systems with no net magnetization is of considerable interest for both fundamental physics and spintronic applications. Of particular interest is the two-dimensional van der Waals antiferromagnet FeTe that has an unusual fully magnetically compensated bicollinear AF structure and exhibits pronounced Kondo interaction leading to strong band renormalization. Here, we investigate the AHE in epitaxial FeTe thin films grown by molecular beam epitaxy. A large anomalous Hall conductivity is exhibited below the Neel temperature (T_N ~ 60 K) and, strikingly, becomes nonlinear at high fields within a narrow temperature window around 49 K, deviating from conventional AHE scaling behavior versus its longitudinal conductivity. Linear fits reveal a pronounced negative peak in the intercept, accompanied by a field-induced canted magnetic moment. The AHE responses are related to the Berry curvature derived from FeTe's topological band structure, highlighting the intricate interplay between topology, magnetism, and electronic transport.

Instantaneous Quantum Polynomial-time (IQP) circuits are a candidate for demonstrating near-term quantum advantage, as their sampling task is believed to be classically hard in the ideal theoretical setting under standard complexity-theoretic assumptions. In noisy implementations, however, this hardness can disappear once circuit depth exceeds a noise-dependent critical threshold. We show that qubit connectivity is a key parameter in this transition, since sparse architectures require additional routing to implement long-range interactions, thereby increasing compiled circuit depth. To make this explicit, we present a connectivity-aware analysis of compiled IQP circuits. For a fixed abstract IQP instance, different hardware connectivity graphs yield different compiled depths and thus different effective positions relative to the noisy-IQP simulatability boundary. We quantify this architecture-dependent shift using the compiled depth overhead and the corresponding simulatability margin. We combine analytic depth estimates for sparse geometries, including the two-dimensional grid, with native-gateset-aware compilation experiments across seven hardware-grounded experimental device models derived from publicly available topologies. To compare these device models under a unified empirical framework, we approximate the effective noise level primarily through reported two-qubit gate error rates. This lets us compare how much effective noise sparse and fully connected architectures can tolerate for the same position relative to the noisy-IQP simulatability boundary. Our results show that sparse connectivity requires a lower effective noise level to sustain the same margin relative to the noisy-IQP simulatability boundary, and they provide a quantitative framework for determining when compiled IQP experiments are likely to remain outside, or instead enter, the classically simulatable regime.

In the realm of computer-aided design (CAD) software, the intersection of B-spline surfaces stands as a fundamental operation. Despite the extensive history of surface intersection algorithms, the challenge of handling complex intersection topologies persists. While subdivision algorithms have demonstrated strong robustness in computing surface/surface intersection and are capable of addressing singular cases, determining the topology of the intersection obtained through these methods is a key factor for calculating correct intersection, and remains a difficult issue. To address this challenge, we propose a Mapper-based method for determining the topology of the intersection between two B-spline surfaces. Our algorithm is designed to efficiently handle various common and complex intersection topologies. Experimental results verify the robustness and topological correctness of this method.

We investigate the mutual information harvesting of two circularly accelerated detectors that interact with the massless scalar fields near a reflecting boundary. We consider that the two detectors share a common rotational axis with the same acceleration and trajectory radius. As the interdetector separation increases, the mutual information may exhibit oscillatory behavior at large acceleration and small radius. For a fixed radius, a larger acceleration leads to a larger peak value of the mutual information. Near the boundary, the mutual information may oscillate and the maximum can be obtained. As the acceleration increases, the mutual information in a small interdetector separation first increases and then decreases. For an intermediate interdetector separation, the mutual information may oscillate with the increase of acceleration. For a not large interdetector separation, when we take large acceleration and small radius, as the energy gap increases, the mutual information first decreases, then oscillates, and finally goes to zero. The combination of large acceleration and small radius corresponds to the fast rotation, which significantly modifies the vacuum fluctuations of the field, leading to the oscillatory behavior. Furthermore, the oscillation intensifies near the boundary, which indicates that it is related to the coherent superposition of boundary reflections.

Knowledge-intensive text usually contains fruitful entities and complex relationships, such as academic articles and scientific exposition. Reading and comprehending such texts often demands considerable time and mental effort to track the relationships between entities. To reduce the burden, we present GraphTide, a visualization technique that progressively constructs nested entity-relationship graphs with animation to support the understanding of complex text. Our method features an on-demand entity-relationship decomposition pipeline that constructs nested graphs to represent intra- and inter-sentence relationships. Moreover, we propose a structure-aware force-directed layout optimization algorithm to enhance structural clarity. Sentences and their associated entities are incrementally revealed through animated transitions, helping users maintain context as the narrative unfolds. A user study shows that GraphTide significantly improves users' comprehension of knowledge-intensive texts compared to traditional graph-based techniques and static nested graph representations.

Future sixth-generation (6G) networks require high spectral efficiency (SE), massive connectivity, and stringent reliability under imperfect channel state information at the transmitter. Rate-splitting multiple access (RSMA) addresses part of this challenge by flexibly managing interference through common and private message streams, while fluid antenna systems (FAS) offer low-cost spatial diversity by dynamically reconfiguring antenna positions within a compact aperture. In this paper, we first classify FAS-enabled multiple access systems from the perspectives of FAS deployment, objectives, and antenna configuration, along with some comparisons with benchmark schemes, thereby exhibiting the inherent efficiency of FAS-RSMA. Moreover, we reveal the mutually enhancing mechanism between FAS and RSMA: FAS strengthens the weakest effective link and improves the beamforming design in RSMA, whereas RSMA turns FAS-induced spatial diversity into robust interference management under diverse channel conditions. In addition, we identify representative 6G scenarios and highlight major research challenges in joint beamforming-antenna position design, channel estimation, and hardware design. Furthermore, case studies quantify the gains of FAS-RSMA over the fixed-position antenna (FPA) system with RSMA and NOMA baselines, which validates that FAS-RSMA is a strong candidate for interference-limited access in 6G systems.

In massive machine-type communication (mMTC) applications, a key challenge is joint device activity detection and channel estimation (JADCE) under grant-free random access, as a massive number of devices with sporadic traffic seek to connect to the base station. We address JADCE for massive random access using a covariance learning-based sparse Bayesian learning (SBL) approach. Specifically, we first use the successive convex approximation (SCA) framework to partially linearize the scaled negative log-likelihood function (LLF) of the data, then minimize it to estimate the sparse vector of devices' signal powers. After identifying active devices from these power estimates, empirical Bayesian estimation is used to obtain channel estimates. Simulation results demonstrate the efficiency and performance superiority of the proposed CL-SCA method compared to other existing methods.

We generalize the Abel--Hurwitz identities to an almost entirely noncommutative setting. Namely, let $V$ be a finite set of size $n$, and let $\mathbb{L}$ be any noncommutative ring. For each $s\in V$, let $x_{s}\in\mathbb{L}$. Set $x\left( S\right) :=\sum_{s\in S}x_{s}$ for any $S\subseteq V$. Let $X$ and $Y$ be two elements of $\mathbb{L}$ such that $X+Y$ lies in the center of $\mathbb{L}$. Then, we show that% \begin{align*} & \sum_{S\subseteq V}\left( X+x\left( S\right) \right) ^{\left\vert S\right\vert }\left( Y-x\left( S\right) \right) ^{n-\left\vert S\right\vert }=\sum_{\substack{i_{1},i_{2},\ldots,i_{k}\in V\text{ distinct}% }}\left( X+Y\right) ^{n-k}x_{i_{1}}x_{i_{2}}\cdots x_{i_{k}};\\ & \sum_{S\subseteq V}X\left( X+x\left( S\right) \right) ^{\left\vert S\right\vert -1}\left( Y-x\left( S\right) \right) ^{n-\left\vert S\right\vert }=\left( X+Y\right) ^{n};\\ & \sum_{S\subseteq V}X\left( X+x\left( S\right) \right) ^{\left\vert S\right\vert -1}\left( Y-x\left( S\right) \right) ^{n-\left\vert S\right\vert -1}\left( Y-x\left( V\right) \right) =\left( X+Y-x\left( V\right) \right) \left( X+Y\right) ^{n-1}. \end{align*} (Negative powers are understood to be cancelled by other factors.)

FPGAs are well-suited for dataflow architectures that process data in a streaming or pipelined manner, thus satisfying the high computational and communication demands of emerging applications. However, manually implementing an efficient dataflow architecture for large-scale applications is still challenging, even for specialists who use high-level synthesis (HLS) to simplify FPGA programming. To address this, we introduce CODO, an automated compiler that generates feasible and efficient dataflow accelerators on FPGAs. CODO features a systematic method for detecting and eliminating both coarse-grained and fine-grained dataflow violations. Building on this, CODO performs both on- and off-chip data movement optimizations to maximize transfer efficiency. To guarantee a higher design quality, CODO performs automatic scheduling to generate high-performance dataflow accelerators, ensuring a balanced performance-resource trade-off. Synthesis results show that CODO delivers $1.45\times$ to $4.52\times$ latency speedups on typical computation kernels and $3.7\times$ to $33.8\times$ speedups on DNN models compared to SOTA frameworks. In on-board evaluations, CODO achieves $7.3\times$ average speedup on CNN models and $2.07\times$ average speedup on the GPT-2 model over SOTA frameworks. The compiler is open-sourced at https://github.com/sjtu-zhao-lab/codo-artifact.

Moiré superlattices are generally assumed to act only at the interface where lattice mismatch or twist occurs. Here, we study charge transport in large-angle helical twisted trilayer graphene, where interlayer tunneling is strongly reduced. When only the top monolayer graphene is aligned with hBN, the electronic response reorganizes into a moiré-modulated monolayer and a remaining twisted bilayer graphene subsystem. Despite the absence of any explicit structural moiré in the twisted bilayer, we observe satellite-like features in its electronic response that run parallel to the primary spectrum and are locked to the density scale of the hBN/graphene moiré. These findings indicate that a moiré potential may not be confined to its structural interface and can, through electrostatic coupling, influence a spatially separated Dirac subsystem even in the absence of strong interlayer tunneling.

The production of two isolated photons in high-energy hadron collisions poses a challenge to perturbative QCD because of large corrections through next-to-next-to-leading order (NNLO). We present novel next-to-next-to-next-to-leading order ($\text{N}^3$LO) predictions and finally demonstrate perturbative convergence for this process. We discuss the considerable computational challenges and phenomenological results for the Large Hadron Collider.

Recrystallized tungsten (W) samples were irradiated by 20 MeV self-ions at 1350 K to peak damage doses in the range of 0.001-2.3 dpa. The irradiation-induced defects were then decorated with deuterium (D) by a gentle D plasma exposure ($<5$ eV/D, $5.6 \times 10^{19}$ $\text{D} / (\text{m}^2 \text{s})$) at 370 K. The D depth profiles in the samples were measured using $\rm D(^{3}He,p)α$ nuclear reaction analysis. The maximum trapped D concentration evolves differently with the damage dose compared with the previously studied irradiations at 290 K and 800 K. At the damage doses below 0.1 dpa, the D concentrations are lower than those after the irradiation at 800 K. At higher damage doses, the D concentrations exceed the 800 K values and reach 1.7 at.% at 2.3 dpa, showing no clear tendency towards saturation. Transmission electron microscopy revealed the presence of nm-sized voids in the samples irradiated at 1350 K, in contrast to the ones irradiated at 290 K and 800 K. Thermal desorption spectroscopy (TDS) indicates that the dominant D trapping sites are different compared to the irradiations at 290 K and 800 K. Reaction-diffusion simulations show that the TDS spectra can be described by assuming that D is trapped as $\rm D_2$ gas in the void volume and as D atoms at the void surface.

LiGa$_5$O$_8$ in a spinel type structure has recently been claimed to be an unintentional p-type ultra-wide-band-gap oxide semiconductor. While previous computational work did not yet identify the origin of p-type doping and in fact predicted insulating behavior by compensation of deep acceptors by shallow donors, defect characterization in terms of its optical signatures remains important. Rather than focusing on thermodynamics transition levels, as in earlier work, this present paper focuses on the vertical transitions in a defect configuration diagram of defects in different charge states, representing absorption and emission processes involving carrier capture/emission from/to band edges. In addition, the structural relaxation of several native defects is revisited by allowing for more complex symmetry-breaking distortions in an effort to reconcile conflicting results in the previous literature. Special attention is given to the Li vacancy because it is the shallowest native acceptor. For this defect, the previously reported transition levels are revised on the basis of symmetry-breaking relaxations. The structural relaxations, band structures, and densities of states are compared between the symmetry-broken polaronic and symmetry-conserving non-polaronic states. Finally, we also study carbon impurities, which are likely to originate from growth methods involving organic precursors.

The functionality of catalysts, enzymes, and supramolecular assemblies emerges not from individual molecules alone, but from the subtle interplay between multiple components arranged in complex systems. Designing such systems is a grand challenge, the combinatorial explosion of possible chemical compositions and spatial arrangements makes brute-force exploration infeasible, while many current generative approaches remain limited to isolated molecules. In this work, we introduce a hierarchical generative optimization framework that overcomes this barrier by coupling a genetic algorithm for configurational search with a generative model for molecular design. This closed-loop approach enables simultaneous refinement of geometry and composition, efficiently steering discovery toward systems with targeted functionality. As a proof of concept, we design catalytic environments for the Claisen rearrangement of p-tolyl ether by optimizing surrounding components around a fixed reference transition-state geometry. Despite this constraint during the search phase, post-hoc validation via Climbing-Image Nudged Elastic Band calculations confirm a 30% reduction in activation barrier. Beyond this example, our framework provides a general strategy for data-driven discovery of functional multi-component systems, opening the door to automated design of catalysts, enzyme active sites, and advanced materials. Scientific contribution. The study presents a closed loop generative framework that enables joint optimization of molecular components and their spatial organization in multi-component systems. The method moves generative molecular design beyond single molecules toward larger and more complex systems.