The former leader of the AI ethics team at Google, who was fired for raising the alarm about bias, Timnit Gebru, chronicles her journey as a refugee and Black woman in STEM, delivering an inspiring vision for better, human-centered, equitable technology.

Pollination is a key ecological process sustaining biodiversity and food security, yet global patterns of plant--pollinator specialization remain unresolved. Using a global dataset of quantitative plant--pollinator networks (>3,400 networks, >110,000 interactions), we show that specialization is latitudinally structured, but not as a single directional pattern. We find no consistent monotonic increase or decrease in specialization towards the Equator. Instead, network-level and pollinator specialization peak at low northern latitudes near the tropical--subtropical boundary, whereas plant specialization is hemispherically asymmetric. Climate is most often the best-supported predictor of specialization compared with latitude, biodiversity and environmental productivity. Network-level specialization declines with increasing mean annual temperature, with strong group-specific differences. Precipitation exerts nonlinear effects: plant specialization peaks at intermediate precipitation but is lower at the wettest sites, whereas pollinator specialization is lower at intermediate precipitation, although often in interaction with temperature. Pollinator groups diverge strongly, especially in contrasting precipitation-related patterns between birds and insects. These findings show that global plant--pollinator specialization is structured by latitude and climate in complex and taxon-specific ways, with implications for uneven responses of pollination networks to future climatic change.

The rapid emergence of increasingly agentic AI systems poses a structural challenge to the future of meaningful human work. As AI expands from task automation to autonomous goal pursuit, preserving human agency can no longer rely on ethical guidelines alone. This manifesto advances Hybrid Intelligence (HI) as a multi-level socio-technical transformation model aimed at keeping human judgment economically indispensable, by developing human–AI workflows that outperform pure automation across a broad range of applications. Headline results in which AI alone beats AI-assisted professionals test automation against unscaffolded oversight rather than against hybrid intelligence; the decisive comparison has yet to be run, and this collection builds the apparatus to run it. Drawing on more than fifty interdisciplinary roundtables and two foresight workshops, we, the HI Manifesto consortium of major European HI research centers and 60 scholars and AI practitioners, articulate an urgent mission: to design interfaces, governance frameworks, institutional policies, and market mechanisms such that human value creation scales through competitive human–AI collaboration rather than displacement. This does not require humans to approve every AI output; it requires that human contextual judgment be structurally embedded in value-creating workflows, irreducible to symbolic oversight, and that such workflows constitute the majority rather than a fringe of the future economy. We reinterpret complementarity as a dynamic frontier between prediction and judgment that shifts as contextual knowledge becomes machine-operational and that must be actively managed through deliberate workflow design. On this foundation we contribute: (1) a durable, knowledge-structure-grounded definition of complementarity with operational classification tests; (2) prescriptive multi-level HI definitions distilled from ten consensus criteria and constructed as exclusion instruments against HI-washing, with qualified terms spanning interface, practice, organization, and society; (3) a six-level interface classification from autonomous AI to collaborative HI; (4) a five-dimensional use-case maturity matrix applied to iconic and partner cases, revealing distinct HI business-model patterns; and (5) a foresight analysis of four AI futures identifying the transition's drivers and its valley of death. Four companion papers and a cross-cutting interaction framework (FERC) operationalize the interface, organizational, macroeconomic, and educational levels, including first field results showing the framework is buildable by ordinary commercial teams. We conjecture that granular hybrid workflows can reduce dependence on frontier-scale compute, with competitiveness, sustainability, and sovereignty as the dividend, and we close by extending the program into a constellation of Hybrid Intelligence Missions with concrete institutional calls to action. HI thus emerges not as a tool choice but as a systemic design paradigm for safeguarding meaningful work and human agency in the age of agentic AI.

Staphylococcus aureus (S. aureus) is an increasingly recognized intracellular pathogen, yet infection outcomes vary with bacterial isolate and host cell type. Many mechanisms underlying these differences remain poorly understood. This study investigated how distinct intracellular S. aureus isolates influence host signaling programs and infection outcomes by modulating cell death pathways and TNF-R1 dependent regulation of host cell fates across different human cell lines.

Aging is a major risk factor for cardiovascular diseases and is often accompanied by systemic alterations of the immune system, such as the accumulation of terminally differentiated T cells, clonal hematopoiesis of intermediate potential, and dysregulated production of pro-inflammatory cytokines. However, how the aging immune system impacts post-myocardial infarction (MI) repair remains poorly understood. In the present study, we compared the post-MI inflammatory responses in 2- and 18-month-old C57BL/6 J mice of both sexes and monitored the distribution of interferon-gamma (IFN-γ) producing cells in these conditions using Ifng-YFP reporter mice. Our results show a conserved IFN-γ production signature both in mice and humans during physiological aging. In mice, the aging myocardium exhibited increased pro-inflammatory gene expression signature with increased recruitment of IFN-γ-expressing T cells following MI. Moreover, while this age-related inflammation had little impact in the acute post-MI responses, a persistent IFN-γ-production signature observed in elderly mice was associated with an aggravation of chronic adverse cardiac remodeling. Taken together, our results indicate that age-related smoldering inflammation may fuel the long-term progression of ischemic heart failure.

As large language models (LLMs) increasingly serve as a technology basis for conversational health behaviour interventions, understanding how users evaluate and respond to these systems becomes critical for their acceptance. When LLM-based coaches are anthropomorphically designed – e.g. through human names, avatars, and messaging interfaces – they may trigger interpersonal trust expectations traditionally reserved for human relationships. This study investigates whether and to what extent the interpersonal trust antecedents of ability, benevolence, and integrity influence trust-related user responses in LLM-based health coaching for fitness and nutrition. Using a vignette-based survey design and combining partial least-squares structural equation modelling (PLS-SEM) with necessary condition analysis (NCA), we identify which antecedents are essential for perceived usefulness and intention to adopt. Results show that integrity is a necessary condition for both outcomes, with a medium to large effect, and that ability significantly influences perceived usefulness but is not necessary. Benevolence, however, was found to be neither a necessary nor a significant predictor in this context. These findings offer theoretical insights into the selective transfer of interpersonal trustworthiness beliefs to AI and provide practical guidance for the design of trustworthy anthropomorphic coaching systems.

Reading comprehension is essential for accessing scientific knowledge, yet many students, especially non native readers, struggle to understand complex academic texts. Comprehension theory shows that understanding requires the coordinated interaction of multiple subskills and the construction of a coherent situation model. Existing reading support systems typically address only isolated aspects such as summarization or simplification, resulting in superficial understanding. We present PaperMate, a canvas based reading support system designed to support situation model construction. Grounded in the Active Reader View, the artifact integrates simplification, structural guidance, essentialization, and contextual clarification within a unified and user controlled interface. This enables readers to flexibly combine support based on individual needs. The artifact was developed through multiple iterative design cycles including proof of concept, proof of value, laboratory studies, and field tests, and is continuously refined. The work demonstrates how holistic reading support can be operationalized in a concrete artifact and contributes to theory grounded design of reading support systems.

Purpose In response to various calls for more research extending our understanding of trust in IT artifacts, we synthesize the empirical evidence of prior studies in the rich stream of trust-related research to gain a more consistent picture regarding the role of trust antecedents in human–technology relationships. Design/methodology/approach We conduct an exploratory meta-analysis of 374 primary studies reported in 348 articles covering 26 years of trust research to gain deeper insights into the role of the antecedents to trust in IT artifacts. Our research framework combines 19 independent variables conceptualizing human- and system-like trusting beliefs, technology acceptance factors, system characteristics, user characteristics and contextual factors with perceived risk, trusting intention and trusting behavior as the dependent variables. Findings The findings confirm the pervasive role of the human- and system-like trusting beliefs ability/competence, benevolence, reliability and helpfulness as well as the technology acceptance factors perceived enjoyment and perceived usefulness in building trust and fostering trusting behavior. In addition, the findings indicate the importance of the system characteristics security and privacy in forming trust and helping users overcome the perceived degree of uncertainty on the way to fostering behavioral and continuous intention. Originality/value Based on the findings, we discuss the implications for research and practice, propose design recommendations for increasing trust in IT artifacts and provide future research avenues to explore. Considering that trust significantly influences IT adoption, our study's integrated perspective on trust is crucial for deepening our understanding of key factors in fostering trust between humans and technology.

Trust violations in sociotechnical systems are becoming increasingly visible as AI systems take on advisory and decision-making roles in everyday life. When these systems fail, users must interpret what went wrong and decide whether to continue relying on them. Yet IS research has largely focused on trust formation, leaving the processes through which trust is broken, and potentially repaired, undertheorized. Meanwhile, adjacent fields have produced fragmented insights into trust violations and the mechanisms used to repair them. This paper presents a systematic literature synthesis of 62 studies to understand how trust violations and repair are conceptualized when technology plays a meaningful role. We identify four core stages that structure trust breakdown and recovery: the violation incident, the trustor sensemaking, the violator reaction, and the recalibrated trust outcome. We consolidate these insights into a process-oriented framework and guiding questions, outlining implications for human-AI collaboration and future IS research.

Reading complex informational text is essential for comprehending global scientific knowledge, yet many nonnative readers from practice-oriented pathways struggle to comprehend such materials and require adaptive support that existing systems do not provide. However, past research lacks prescriptive design knowledge on how to effectively design such support systems. Hence, we present results from our design science research endeavor to design and evaluate an adaptive reading support system grounded in comprehension theory. We built on design principles and instantiated them as adaptive hints and support functions typically scattered across separate tools. We evaluated the systems in an online experiment with 84 participants across three conditions: control, conversational interface, and canvas interface. The results show that a conversational interface grounded in ICAP improves comprehension, while canvas-based instantiations of prior design principles require careful design to manage their higher interaction complexity, illustrating how system design grounded in linguistic theory can support the processing of complex texts. The study contributes design knowledge for effective comprehension support systems.

The novel paradigm of Generative user interfaces introduces a shift from predefined, static interfaces toward dynamically generated and contextually adaptive user experiences. While personalization in Information Systems (IS) is well established, it has largely focused on adapting content and functionality, whereas the user interface itself often remains predefined and static. Generative user interfaces address this limitation by leveraging generative artificial intelligence to generate front-end code, constructing and adapting interface components at runtime based on users’ needs, preferences, and context. This promises to not only enhance personalization but also disrupt current IS development. Yet, this emerging implementation raises fundamental questions for IS design and development. Although Generative user interfaces have recently attracted research interest in Human–Computer Interaction, there remains a lack of shared conceptual understanding of their design characteristics and implications for IS development. To address this gap, we develop a taxonomy that identifies these characteristics based on a Socio-Technical Systems perspective. We validate our taxonomy with experts and formulate a research agenda for Generative user interfaces, using the taxonomy as a conceptual foundation to guide future IS research.

Western honeybees (Apis mellifera) are remarkable among insects for their ability to generate heat endogenously and maintain a stable hive microclimate. This process relies on the activity of the indirect flight muscles, but the direct relationship between thermogenesis and carbohydrate availability has not been systematically investigated. To investigate this relationship, we developed a controlled individual-level thermography assay that provides a standardized framework for studying honeybee thermogenesis at the individual level. Starved bees showed little or no thermogenesis, whereas sucrose-fed bees exhibited substantially increased thoracic temperatures and thermogenic activity, consistent with a threshold-like activation of thermogenesis. Additionally, thermogenic performance was significantly lower in bees fed a sucralose solution used as a putatively non-nutritive control than in bees that received the same volume of sucrose solution. Together, these findings support the hypothesis that thermogenesis depends on the availability of recently ingested carbohydrates.

Reliable prediction of time-varying channel state information (CSI) is essential for efficient wireless communication. Each CSI frame is a matrix-valued representation of the wireless channel response, and a sequence of CSI frames forms a temporal channel trajectory. Modern deep learning-based CSI predictors, however, often provide only point predictions and lack calibrated uncertainty estimates. This limitation is particularly problematic in multi-step CSI prediction, where the target is a sequence of future CSI matrices, and downstream decisions such as beamforming or scheduling may fail if any part of the predicted trajectory is unreliable. We propose trajectory-adaptive calibration and error profiling with conformal risk control (TRACE-CRC), a method for trajectory-aware uncertainty quantification in multi-step CSI prediction. TRACE-CRC constructs Frobenius-norm uncertainty balls around predicted CSI matrices and controls the risk that at least one future frame is uncovered. Instead of calibrating each future step independently, TRACE-CRC combines future-step-dependent error profiling, trajectory difficulty stratification, and learn-then-test (LTT) risk control. Empirically, TRACE-CRC achieves reliable trajectory-level coverage with substantially smaller uncertainty balls than conservative multi-step corrections, while avoiding the trajectory undercoverage of compact stepwise and adaptive conformal base-lines.

A semiclassical framework models quantum quadrupoles as biaxial tensors, showing how angular-momentum-dependent geometries dictate generalized nematic phases and quadrupolar spin liquids in frustrated magnets.

Topological superconductivity and Majorana zero modes emerge from two feasible platforms, quasi-1D nanowires and vortex-line heterostructures, when crystal anisotropy and quantum confinement act as tuning parameters.

Transition-metal doping enables the introduction of spin functionality into halide double perovskites, while simultaneously modifying optical properties. Here, we combine controlled single-crystal growth, optical characterization, comprehensive electron paramagnetic resonance (EPR) spectroscopy, and first-principles modeling to identify the microscopic nature of Fe-related centers in Fe-doped \($\mathrmCs_2\mathrmAgBiBr_6$\). Single crystals with nominal \($\mathrmF\mathrme^3+$\) concentrations up to 15% in the precursor stage were grown using a controlled-cooling method, yielding reproducible Fe incorporation up to 0.1% with respect to Bi, without secondary phases. Despite this low concentration, Fe doping introduces electronic states that influence optical absorption and photoluminescence. EPR measurements reveal an S = 5/2 \($\mathrmF\mathrme^3+$\)-related center whose anisotropy follows the cubic-to-tetragonal phase transition below 120 K. Angular-dependent EPR resolves two configurations of this nearly axial spin center, with principal axes rotated by 90ifmmode^circelsetextdegreefi and aligned with the a/b plane of the tetragonal lattice. Density-functional calculations attribute these centers to impurity-vacancy complexes, most likely \($\mathrmFe_\mathrmBi\text\ensuremath-\mathrmV_\mathrmBr$\), that stabilize in a basal configuration of the low-temperature phase. This approach resolves vacancy-coupled defect orientations, narrowing possible models to \($\mathrmF\mathrme^3+$\)-vacancy complexes and establishing them as stable, orientation-sensitive spin probes of structural symmetry in halide double perovskites, while providing a microscopic basis for tuning their magnetic and optical responses.

A study of the counter-Nagaoka problem of a single hole in an infinite-\($U$\) Hubbard model on the kagome lattice reveals a new type of polaron, a resonating-valence-bond polaron that morphs into semiclassical order as polarization is reduced.

Building on earlier reports of superconductivity in bulk PtSb, we present a systematic study of superconductivity in epitaxial PtSb(0001) thin films grown on \($\textSrF_2(111)$\). Electrical transport measurements reveal a superconducting transition at \($T_\mathrmc=1.72\phantom\rule0.16em0ex\mathrmK$\). The field-induced broadening of the transition and the temperature dependence of the upper critical fields are consistent with type-II superconductivity. We determine the upper critical fields for magnetic fields applied perpendicular and parallel to the film plane and parametrize their temperature dependence using an anisotropic Ginzburg-Landau approach. For a representative film in the intermediate thickness regime (\($d=50\phantom\rule0.16em0ex\mathrmn\mathrmm$\)), we obtain coherence lengths of \($\ensuremath\xi_ab\ensuremath\approx56\phantom\rule0.16em0ex\mathrmn\mathrmm$\) and \($\ensuremath\xi_c\ensuremath\approx14\phantom\rule0.16em0ex\mathrmn\mathrmm$\). Current-voltage characteristics show sizable critical currents, with a critical current density reaching \($J_\mathrmc\ensuremath\approx6\ifmmode\times\else\texttimes\fi10^4\phantom\rule0.16em0ex\mathrmA/\textcm^2$\) at \($0.5\phantom\rule0.16em0ex\mathrmK$\). These results establish the superconducting properties of epitaxial PtSb thin films and provide a basis for lattice-matched heterostructures within the NiAs-type materials family.

We introduce many-body caging as a novel mechanism for nonthermal behavior in quantum matter. We define many-body cages as eigenstates that, through quantum interference, become localized on a subgraph of the many-body state graph. These many-body cages can lead to the formation of flat bands in the many-body spectrum at characteristic, system-independent energies. These flat bands can realize a novel type of glassy eigenspectrum order in the absence of disorder, which we quantify by a band-overlap order parameter with an intricate, possibly fractal, distribution over the many-body state graph. We further show that these many-body cages exhibit distinctive signatures in experimentally accessible quantities, such as through a nonvanishing long-time memory of the initial condition, and many-body Rabi oscillations set by the characteristic flat-band energies. While our predictions in principle apply to any constrained quantum system, we demonstrate them here for 2D lattice gauge theories and models relevant for current experiments in Rydberg atoms. We expect that these many-body cages offer a promising route to realize nonequilibrium quantum states with novel properties.

Transverse thermoelectric devices offer simplified geometries with flexible designs for next-generation cryogenic technologies. However, mechanisms on geometrical amplification of transverse thermoelectric effects remain elusive. Here we establish a general framework linking reduced dimensionality with its geometrical anisotropy to enhanced transverse thermoelectric responses. Taking quasi-one-dimensional Li0.9Mo6O17 as a model platform, we combine theoretical modeling with systematic experiments to reveal extraordinarily large Nernst and Ettingshausen signals, reaching 11,430 \($\mu$\)V/K (25\thinspaceK, 9\thinspaceT) and 0.0063thinspaceKthinspacetextperiodcenteredthinspacem/A (43.1\thinspaceK, 9\thinspaceT), respectively, together with a high power factor of 814 \($\mu$\)W cm−1 K−2 at 25thinspaceK and 9thinspaceT. Dimensional reduction amplifies the Seebeck-driven electric field and enhances the reduced mobility through anisotropic Fermi surface geometry. These effects yield the transverse thermoelectric response far exceeding those of higher-dimensional systems with even ultrahigh carrier mobilities. Our findings highlight enhanced geometrical anisotropy as a powerful principle of design for engineering high-performance transverse thermoelectric systems.

NiS2, a material long recognized for its non-collinear magnetic order and correlated insulating behavior, has until now been entirely overlooked as a potential host of topological phenomena. Bulk NiS2 is known to be a Mott or charge-transfer insulator, yet its surface exhibits anomalous finite conductivity of unclear origin. Additionally, a definitive consensus regarding its low-temperature magnetic transition remains elusive. In this work, we provide a unified explanation for both of these longstanding puzzles. First, we present a refined characterization of NiS2's magnetic phases by proposing a novel ground state based on a rigorous symmetry analysis of high-resolution neutron scattering data. Second, through high-resolution scanning tunneling microscopy and spectroscopy (STM/STS), we reveal the presence of robust edge states on both Ni- and S-terminated surfaces, which persist even under applied magnetic fields. Using ab initio calculations combined with topological analysis, we attribute these edge states to obstructed atomic charges arising from the bulk's obstructed nature. Altogether, this study not only sheds new light on the physics of NiS2, but also establishes a solid experimental and theoretical foundation for exploring the interplay between topology and electronic correlations.

The electronic properties of materials exhibiting giant thermopower with a thermopower that exceeds 2000 ensuremathmuV/K, such as EuCd\($_2$\)P\($_2$\), are strongly temperature dependent.

Music recommendation has been a core area of interest within the recommender systems community since its early days. With the rise of music streaming platforms, algorithmic recommendations have become central to the music industry. However, critical challenges remain concerning the design, evaluation, and societal impact of music recommender systems. This fourth edition of the Music Recommender Systems Workshop (MuRS) centers on the growing influence of generative content on music recommendation. The rapid influx of AI-generated music transforms the streaming ecosystem, raising critical questions about discoverability, authenticity, and the curational role of recommender systems. The challenges and opportunities associated with AI-generated content extend beyond music and recommender systems, demanding transparent, fair, and accountable recommendation frameworks that reflect the interests of the diverse set of stakeholders.

Quantum spin Hall insulators, with dissipationless edge channels protected by time-reversal symmetry, are promising for low-power and quantum devices. However, experimentally realized examples with sizable nontrivial gaps remain scarce. Monolayer Si2Te2 has been theoretically predicted to host a room-temperature quantum spin Hall phase, but the absence of a bulk analog has hindered its experimental realization. Here we show that HfTe2 provides an ideal van der Waals template for the epitaxial growth of strain-free monolayer Si2Te2 while preserving its nontrivial topological phase. Scanning tunneling microscopy and spectroscopy reveal an intact (1\texttimes1) lattice and a bulk band gap of  textasciitilde 300 meV, consistent with first-principles calculations. Moreover, pronounced edge states extending textasciitilde 2.0 nm from the island boundary are observed in monolayer Si2Te2, exhibiting characteristics expected for topological edge modes. Our results establish monolayer Si2Te2 as a large-gap two-dimensional topological-insulator platform for topological phenomena and device concepts at elevated temperatures.

Here, the authors revisit the question of the logarithmic renormalization of the Fermi velocity in graphene. Using large-scale Quantum Monte Carlo simulations of lattices with up to \($2\ifmmode\times\else\texttimes\fi10^4$\) interacting electrons, they directly connect experimental data with unbiased, nonperturbative calculations starting from a microscopic Hamiltonian. Their results reveal that even random-phase approximation corrections are quantitatively insufficient at realistic interaction strengths for suspended graphene, while continuum perturbation theory misses important lattice-scale effects. Remarkably, the optical conductivity still remains constant pointing to exact cancellation of different corrections.

Traditional theories of electron transport in crystals are based on the Boltzmann equation and do not capture physics arising from quantum coherence. We introduce a transport formalism based on orbital Wigner functions, which accurately captures quantum coherent physics in multiband fermionic systems. We illustrate the power of this approach compared with traditional semiclassical transport theory by testing it numerically against microscopic simulations of one-dimensional, noninteracting, two-band systems---the simplest systems capable of exhibiting interorbital coherence. We show that orbital Wigner functions accurately capture strongly nonequilibrium features of electron dynamics that lie beyond conventional Boltzmann theory, such as the ballistic transport of a relative phase between microscopic orbitals and topological Thouless pumping of charge, both at nonzero temperature and away from the adiabatic limit. Our approach is motivated in part by modern ultracold atom experiments that can prepare and measure far-from-equilibrium charge transport and phase coherence in multiband fermionic systems, calling for correspondingly precise theories of transport. The quantitative accuracy exhibited by our approach, together with its capacity to capture nontrivial physics even at the ballistic scale, establishes orbital Wigner functions as an ideal starting point for developing a fully systematic theory of transport in crystals.

We study twisted bilayer \($\mathrmWSe_2$\) within a continuum moir'e model and apply a method for treating finite geometries directly in the continuum framework, thereby avoiding the limitations associated with purely momentum-space formulations and Wannier obstructions. By projecting a confinement potential onto bulk moir'e eigenstates, we obtain a real-space description of edge physics without lattice models. Applying this approach to nanoribbons, we demonstrate chiral edge modes consistent with bulk Chern numbers and reveal their moir'e-scale character. In the magic-angle regime, these states are strongly localized, exhibit layer-polarized counterpropagating modes, and are electrically tunable via a displacement field, enabling control of localization, hybridization, and topological transitions. Our results establish a general framework for boundary physics in topological moir'e materials.

Transport properties of strongly correlated materials have contributions from quasiparticle excitations such as electrons and holes as well as emerging collective excitations such as plasmonic soundlike modes which are sustained by interactions. It was previously shown by Pongsangangan et al. [Phys. Rev. B 106, 205127 (2022)] that the thermal excitation of the long-lived plasmons in graphene provides a substantial contribution to heat and momentum transport in the interaction-dominated regime. Detailed information on these excitations is therefore necessary for the understanding of hydrodynamic transport with quantitative precision. On the other hand, dynamics of graphene plasmons is usually studied using perturbation theory within the Dirac cone approximation, thus neglecting the effects of a finite Brillouin zone (BZ) and higher-order perturbative corrections. Both these effects can be, however, significant for strong-interacting systems including free-standing graphene, where the effective coupling constant can reach values up to \($\ensuremath\alpha=\frace^2\ensuremath\hbarv_\mathrmF=2$\). Consequently, in this paper, we study the behavior of plasmons in half-filled free-standing graphene using unbiased quantum Monte Carlo (QMC) calculations. We confirm the existence of well-defined resonance peaks for plasmons around the \($\mathrm\ensuremath\Gamma$\) point. Comparison with the random phase approximation (RPA) calculation for the honeycomb lattice shows qualitative agreement with QMC. However, RPA yields more stable plasmon modes, while QMC shows enhanced broadening due to nonperturbative interaction effects naturally included in the simulations. To account for this effect, we generalize the calculation of lattice RPA by dressing the fermionic propagator by a constant lifetime. As a result, the plasmon frequencies are shifted toward higher energies, and the spectrum is broadened, yielding better agreement with QMC. Our findings highlight the need to account for both a finite BZ and strong interaction effects when developing theories of electronic transport in free-standing graphene.

Chirality induced by rolling a two-dimensional material into a spiral geometry reshapes its electronic band structure. In this work we theoretically investigate the topological properties of carbon nanoscrolls under an axial magnetic field, focusing on structures in which chirality is encoded through shifted edge alignments. In contrast to unshifted structures, where mirror symmetry pins the Dirac cones to half a flux quantum, chiral carbon nanoscrolls lack this symmetry, and Dirac cones emerge at magnetic flux values away from half a flux quantum. We demonstrate that these Dirac cones are topologically protected by combined inversion--time-reversal symmetry and remain robust even when sublattice symmetry is broken. Furthermore, we show that the number of Dirac cones and their real-space probability distributions depend on the number of turns and the magnetic field strength. Our study elucidates the role of chirality in the band topology of nanoscroll geometries.

We have studied the magnetic properties of clinoatacamite \($\mathrmCu_2\mathrmCl(\mathrmOH)_3$\), the parent compound of the quantum spin liquid candidate herbertsmithite and a long-standing puzzle among frustrated quantum magnets. As we reveal using density-functional theory, clinoatacamite belongs to the class of distorted kagome antiferromagnets with the kagome plane being embedded into a low-symmetry crystal structure. By means of thermodynamic measurements, muon spin rotation and relaxation, as well as neutron diffraction on single crystals, we find in zero magnetic field complex behavior below \($T_\mathrmI=18.1\text \text \mathrmK$\) which unfolds in three temperature regions I--III. We propose this complexity in multicritical clinoatacamite to arise from the competition of antiferromagnetic ordering modes from the underconstrained manifold of modes, which can lead to a metamagnetic texture in zero field.

Recommendation and personalization research is no longer confined to computer science. Relevant questions are actively explored across disciplines including information systems, communication science, management, cognitive science, psychology, education, media studies, marketing, and the social sciences, often grounded in distinct theories, methods, and evaluation traditions. “Beyond Algorithms: A Workshop on the Interdisciplinarity of Recommender Systems” (BEYOND 2026) aims to create a dedicated space at ACM RecSys 2026 to connect these perspectives. The workshop highlights how insights from adjacent fields can complement, extend, and sometimes challenge computing-centric approaches. It explicitly welcomes contributions studying recommendation, personalization, and related phenomena outside core computing venues, as well as integrative work that bridges computational and non-computational approaches. Through presentations and interactive discussions, we aim to surface diverse theoretical frameworks, methodological traditions, and evaluation paradigms. Our goal is to foster dialogue across disciplinary boundaries, make disciplinary assumptions explicit, and identify shared research challenges and opportunities for collaboration. By strengthening ties between the RecSys community and neighboring research traditions, BEYOND 2026 seeks to broaden participation and visibility for interdisciplinary research and to position RecSys as a hub for both technical innovation and cross-disciplinary exchange.

Mathematical argumentation is a core competence that integrates mathematical reasoning with broader critical thinking skills, enabling students to construct coherent and logically structured arguments to evaluate mathematical claims. This skill is essential for undergraduate students across disciplines such as information systems, mathematics, and computer science—yet many struggle to develop it. Recent advances in Large Language Models (LLMs) and multi-agent systems offer a promising opportunity to address these challenges by supporting students in refining their mathematical argumentation within a tutorial setting. In this Research-in-Progress paper, we adopt a Design Science Research approach to generate design knowledge for an LLM-based tutoring tool aimed at fostering students’ argumentation skills. Drawing on user requirements and meta-requirements derived from user interviews and theoretical foundations, we derive four design principles and propose a prototype of a multi-agent LLM-based tutoring tool. The system delivers formative feedback on students’ proofs and provides tutorial guidance through step-by-step explanations, grounded in theories of formative feedback and tutorial guidance.

Ancestral recombination graphs (ARGs) are an increasingly important component of population and statistical genetics. The tskit library has become key infrastructure for the field, providing an expressive and general representation of ARGs together with a suite of efficient fundamental operations. In this note, we announce tskit version 1.0, describe its underlying rationale, and document its stability guarantees. These guarantees provide a foundation for durable computational artefacts and support long-term reproducibility of code and analyses.

Biobank-scale genomic analyses are increasingly constrained by computational costs, as hundreds of thousands to millions of samples and variants must be analyzed together. The genotype representation graph (GRG) compactly encodes population genetic variation to accelerate computation, but the current approach does not exploit modern accelerator architectures. This work introduces Mikado, a new methodology for expressing GRG-based computation, previously performed by graph traversal, using sparse linear algebra primitives. Under a reverse topological ordering of the graph nodes, the GRG adjacency matrix is strictly block-lower-triangular, and the genotype matrix-vector product becomes a sparse triangular solve that can be further decomposed into a pipelined sequence of blocked sparse matrix-vector multiplies. By decoupling computation from graph representation, our approach exposes fine-grained parallelism and enables hardware-optimized sparse primitives on GPUs. Mikado achieves an order-of-magnitude speedup and cost savings for PCA and BOLT-LMM compared with the original GRG traversal approach, including on All of Us cohorts. It provides a scalable, hardware-portable, researcher-friendly tool for population genetics at biobank scale.Competing Interest StatementThe authors have declared no competing interest.NSF, IIS-243580NIH, R35-GM150579

Ancestral recombination graphs (ARGs) are an attractive means for quantitative genetic analysis of complex traits because they encode the realized genetic relatedness between a sample of individuals in the presence of genetic drift, recombination, and mutation. Data structures for efficiently storing ARGs can also be used to rapidly process millions of genomes, and are thus promising for fitting linear mixed models to large phenotype and genome datasets. Here, we study the problems of variance component estimation and prediction of genetic values with ARGs, by describing a generative model of complex traits with additive effects on an ARG, and then developing algorithms that use the ARG to solve these problems efficiently on biobank-scale datasets. We observe nearly linear scaling of runtime with sample size, which is achieved by using the succinct tree sequence representation of the ARG for implicit matrix-vector products, along with modern randomized linear algebra algorithms. We estimate variance components using restricted maximum likelihood, which we find performs substantially better than the Haseman–Elston method. In simulation tests, both variance component estimation and prediction of genetic values (using the best linear unbiased predictor) perform nearly as well with inferred ARGs as with true ARGs. We also discuss interpretations of the variance component estimates as mutational variance and additive genetic variance. We provide an implementation of the algorithms as a Python package tslmm, which leverages the tree sequence library tskit.

Foreshadowing haplotype-based methods of the genomics era, it is an old observation that the “junction” between two distinct haplotypes produced by recombination is inherited as a Mendelian marker. In a genealogical context, this recombination-mediated information reflects the persistence of ancestral haplotypes across local genealogical trees in which they do not represent coalescences. We show how these non-coalescing haplotypes (“locally-unary nodes”) may be inserted into ancestral recombination graphs, a compact but information-rich data structure describing the genealogical relationships among recombinant sequences. The resulting ancestral recombination graphs are smaller, faster to compute with, and the additional ancestral information that is inserted is nearly always correct where the initial ancestral recombination graph is correct. We provide efficient algorithms to infer locally-unary nodes within existing ancestral recombination graphs, and explore some consequences for ancestral recombination graphs inferred from real data. To do this, we introduce new metrics of agreement and disagreement between ancestral recombination graphs that, unlike previous methods, consider ancestral recombination graphs as describing relationships between haplotypes rather than just a collection of trees.

Ultra-rare variants dominate whole-genome sequencing datasets, yet their interpretation is limited by allele frequency, which provides little information at very low counts and is highly sensitive to uneven ancestry representation. Allele age offers an ancestry-agnostic alternative but existing methods do not scale to biobank-sized cohorts. Here we present a scalable variational algorithm for dating Ancestral Recombination Graphs (ARGs), imple- mented in tsdate, together with new distributed methods enabling practical biobank-scale ARG inference using tsinfer. Applied to 47,535 genomes from the Genomics England 100,000 Genomes Project, we infer contiguous ARGs spanning 206 Mb and estimate ages for 23.2 million variants, including 11.8 million singletons. ARG-based allele ages remain accurate under extreme sampling imbalance and, in real data, reveal signatures of purifying selection and clinically relevant heterogeneity among variants with identical observed frequencies. Estimates for recent mutations are precise only at large sample sizes, highlighting the infor- mation accessible in the haplotype structure of large datasets. Biobank-scale ARGs therefore enable robust, ancestry-agnostic age estimation for ultra-rare variation with broad utility for statistical and clinical genomics.

Ancestral recombination graphs (ARGs) are the focus of much ongoing research interest. Recent progress in inference has made ARG-based approaches feasible across of range of applications, and many new methods using inferred ARGs as input have appeared. This progress on the long-standing problem of ARG inference has proceeded in two distinct directions. First, the Bayesian inference of ARGs under the Sequentially Markov Coalescent (SMC), is now practical for tens-to-hundreds of samples. Second, approximate models and heuristics can now scale to sample sizes two to three orders of magnitude larger. Although these heuristic methods are reasonably accurate under many metrics, one significant drawback is that the ARGs they estimate do not have the topological properties required to compute a likelihood under models such as the SMC under present-day formulations. In particular, heuristic inference methods typically do not estimate precise details about recombination events, which are currently required to compute a likelihood. In this article, we present a backwards-time formulation of the SMC (conventionally regarded as an along-the-genome process) and derive a straightforward definition of the likelihood of a general class of ARG under this model. We show that this formulation does not require precise details of recombination events to be estimated, and is robust to the presence of polytomies. We discuss the possibilities for ARG inference that this new formulation opens.

Simulation-based inference methods are increasingly being used in population genetics due to their flexibility and ability to be applied in settings where likelihood-based methods are intractable. Perhaps the best known such method is Approximate Bayesian Computation (ABC); however, its popularity is offset by its shortcomings which include computational expense and an unfortunate inability to efficiently fit models to high-dimensional summaries of the data. An alternative approach that solves these issues is supervised machine learning (ML); however, ML methods generally do not yield Bayesian uncertainty estimates of the quantities they predict. Here, we apply a recently introduced method, neural posterior estimation, that combines the best facets of ABC and supervised ML by training a neural network to estimate the posterior distribution of a population genetics model. We first compare neural posterior estimation with other inference methods for a variety of population genetic tasks, and show that neural posterior estimators yield posterior distributions with high accuracy and efficiency. We compare learned posterior distributions given raw genotypes and various summary statistics as input data. Additionally, we apply neural posterior estimation for demographic inference for simple and more complex models to highlight its application, including an analysis of demographic history in Drosophila melanogaster. Finally, we provide a user friendly workflow that enables others to perform neural posterior estimation on their own genetic data.

ABSTRACT Estimating the size of wild populations is a critical priority for ecologists and conservation biologists, but tools to do so are often labour intensive and expensive. A promising set of newer approaches are based on genetic data, which can be cheaper to obtain and less invasive than information from more direct observation. One of these approaches is close-kin mark-recapture (CKMR), a type of method that uses genetic data to identify kin pairs and estimates population size from these pairs. Although CKMR methods are promising, one limitation to using them more broadly is a lack of CKMR models that can deal with spatially structured populations and spatial heterogeneity in sampling. In this paper, we introduce a spatially explicit approach to CKMR that uses individual-based simulation in concert with a deep convolutional neural network to estimate population sizes. Using simulations, we show that our method, CKMRnn, is highly accurate, even in the face of spatial heterogeneity in sampling and spatial population structure, and demonstrate that it can account for potential confounders such as unknown population histories. Finally, to demonstrate the accuracy of our method in an empirical system, we apply CKMRnn to data from a Ugandan elephant population, and show that point estimates from our method recapitulate those from traditional estimators but that the confidence interval on our estimator is approximately 30% narrower.

This study evaluates the impact of transitioning from FP64 to 8-bit floating-point (FP8 E4M3) for image reconstruction using autoencoders. To address energy and memory constraints in edge computing, we apply Post-Training Quantization with dynamic scaling (range 448). Although quantization increases Mean Squared Error slightly, perceptual quality remains acceptable. Memory use per parameter drops by 87.5%, drastically reducing computational load while preserving utility. The findings confirm that low-precision formats enable efficient AI deployment on resource-limited devices, allowing embedded vision systems to operate with significantly lower energy overhead

Microservices have become a central architectural style for cloud-based systems, driving increased interest in their performance characteristics. However, the limited availability of real-world microservice applications for academic research constrains empirical studies in this area, particularly those using application-agnostic, machine learning-based approaches. Prior work has explored synthetic generation and simulation as alternative means of obtaining performance data; however, their benefits when used alongside real-world measurements are less understood. Using the saturation predictor Monitorless in a case study, we evaluate two scenarios. Our results indicate that generated and simulated applications improve predictive performance on previously unseen, real-world systems. In our case, generated applications reduced false positives, simulated applications enhanced robustness, while their combination yielded the most balanced predictor. The findings suggest that synthetic and simulated applications can effectively complement real-world data, enabling more diverse and robust evaluation of performance models.

Microservices have become a central architectural style for cloud-based systems, driving increased interest in their performance characteristics. However, the limited availability of real-world microservice applications for academic research constrains empirical studies in this area, particularly those using application-agnostic, machine learning-based approaches. Prior work has explored synthetic generation and simulation as alternative means of obtaining performance data; however, their benefits when used alongside real-world measurements are less understood. Using the saturation predictor Monitorless in a case study, we evaluate two scenarios. Our results indicate that generated and simulated applications improve predictive performance on previously unseen, real-world systems. In our case, generated applications reduced false positives, simulated applications enhanced robustness, while their combination yielded the most balanced predictor. The findings suggest that synthetic and simulated applications can effectively complement real-world data, enabling more diverse and robust evaluation of performance models.

Microservices have become a central architectural style for cloud-based systems, driving increased interest in their performance characteristics. However, the limited availability of real-world microservice applications for academic research constrains empirical studies in this area, particularly those using application-agnostic, machine learning-based approaches. Prior work has explored synthetic generation and simulation as alternative means of obtaining performance data; however, their benefits when used alongside real-world measurements are less understood. Using the saturation predictor Monitorless in a case study, we evaluate two scenarios. Our results indicate that generated and simulated applications improve predictive performance on previously unseen, real-world systems. In our case, generated applications reduced false positives, simulated applications enhanced robustness, while their combination yielded the most balanced predictor. The findings suggest that synthetic and simulated applications can effectively complement real-world data, enabling more diverse and robust evaluation of performance models.

Microservices have become a central architectural style for cloud-based systems, driving increased interest in their performance characteristics. However, the limited availability of real-world microservice applications for academic research constrains empirical studies in this area, particularly those using application-agnostic, machine learning-based approaches. Prior work has explored synthetic generation and simulation as alternative means of obtaining performance data; however, their benefits when used alongside real-world measurements are less understood. Using the saturation predictor Monitorless in a case study, we evaluate two scenarios. Our results indicate that generated and simulated applications improve predictive performance on previously unseen, real-world systems. In our case, generated applications reduced false positives, simulated applications enhanced robustness, while their combination yielded the most balanced predictor. The findings suggest that synthetic and simulated applications can effectively complement real-world data, enabling more diverse and robust evaluation of performance models.

Microservices have become a central architectural style for cloud-based systems, driving increased interest in their performance characteristics. However, the limited availability of real-world microservice applications for academic research constrains empirical studies in this area, particularly those using application-agnostic, machine learning-based approaches. Prior work has explored synthetic generation and simulation as alternative means of obtaining performance data; however, their benefits when used alongside real-world measurements are less understood. Using the saturation predictor Monitorless in a case study, we evaluate two scenarios. Our results indicate that generated and simulated applications improve predictive performance on previously unseen, real-world systems. In our case, generated applications reduced false positives, simulated applications enhanced robustness, while their combination yielded the most balanced predictor. The findings suggest that synthetic and simulated applications can effectively complement real-world data, enabling more diverse and robust evaluation of performance models.

Microservices have become a central architectural style for cloud-based systems, driving increased interest in their performance characteristics. However, the limited availability of real-world microservice applications for academic research constrains empirical studies in this area, particularly those using application-agnostic, machine learning-based approaches. Prior work has explored synthetic generation and simulation as alternative means of obtaining performance data; however, their benefits when used alongside real-world measurements are less understood. Using the saturation predictor Monitorless in a case study, we evaluate two scenarios. Our results indicate that generated and simulated applications improve predictive performance on previously unseen, real-world systems. In our case, generated applications reduced false positives, simulated applications enhanced robustness, while their combination yielded the most balanced predictor. The findings suggest that synthetic and simulated applications can effectively complement real-world data, enabling more diverse and robust evaluation of performance models.

Computational analysis of a large number of genomes requires a data structure that can represent the dataset compactly while also enabling efficient operations on variants and samples. However, encoding genetic data in existing tabular data structures and file formats has become costly and unsustainable. Here we introduce the genotype representation graph (GRG), a fully connected hierarchical data structure that losslessly encodes phased whole-genome polymorphisms. Exploiting variant-sharing across samples enables GRG to compress 200,000 UK Biobank phased human genomes to 5--26thinspacegigabytes per chromosome, also enabling graph-traversal algorithms to reuse computed values in random access memory. Constructing and processing GRG files scales to a million whole genomes. Using allele frequencies and association effects as examples, we show that computation on GRG via graph traversal runs the fastest among all tested alternatives. GRG-based algorithms have the potential to increase the scalability and reduce the cost of analyzing large genomic datasets.

Variant Call Format (VCF) is the standard file format for interchanging genetic variation data and associated quality control metrics. The usual row-wise encoding of the VCF data model (either as text or packed binary) emphasizes efficient retrieval of all data for a given variant, but accessing data on a field or sample basis is inefficient. The Biobank-scale datasets currently available consist of hundreds of thousands of whole genomes and hundreds of terabytes of compressed VCF. Row-wise data storage is fundamentally unsuitable and a more scalable approach is needed.Zarr is a format for storing multidimensional data that is widely used across the sciences, and is ideally suited to massively parallel processing. We present the VCF Zarr specification, an encoding of the VCF data model using Zarr, along with fundamental software infrastructure for efficient and reliable conversion at scale. We show how this format is far more efficient than standard VCF-based approaches, and competitive with specialized methods for storing genotype data in terms of compression ratios and single-threaded calculation performance. We present case studies on subsets of 3 large human datasets (Genomics England: \\($n\$\)=78,195; Our Future Health: \\($n\$\)=651,050; All of Us: \\($n\$\)=245,394) along with whole genome datasets for Norway Spruce (\\($n\$\)=1,063) and SARS-CoV-2 (\\($n\$\)=4,484,157). We demonstrate the potential for VCF Zarr to enable a new generation of high-performance and cost-effective applications via illustrative examples using cloud computing and GPUs.Large row-encoded VCF files are a major bottleneck for current research, and storing and processing these files incurs a substantial cost. The VCF Zarr specification, building on widely used, open-source technologies, has the potential to greatly reduce these costs, and may enable a diverse ecosystem of next-generation tools for analysing genetic variation data directly from cloud-based object stores, while maintaining compatibility with existing file-oriented workflows.

Die IG Metall schlage eine Vier-Tage-Woche vor - und spreche lediglich "von einem gewissen Lohnausgleich". Ist das realistisch? Arbeitsreduzierung habe schon einmal gut bei fast vollem Lohnausgleich funktioniert: bei VW. Vor 25 Jahren habe die Vier-Tage-Woche Arbeitsplätze gerettet und die Lebensqualität verbessert.

Wireless communication is the most effective communication to convey audio or video information among the population. It enables the masses to connect throughout the world. Wireless technologies improve the lifestyle of individuals in rural and poor communication areas. In this view, the quality of a reliable signal can be enhanced by minimizing carrier interference. In this paper bit error rate of an image, signal is transmitted over fading channel is analyzed using orthogonal frequency multiplexing and channel estimation technique. An Orthogonal Frequency Multiplexing (OFDM) provides prominent bandwidth effectiveness and improved immunity to the fading environments. In OFDM, the data is modulated using multiple numbers of subcarriers that are orthogonal to each other. A cyclic prefix is infixed between OFDM symbols to annihilate the inter symbol interference (ISI) and inter-carrier interference (ICI). The least square channel estimation method is used to minimize the effect of multipath fading. An image, signal is modulated using BPSK, QPSK, 16QAM and 64QAM digital modulation schemes with OFDM and channel estimation and transmitted over AWGN and fading channel. The objective of this work is to improve the signal to noise ratio by reducing interference. The bit error rate vs. signal to noise ratio for BPSK, QPSK, 16QAM and 64QAM without channel estimation at 5dB is 0.4948, 0.4987, 0.4965 and 0.4983 and with channel estimation is 0.099, 0.2600, 0.3900 and 0.4300 respectively. The bit error rate obtained with BPSK, QPSK, 16QAM and 64QAM without channel estimation at SNR of 10dB is 0.4964, 0.4985, 0.4957 and 0.4982 and with channel estimation is 0.033, 0.19, 0.34 and 0.38 respectively. The bit error rate obtained with BPSK, QPSK, 16QAM and 64QAM without channel estimation at SNR of 15dB is 0.4938, 0.4985, 0.4953 and 0.4979 and with channel estimation is 0.0072, 0.1900, 0.3241 and 0.3762 respectively. The error rate is minimized with channel estimation. The error rate increases with the order of modulation and it is noticed that the error rate is minimum with the BPSK modulator and is maximum with 64QAM.

Die IG Metall schlage eine Vier-Tage-Woche vor - und spreche lediglich "von einem gewissen Lohnausgleich". Ist das realistisch? Arbeitsreduzierung habe schon einmal gut bei fast vollem Lohnausgleich funktioniert: bei VW. Vor 25 Jahren habe die Vier-Tage-Woche Arbeitsplätze gerettet und die Lebensqualität verbessert.

Die IG Metall schlage eine Vier-Tage-Woche vor - und spreche lediglich "von einem gewissen Lohnausgleich". Ist das realistisch? Arbeitsreduzierung habe schon einmal gut bei fast vollem Lohnausgleich funktioniert: bei VW. Vor 25 Jahren habe die Vier-Tage-Woche Arbeitsplätze gerettet und die Lebensqualität verbessert.

Success of any software system largely looms upon its vigilance efficiency that prompts organizations to meet the set of objectives in the arena of networks. In the highly competitive world, everything appears to be vulnerable; information system is also not an exception to this fact. The security of information system has become a cause of great concern. On the contrary, till time the software security engineers are trying hard to develop fully protected and highly secured information systems but all these developments are at nascent stages. It is quite revelling that in the earlier research studies, little attention is paid to highlight an accurate status of the security alertness for developed software. Hence, keeping all these factors at the backdrop, this paper is an attempt to propose a holistic Security Maturity Model (SMM), in which five levels/stars have been developed, driven on the strength of the security vigilance occurring at the various stages for any software. SMM is in its conceptual stage; the detailed steps will certainly require time to be developed so that every software system can reap out the benefits of this model. To categorize/discriminate the level of potency, SMM will be highlighted through appropriate ranking/star system. It is hoped that if SMM will be followed in its true letter and sprit; undoubtedly, this will restore the clients’ trust and confidence on the software as well as their corresponding vendors. Moreover, this will also enable software industry to follow transparent and ethical practices.