Evaluating alignment in language models requires testing how they behave under realistic pressure, not just what they claim they would do. While alignment failures increasingly cause real-world harm, comprehensive evaluation frameworks with realistic multi-turn scenarios remain lacking. We introduce an alignment benchmark spanning 904 scenarios across six categories -- Honesty, Safety, Non-Manipulation, Robustness, Corrigibility, and Scheming -- validated as realistic by human raters. Our scenarios place models under conflicting instructions, simulated tool access, and multi-turn escalation to reveal behavioural tendencies that single-turn evaluations miss. Evaluating 24 frontier models using LLM judges validated against human annotations, we find that even top-performing models exhibit gaps in specific categories, while the majority of models show consistent weaknesses across the board. Factor analysis reveals that alignment behaves as a unified construct (analogous to the g-factor in cognitive research) with models scoring high on one category tending to score high on others. We publicly release the benchmark and an interactive leaderboard to support ongoing evaluation, with plan
Petri net unfoldings are a useful tool to tackle state-space explosion in verification and related tasks. Moreover, their structure allows to access directly the relations of causal precedence, concurrency, and conflict between events. Here, we explore the data structure further, to determine the following relation: event a is said to reveal event b iff the occurrence of a implies that b inevitably occurs, too, be it before, after, or concurrently with a. Knowledge of reveals facilitates in particular the analysis of partially observable systems, in the context of diagnosis, testing or verification; it can also be used to generate more concise representations of behaviours via abstractions. The reveals relation was previously introduced in the context of fault diagnosis, where it was shown that the reveals relation was decidable: for a given pair a,b in the unfolding U of a safe Petri net N, a finite prefix P of U is sufficient to decide whether or not a reveals b. In this paper, we first considerably improve the bound on |P|. We then show that there exists an efficient algorithm for computing the relation on a given prefix. We have implemented the algorithm and report on experimen
The concept of ergotropy was previously introduced as the maximum extractable work from a quantum state. Its enhancement, which is induced by quantum correlation via projective measurement, was formulated as the daemonic ergotropy. In this work, we investigate the ergotropy in the presence of quantum correlation via weak measurement because of its elegant effects on the measured system. By considering a bipartite correlated quantum system consisting of main and ancillary systems, we demonstrate that the extractable work by the non-selective weak measurement on the ancilla is always equal to the situation captured by the strong measurement. However, the selective weak measurement interestingly reveals more work than the daemonic ergotropy and the ergotropy of the total system is greater than or equal to the daemonic ergotropy. Moreover, it is shown that for Bell diagonal states, at the cost of losing quantum correlation, the total extractable and thus non-local extractable works can be increased by using measurement. Also, we find that there is no direct relationship between quantum correlation and non-local extractable work for these cases.
Normal molecular dynamics simulations are usually unable to simulate chemical reactions due to the low probability of forming the transition state. Therefore, enhanced sampling methods are implemented to accelerate the occurrence of chemical reactions. In this investigation, we present an application of metadynamics in simulating an organic multi-step cascade reaction. The analysis of the reaction trajectory reveals the barrier heights of both forward and reverse reactions. We also present a discussion of the advantages and disadvantages of generating the reactive pathway using molecular dynamics and the intrinsic reaction coordinate (IRC) algorithm.
Correlations of partitioned particles carry essential information about their quantumness. Partitioning full beams of charged particles leads to current fluctuations, with their autocorrelation (namely, shot noise) revealing the particle' charge. This is not the case when the partitioned particle beams are diluted. Bosons or fermions will exhibit particles antibunching (due to their sparsity and discreteness). However, when diluted anyons, such as the quasiparticles in fractional quantum Hall states, are partitioned in a narrow constriction, their autocorrelation reveals an essential aspect of their exchange statistics: their braiding phase. Here, we describe detailed measurements of weak partitioned, highly diluted, one-dimension-like edge modes of the one-third filling fractional quantum Hall state. The measured autocorrelation agrees with our theory of braiding anyons in the time-domain (instead of braiding in space); with a braiding phase 2$θ$=2$π$/3, without any fitting parameters. Our work offers a relatively straightforward and simple method to observe the braiding statistics of other exotic anyonic states, such as non-abelian states, without resorting to complex interferenc
The behaviour of classical mechanical systems is characterised by their phase portraits, the collections of their trajectories. Heisenberg's uncertainty principle precludes the existence of sharply defined trajectories, which is why traditionally only the time evolution of wave functions is studied in quantum dynamics. These studies are quite insensitive to the underlying structure of quantum phase space dynamics. We identify the flow that is the quantum analog of classical particle flow along phase portrait lines. It reveals hidden features of quantum dynamics and extra complexity. Being constrained by conserved flow winding numbers, it also reveals fundamental topological order in quantum dynamics that has so far gone unnoticed.
An exceptionally preserved snake fossil from Brazil reveals that early snakes were far more diverse than scientists once thought。 Detailed brain reconstructions show that Tametara mirim was adapted for burrowing, while another ancient snake was suited to life above ground。 Other fossils point to marine lifestyles as well
Metabolic flux balance analyses are a standard tool in analysing metabolic reaction rates compatible with measurements, steady-state and the metabolic reaction network stoichiometry. Flux analysis methods commonly place unrealistic assumptions on fluxes due to the convenience of formulating the problem as a linear programming model, and most methods ignore the notable uncertainty in flux estimates. We introduce a novel paradigm of Bayesian metabolic flux analysis that models the reactions of the whole genome-scale cellular system in probabilistic terms, and can infer the full flux vector distribution of genome-scale metabolic systems based on exchange and intracellular (e.g. 13C) flux measurements, steady-state assumptions, and target function assumptions. The Bayesian model couples all fluxes jointly together in a simple truncated multivariate posterior distribution, which reveals informative flux couplings. Our model is a plug-in replacement to conventional metabolic balance methods, such as flux balance analysis (FBA). Our experiments indicate that we can characterise the genome-scale flux covariances, reveal flux couplings, and determine more intracellular unobserved fluxes in
The understanding of synchronization ranging from natural to social systems has driven the interests of scientists from different disciplines. Here, we have investigated the synchronization dynamics of the Kuramoto dynamics departing from the fully synchronized regime. We have got the analytic expression of the dynamical correlation between pairs of oscillators that reveals the relation between the network dynamics and the underlying topology. Moreover, it also reveals the internal structure of networks that can be used as a new algorithm to detect community structures. Further, we have proposed a new measure about the synchronization in complex networks and scrutinize it in small-world and scale-free networks. Our results indicate that the more heterogeneous and "smaller" the network is, the more closely it would be synchronized by the collective dynamics.
Accessing unexplored conditions in crystal growth often reveals remarkable surprises and new regimes of physical behavior. In this work, performing molecular beam epitaxy of the technologically important superconductor NbN at temperatures greater than 1000$^\circ$C, higher than in the past, is found to reveal persistent RHEED oscillations throughout the growth, atomically smooth surfaces, normal metal resistivities as low as 37$μΩ$-cm and superconducting critical temperatures in excess of 15 K. Most remarkably, a reversal of the sign of the Hall coefficient is observed as the NbN films are cooled, and the high material quality allows the first imaging of Abrikosov vortex lattices in this superconductor.
Movement is a fundamental behaviour of organisms that brings about beneficial encounters with resources and mates, but at the same time exposes the organism to dangerous encounters with predators. The movement patterns adopted by organisms should reflect a balance between these contrasting processes. This trade-off can be hypothesized as being evident in the behaviour of plankton, which inhabit a dilute 3D environment with few refuges or orienting landmarks. We present an analysis of the swimming path geometries based on a volumetric Monte Carlo sampling approach, which is particularly adept at revealing such trade-offs by measuring the self-overlap of the trajectories. Application of this method to experimentally measured trajectories reveals that swimming patterns in copepods are shaped to efficiently explore volumes at small scales, while achieving a large overlap at larger scales. Regularities in the observed trajectories make the transition between these two regimes always sharper than in randomized trajectories or as predicted by random walk theory. Thus real trajectories present a stronger separation between exploration for food and exposure to predators. The specific scale
The London moment is the magnetic moment acquired by a rotating superconductor. We propose that the London moment reveals the following fundamental properties of the superconducting state: (i) superconductors (unlike normal metals) know the $sign$ of the charge carriers, (ii) the superconducting charge carriers are $free$ electrons, (iii) electrons are expelled from the interior to the surface in the transition to the superconducting state, (iv) superfluid electrons occupy orbits of radius $2λ_L$ ($λ_L=$London penetration depth), and (v) a spin current exists in the ground state of superconductors. These properties are consistent with the Meissner effect, however the Meissner effect does not $directly$ reveal the sign of the charge carriers nor the fact that the carrier's mass is the free electron mass nor the fact that a spin current exists in superconductors. Note also that within the BCS theory of superconductivity none of the key properties of superconductors listed above are predicted. Instead, these properties are predicted by the theory of hole superconductivity.
We build upon a simple micro-founded model of asset trading proposed by Kyle (1985) to study under what conditions a trader who is privately informed of the future return of the asset may want to share her information with other traders. Despite what conventional wisdom suggests, we show that in the unique equilibrium of the game the informed trader reveals her information with positive probability. A consequence of it is that, in contrast with the corresponding no-communication benchmark, the equilibrium price need not be fully revealing of the asset's return, even if traders are risk neutral. This, in turn, has significant implications on the distribution of the social surplus. While our model initially assumes that inter-agent communication is restricted by an arbitrarily given social network, we also study which such networks arise when links are endogenously formed through traders' prior connection decisions.
Masked diffusion models (MDMs), which leverage bidirectional attention and a denoising process, are narrowing the performance gap with autoregressive models (ARMs). However, their internal attention mechanisms remain under-explored. This paper investigates the attention behaviors in MDMs, revealing the phenomenon of Attention Floating. Unlike ARMs, where attention converges to a fixed sink, MDMs exhibit dynamic, dispersed attention anchors that shift across denoising steps and layers. Further analysis reveals its Shallow Structure-Aware, Deep Content-Focused attention mechanism: shallow layers utilize floating tokens to build a global structural framework, while deeper layers allocate more capability toward capturing semantic content. Empirically, this distinctive attention pattern provides a mechanistic explanation for the strong in-context learning capabilities of MDMs, allowing them to double the performance compared to ARMs in knowledge-intensive tasks. All codes and datasets are available at https://github.com/NEUIR/Attention-Floating.
We present a detailed, time-resolved analysis of the Fe K band of the Seyfert 1.5 galaxy NGC 3516 observed with XRISM. The 249 ks observation spanning $\sim$310 ks in elapsed time reveals an exceptionally rich and time-variable absorption spectrum. Six distinct absorption components are detected across multiple ionization states, spanning more than an order of magnitude in ionization parameter and a wide range of systemic velocities, from a potential inflow ($+4300~\rm km~s^{-1}$) to a mildly relativistic ultra-fast outflow ($-9800~\rm km~s^{-1}$). Despite their diversity, the components exhibit relatively small broadening ($\lesssim$$400~\rm km~s^{-1}$), implying comparable internal dynamics within a medium of a complex structure. Time-resolved spectroscopy reveals pronounced variability in three highly ionized absorbers, with Fe XXV$-$Fe XXVI features that appear and disappear on timescales of tens of kiloseconds. This behavior likely reflects a combination of geometrical transits of clumpy gas and ionization-state changes driven by continuum variability. An additional temporary absorption feature in the red wing of the Fe K$α$ line, consistent with Fe XXV absorption, indicates a
Consumer heterogeneity in revealed-preference data is larger than bilateral rationality tests can reveal. We construct a continuous nonparametric metric of this hidden heterogeneity by repeatedly subsampling choices, partitioning agents into groups whose pooled data are jointly rationalisable under a chosen consistency criterion and recording how often each pair is co-classified. The resulting kernel is positive semi-definite, embeds the population in a Hilbert space, and induces a metric with the triangle inequality. Under a necessary-and-sufficient contrast-rank condition, its spectral structure recovers latent preference types. Inference on demographic correlates proceeds via a Monte-Carlo-conditional test and a finite-sample-valid permutation test. Applied to US grocery scanner data, the construction reveals a joint-rationality gap of 0.62 between near-saturated pairwise compatibility and population-level co-typing; binary lottery data yield a comparable gap of 0.38. Standard demographics organise only a modest part of the scanner kernel structure.
We discuss a randomized strong rank-revealing QR factorization that effectively reveals the spectrum of a matrix $\textbf{M}$. This factorization can be used to address problems such as selecting a subset of the columns of $\textbf{M}$, computing its low-rank approximation, estimating its rank, or approximating its null space. Given a random sketching matrix $\pmbΩ$ that satisfies the $ε$-embedding property for a subspace within the range of $\textbf{M}$, the factorization relies on selecting columns that allow to reveal the spectrum via a deterministic strong rank-revealing QR factorization of $\textbf{M}^{sk} = \pmbΩ\textbf{M}$, the sketch of $\textbf{M}$. We show that this selection leads to a factorization with strong rank-revealing properties, making it suitable for approximating the singular values of $\textbf{M}$.
This study investigates the mechanisms of Surveillance Capitalism, focusing on personal data transfer during web navigation and searching. Analyzing network traffic reveals how various entities track and harvest digital footprints. The research reveals specific data types exchanged between users and web services, emphasizing the sophisticated algorithms involved in these processes. We present concrete evidence of data harvesting practices and propose strategies for enhancing data protection and transparency. Our findings highlight the need for robust data protection frameworks and ethical data usage to address privacy concerns in the digital age.
Simple commit-reveal beacons are vulnerable to last-revealer strategies, and existing descriptions often leave accountability and recovery mechanisms unspecified for practical deployments. We present Commit-Reveal$^2$, a layered design for blockchain deployments that cryptographically randomizes the final reveal order, together with a concrete accountability and fallback mechanism that we implement as smart-contract logic. The protocol is architected as a hybrid system, where routine coordination runs off chain for efficiency and the blockchain acts as the trust anchor for commitments and the final arbiter for disputes. Our implementation covers leader coordination, on-chain verification, slashing for non-cooperation, and an explicit on-chain recovery path that maintains progress when off-chain coordination fails. We formally define two security goals for distributed randomness beacons, unpredictability and bit-wise bias resistance, and we show that Commit-Reveal$^2$ meets these notions under standard hash assumptions in the random-oracle model. In measurements with small to moderate operator sets, the hybrid design reduces on-chain gas by more than 80% compared to a fully on-chain
Energy transfer across scales is fundamental in fluid dynamics, linking large-scale flow motions to small-scale turbulent structures in engineering and natural environments. Triadic interactions among three wave components form complex networks across scales, challenging understanding and model reduction. We introduce Triadic Orthogonal Decomposition (TOD), a method that identifies coherent flow structures optimally capturing spectral momentum transfer, quantifies their coupling and energy exchange in an energy budget bispectrum, and reveals the regions where they interact. TOD distinguishes three components--a momentum recipient, donor, and catalyst--and recovers laws governing pairwise, six-triad, and global triad conservation. Applied to unsteady cylinder wake and wind turbine wake data, TOD reveals networks of triadic interactions with forward and backward energy transfer across frequencies and scales.