From the onset of fundamental statistical mechanical constructs formulated in the late 19th century, alchemical free-energy methods slowly emerged and transitioned to become operational tools of biomolecular simulation applicable to a wide range of problems including protein-ligand binding for drug discovery research. This article reconstructs how statistical mechanical approaches such as thermodynamic integration and free-energy perturbation were reconfigured in the early 1980's to address the complexities of increasingly heterogeneous biomolecular systems. Drawing on oral history interviews and primary literature, the study examines the technical, institutional, theoretical, and infrastructural conditions under which these methods were implemented, and became progressively operational. These conditions encompassed the consolidation of lab-specific software infrastructures, the formulation of practical simulation protocols, as well as essential statistical mechanical clarifications. From this perspective, the progress of free-energy methods proceeded less from a unified convergence than from an iterative troubleshooting process of alignment involving practical and theoretical considerations. The aim of the present article is to offer a historically grounded account of how free-energy techniques acquired practical and functional reliability.

A central challenge in cognitive neuroscience is to explain how semantic and episodic memory, two major forms of declarative memory, typically associated with cortical and hippocampal processing, interact to support learning, recall, and imagination. Despite significant advances, we still lack a unified computational framework that jointly accounts for core empirical phenomena across both semantic and episodic processing domains. Here, we introduce the Generative Episodic-Semantic Integration System (GENESIS), a computational model that formalizes memory as the interaction between two limited-capacity generative systems: a Cortical-VAE, supporting semantic learning and generalization, and a Hippocampal-VAE, supporting episodic encoding and retrieval within a retrieval-augmented generation (RAG) architecture. GENESIS reproduces hallmark behavioral findings, including generalization in semantic memory, recognition, serial recall effects and gist-based distortions in episodic memory, and constructive episodic simulation, while capturing their dynamic interactions. The model elucidates how capacity constraints shape the fidelity and memorability of experiences, how semantic processing introduces systematic distortions in episodic recall, and how episodic replay can recombine previous experiences. Together, these results provide a principled account of memory as an active, constructive, and resource-bounded process. GENESIS thus advances a unified theoretical framework that bridges semantic and episodic memory, offering new insights into the generative foundations of human cognition.

This paper develops a risk-adjusted alternative to standard optimal policy learning (OPL) for observational data by importing Roy's (1952) safety-first principle into the treatment assignment problem. We formalize a welfare functional that maximizes the probability that outcomes exceed a socially required threshold and show that the associated pointwise optimal rule ranks treatments by the ratio of conditional means to conditional standard deviations. We implement the framework using microdata from the Italian Farm Accountancy Data Network to evaluate the allocation of subsidies under the EU Common Agricultural Policy. Empirically, risk-adjusted optimal policies systematically dominate the realized allocation across specifications, while risk aversion lowers overall welfare relative to the risk-neutral benchmark, making transparent the social cost of insurance against uncertainty. The results illustrate how safety-first OPL provides an implementable, interpretable tool for risk-sensitive policy design, quantifying the efficiency-insurance trade-off that policymakers face when outcomes are volatile.

The Industrial Internet of Things (IIoT) is a transformative paradigm that integrates smart sensors, advanced analytics, and robust connectivity within industrial processes, enabling real-time data-driven decision-making and enhancing operational efficiency across diverse sectors, including manufacturing, energy, and logistics. IIoT is susceptible to various attack vectors, with Advanced Persistent Threats (APTs) posing a particularly grave concern due to their stealthy, prolonged, and targeted nature. The effectiveness of machine learning-based intrusion detection systems in APT detection has been documented in the literature. However, existing cybersecurity datasets often lack crucial attributes for APT detection in IIoT environments. Incorporating insights from prior research on APT detection using provenance data and intrusion detection within IoT systems, we present the CICAPT-IIoT dataset. The main goal of this paper is to propose a novel APT dataset in the IIoT setting that includes essential information for the APT detection task. In order to achieve this, a testbed for IIoT is developed, and over 20 attack techniques frequently used in APT campaigns are included. The performed attacks create some of the invariant phases of the APT cycle, including Data Collection and Exfiltration, Discovery and Lateral Movement, Defense Evasion, and Persistence. By integrating network logs and provenance logs with detailed attack information, the CICAPT-IIoT dataset presents foundation for developing holistic cybersecurity measures. Additionally, a comprehensive dataset analysis is provided, presenting cybersecurity experts with a strong basis on which to build innovative and efficient security solutions.

A lot of recent machine learning research papers have ``open-ended learning'' in their title. But very few of them attempt to define what they mean when using the term. Even worse, when looking more closely there seems to be no consensus on what distinguishes open-ended learning from related concepts such as continual learning, lifelong learning or autotelic learning. In this paper, we contribute to fixing this situation. After illustrating the genealogy of the concept and more recent perspectives about what it truly means, we outline that open-ended learning is generally conceived as a composite notion encompassing a set of diverse properties. In contrast with previous approaches, we propose to isolate a key elementary property of open-ended processes, which is to produce elements from time to time (e.g., observations, options, reward functions, and goals), over an infinite horizon, that are considered novel from an observer's perspective. From there, we build the notion of open-ended learning problems and focus in particular on the subset of open-ended goal-conditioned reinforcement learning problems in which agents can learn a growing repertoire of goal-driven skills. Finally, we highlight the work that remains to be performed to fill the gap between our elementary definition and the more involved notions of open-ended learning that developmental AI researchers may have in mind.

Regular expression (RE) matching is a very common functionality that scans a text to find occurrences of patterns specified by an RE; it includes the simpler function of RE recognition. Here we address RE parsing, which subsumes matching by providing not just the pattern positions in the text, but also the syntactic structure of each pattern occurrence, in the form of a tree representing how the RE operators produced the patterns. RE parsing increases the selectivity of matching, yet avoiding the complications of context-free grammar parsers. Our parser manages ambiguous REs and texts by returning the set of all syntax trees, compressed into a Shared-Packed-Parse-Forest data-structure. We initially convert the RE into a serial parser, which simulates a finite automaton (FA) so that the states the automaton passes through encode the syntax tree of the input. On long texts, serial matching and parsing may be too slow for time-constrained applications. Therefore, we present a novel efficient parallel parser for multi-processor computing platforms; its speed-up over the serial algorithm scales well with the text length. We innovatively apply to RE parsing the approach typical of parallel RE matchers / recognizers, where the text is split into chunks to be parsed in parallel and then joined together. Such an approach suffers from the so-called speculation overhead, due to the lack of knowledge by a chunk processor about the state reached at the end of the preceding chunk; this forces each chunk processor to speculatively start in all its states. We introduce a novel technique that minimizes the speculation overhead. The multi-threaded parser program, written in Java, has been validated and its performance has been measured on a commodity multi-core computer, using public and synthetic RE benchmarks. The speed-up over serial parsing, parsing times, and parser construction times are reported.

Automatic Readability Assessment (ARA), the task of assigning a reading level to a text, is traditionally treated as a classification problem in NLP research. In this paper, we propose the first neural, pairwise ranking approach to ARA and compare it with existing classification, regression, and (non-neural) ranking methods. We establish the performance of our model by conducting experiments with three English, one French and one Spanish datasets. We demonstrate that our approach performs well in monolingual single/cross corpus testing scenarios and achieves a zero-shot cross-lingual ranking accuracy of over 80% for both French and Spanish when trained on English data. Additionally, we also release a new parallel bilingual readability dataset in English and French. To our knowledge, this paper proposes the first neural pairwise ranking model for ARA, and shows the first results of cross-lingual, zero-shot evaluation of ARA with neural models.

The study of spin-glass dynamics, long considered the paradigmatic complex system, has reached important milestones. The availability of single crystals has allowed the experimental measurement of spin-glass coherence lengths of almost macroscopic dimensions, while the advent of special-purpose computers enables dynamical simulations that approach experimental scales. This review provides an account of the quantitative convergence of these two avenues of research, with precise experimental measurements of the expected scaling laws and numerical reproduction of classic experimental results, such as memory and rejuvenation. The article opens with a brief review of the defining spin-glass properties, randomness and frustration, and their experimental consequences. These apparently simple characteristics are shown to generate rich and complex physics. Models are introduced that enable quantitative dynamical descriptions. After a summary of the main numerical results in equilibrium, paying particular attention to temperature chaos, this review examines off-equilibrium dynamics in the absence of a magnetic field and shows how it can be related to equilibrium structures through the fluctuation-dissipation relations. The nonlinear response at a given temperature is then developed, including experiments and scaling in the vicinity of the transition temperature $T_\mathrm{g}$. The consequences of temperature change $\unicode{x2013}$including temperature chaos, rejuvenation, and memory$\unicode{x2013}$ are reviewed. The interpretation of these phenomena requires identifying several length scales relevant to dynamics, which, in turn, generate new insights. Finally, issues for future investigations are introduced, including what is to be nailed down theoretically, why the Ising Edwards-Anderson model is so successful at modeling spin-glass dynamics, and experiments yet to be undertaken.

This roadmap consolidates recent advances while exploring emerging applications, reflecting the remarkable diversity of hardware platforms, neuromorphic concepts, and implementation philosophies reported in the field. It emphasizes the critical role of cross-disciplinary collaboration in this rapidly evolving field.

We analyze the recently observed breakdown of the integer quantum Hall effect in a two-dimensional electron gas embedded in a metallic split-ring resonator. By accounting for both the quantized vacuum field and electrostatic boundary modifications, we identify a mechanism that could potentially explain this breakdown in terms of non-chiral edge channels arising from electrostatic boundary effects. For experimentally relevant geometries, a minimal single-electron model of this mechanism predicts characteristic signatures and energy scales consistent with those observed in experiments. These predictions can be directly tested against alternative, purely vacuum-induced explanations to shed further light on the origin of this puzzling phenomenon.

Glasses are amorphous solids whose constituent particles are caged by their neighbors and thus cannot flow. This sluggishness is often ascribed to the free energy landscape containing multiple minima (basins) separated by high barriers. Here we show, using theory and numerical simulation, that the landscape is much rougher than is classically assumed. Deep in the glass, it undergoes a "roughness transition" to fractal basins. This brings about isostaticity at jamming and marginality of glassy states near jamming. Critical exponents for the basin width, the weak force distribution, and the spatial spread of quasi-contacts at jamming can be analytically determined. Their value is found to be compatible with numerical observations. This advance therefore incorporates the jamming transition of granular materials into the framework of glass theory. Because temperature and pressure control which features of the landscape are experienced, glass mechanics and transport are expected to reflect the features of the topology we discuss here. Hitherto mysterious properties of low-temperature glasses could be explained by this approach.

Optimizing highly complex cost/energy functions over discrete variables is at the heart of many open problems across different scientific disciplines and industries. A major obstacle is the emergence of many-body effects among certain subsets of variables in hard instances leading to critical slowing down or collective freezing for known stochastic local search strategies. An exponential computational effort is generally required to unfreeze such variables and explore other unseen regions of the configuration space. Here, we introduce a quantum-inspired family of nonlocal Nonequilibrium Monte Carlo (NMC) algorithms by developing an adaptive gradient-free strategy that can efficiently learn key instance-wise geometrical features of the cost function. That information is employed on-the-fly to construct spatially inhomogeneous thermal fluctuations for collectively unfreezing variables at various length scales, circumventing costly exploration versus exploitation trade-offs. We apply our algorithm to two of the most challenging combinatorial optimization problems: random k-satisfiability (k-SAT) near the computational phase transitions and Quadratic Assignment Problems (QAP). We observe significant speedup and robustness over both specialized deterministic solvers and generic stochastic solvers. In particular, for 90% of random 4-SAT instances we find solutions that are inaccessible for the best specialized deterministic algorithm known as Survey Propagation (SP) with an order of magnitude improvement in the quality of solutions for the hardest 10% instances. We also demonstrate two orders of magnitude improvement in time-to-solution over the state-of-the-art generic stochastic solver known as Adaptive Parallel Tempering (APT).

The modern technological landscape has trended towards increased precision and greater digitization of information. However, the methods used to record and communicate scientific procedures have remained largely unchanged over the last century. Written text as the primary means for communicating scientific protocols poses notable limitations in human and machine information transfer. In this work, we present the Universal Workflow Language (UWL) and the open-source Universal Workflow Language interface (UWLi). UWL is a graph-based data architecture that can capture arbitrary scientific procedures through workflow representation of protocol steps and embedded procedure metadata. It is machine readable, discipline agnostic, and compatible with FAIR reporting standards. UWLi is an accompanying software package for building and manipulating UWL files into tabular and plain text representations in a controlled, detailed, and multilingual format. UWL transcription of protocols from three high-impact publications resulted in the identification of substantial deficiencies in the detail of the reported procedures. UWL transcription of these publications identified seventeen procedural ambiguities and thirty missing parameters for every one hundred words in published procedures. In addition to preventing and identifying procedural omission, UWL files were found to be compatible with geometric learning techniques for representing scientific protocols. In a surrogate function designed to represent an arbitrary multi-step experimental process, graph transformer networks were able to predict outcomes in approximately 6,000 fewer experiments than equivalent linear models. Implementation of UWL and UWLi into the scientific reporting process will result in higher reproducibility between both experimentalists and machines, thus proving an avenue to more effective modeling and control of complex systems.

Recommender systems continuously interact with users, creating feedback loops that shape both individual behavior and collective market dynamics. This paper introduces a simulation framework to model these loops in online retail environments, where recommenders are periodically retrained on evolving user-item interactions. Using the Amazon e-Commerce dataset, we analyze how different recommendation algorithms influence diversity, purchase concentration, and user homogenization over time. Results reveal a systematic trade-off: while the feedback loop increases individual diversity, it simultaneously reduces collective diversity and concentrates demand on a few popular items. Moreover, for some recommender systems, the feedback loop increases user homogenization over time, making user purchase profiles increasingly similar. These findings underscore the need for recommender designs that balance personalization with long-term diversity.

Query recommendation systems are ubiquitous in modern search engines, assisting users in producing effective queries to meet their information needs. However, these systems require a large amount of data to produce good recommendations, such as a large collection of documents to index and query logs. In particular, query logs and user data are not available in cold start scenarios. Query logs are expensive to collect and maintain and require complex and time-consuming cascading pipelines for creating, combining, and ranking recommendations. To address these issues, we frame the query recommendation problem as a generative task, proposing a novel approach called Generative Query Recommendation (GQR). GQR uses an LLM as its foundation and does not require to be trained or fine-tuned to tackle the query recommendation problem. We design a prompt that enables the LLM to understand the specific recommendation task, even using a single example. We then improved our system by proposing a version that exploits query logs called Retriever-Augmented GQR (RA-GQR). RA-GQr dynamically composes its prompt by retrieving similar queries from query logs. GQR approaches reuses a pre-existing neural architecture resulting in a simpler and more ready-to-market approach, even in a cold start scenario. Our proposed GQR obtains state-of-the-art performance in terms of NDCG@10 and clarity score against two commercial search engines and the previous state-of-the-art approach on the Robust04 and ClueWeb09B collections, improving on average the NDCG@10 performance up to ~4% on Robust04 and ClueWeb09B w.r.t the previous best competitor. RA-GQR further improve the NDCG@10 obtaining an increase of ~11%, ~6\% on Robust04 and ClueWeb09B w.r.t the best competitor. Furthermore, our system obtained ~59% of user preferences in a blind user study, proving that our method produces the most engaging queries.

Mass currents in astrophysics generate gravitomagnetic fields of enormous complexity. Gravitomagnetic helicity, in direct analogy with magnetic helicity, is a measure of entwining of the gravitomagnetic field lines. We discuss gravitomagnetic helicity within the gravitoelectromagnetic (GEM) framework of linearized general relativity. Furthermore, we employ the spacetime curvature approach to GEM in order to determine the gravitomagnetic helicity for static observers in Kerr spacetime.

Magnetic resonance imaging (MRI) scanners have advanced significantly, with a growing use of highfield 3 T systems. This evolution gives rise to safety concerns for healthcare personnel working in proximity to MRI equipment. While manufacturers provide theoretical Gauss line projections, these are typically derived under ideal open-environment conditions and may not reflect real-world installations. For this reason, identical MRI models can produce markedly different fringe field distributions depending on shielding and room configurations. The present study proposes an experimental methodology for the mapping of the fringe magnetic field in the vicinity of three 3 T MRI scanners. Field measurements were interpolated to generate threedimensional magnetic field maps. A comparative analysis was conducted, which revealed notable differences among the scanners. These differences serve to highlight the influence of site-specific factors on magnetic field propagation.

Children with autism spectrum disorder (ASD) experience challenges in grasping social-emotional cues, which can result in difficulties in recognizing emotions and understanding and responding to social interactions. Social-emotional intervention is an effective method to improve emotional understanding and facial expression recognition among individuals with ASD. Existing work emphasizes the importance of personalizing interventions to meet individual needs and motivate engagement for optimal outcomes in daily settings. We design a social-emotional game for ASD children, which generates personalized stories by leveraging the current advancement of artificial intelligence. Via a co-design process with five domain experts, this work offers several design insights into developing future AI-enabled gamified systems for families with autistic children. We also propose a fine-tuned AI model and a dataset of social stories for different basic emotions.

We introduce TURB-Scalar, an open-access database comprising approximately $400$ uncorrelated snapshots of two-dimensional turbulent velocity and passive scalar fields, obtained from the turbulent inverse cascade regime. These data are generated through Direct Numerical Simulations (DNS) of the advection-diffusion equation for a passive scalar, $\theta$, with resolution $N=4096$. The database serves as a versatile benchmark for the development and testing of both physics-based and data-driven modeling approaches. The scalar field exhibits intermittent statistics with universal anomalous scaling, making TURB-Scalar a valuable resource for studying turbulent transport phenomena. The database is available at this http URL.