Search

The flexibility and the variety of computing resources offered by the cloud make it particularly attractive for executing user workloads. However, IaaS cloud environments pose non-trivial challenges in the case of workflow scheduling under deadlines and monetary cost constraints. Indeed, given the typical uncertain performance behavior of cloud resources, scheduling algorithms that assume deterministic execution times may fail, thus requiring probabilistic approaches. However, existing probabilistic algorithms are computationally expensive, mainly due to the greater complexity of the workflow scheduling problem in its probabilistic form, and they hardily scale with the size of the problem instance. In this article, we propose EPOSS, a novel workflow scheduling algorithm for IaaS cloud environments based on a probabilistic formulation. Our solution blends together the low execution latency of state-of-the-art scheduling algorithms designed for the case of deterministic execution times and the capability to enforce probabilistic constraints.Designed with computational efficiency in mind, EPOSS achieves one to two orders lower execution times in comparison with existing probabilistic schedulers. Furthermore, it ensures good scaling with respect to workflow size and number of heterogeneous virtual machine types offered by the IaaS cloud environment. We evaluated the benefits of our algorithm via an experimental comparison over a variety of workloads and characteristics of IaaS cloud environments.

The gravitational waveform emitted by a particle moving in a Schwarzschild geometry can be computed, in perturbation theory, as a double series expansion in the limit where distances are large with respect to the size of the horizon, $z=2M/r\ll 1$ (Post-Minkowskian approximation), but small with respect to the gravitational wavelength, $y=2{\rm i} \omega r\ll 1$ (Post-Newtonian approximation). While in the case of bounded systems the two expansions are linked, for scattering processes they describe complementary regions of spacetime. In this paper, we derive all order formulae for those two expansions of the waveform, where the entire dependence on the $y$ or $z$ variables is resummed under the assumption $\omega M\ll 1$. The results are based on a novel hypergeometric representation of the confluent Heun functions and provide a new derivation of the recently discovered Heun connection formulae. Similar results are found for the non-confluent Heun case. Finally, in the case of circular orbits we compute our formulae at order 30PN and show excellent agreement against solutions obtained via numerical integrations and the MST method., Comment: 21 pages + Appendix + References, 2 Figures, 4 Tables; ancillary files are present, containing two Mathematica codes and relative data that are imported within them

We tackle average-reward infinite-horizon POMDPs with an unknown transition model but a known observation model, a setting that has been previously addressed in two limiting ways: (i) frequentist methods relying on suboptimal stochastic policies having a minimum probability of choosing each action, and (ii) Bayesian approaches employing the optimal policy class but requiring strong assumptions about the consistency of employed estimators. Our work removes these limitations by proving convenient estimation guarantees for the transition model and introducing an optimistic algorithm that leverages the optimal class of deterministic belief-based policies. We introduce modifications to existing estimation techniques providing theoretical guarantees separately for each estimated action transition matrix. Unlike existing estimation methods that are unable to use samples from different policies, we present a novel and simple estimator that overcomes this barrier. This new data-efficient technique, combined with the proposed \emph{Action-wise OAS-UCRL} algorithm and a tighter theoretical analysis, leads to the first approach enjoying a regret guarantee of order $\mathcal{O}(\sqrt{T \,\log T})$ when compared against the optimal policy, thus improving over state of the art techniques. Finally, theoretical results are validated through numerical simulations showing the efficacy of our method against baseline methods.

A new method for the numerical solution of ODEs is presented. This approach is based on an approximate formulation of the Taylor methods that has a much easier implementation than the original Taylor methods, since only the functions in the ODEs, and not their derivatives, are needed, just as in classical Runge-Kutta schemes. Compared to Runge-Kutta methods, the number of function evaluations to achieve a given order is higher, however with the present procedure it is much easier to produce arbitrary high-order schemes, which may be important in some applications. In many cases the new approach leads to an asymptotically lower computational cost when compared to the Taylor expansion based on exact derivatives. The numerical results that are obtained with our proposal are satisfactory and show that this approximate approach can attain results as good as the exact Taylor procedure with less implementation and computational effort.

Multi-task learning is a paradigm that leverages information from related tasks to improve the performance of machine learning. Self-Admitted Technical Debt (SATD) are comments in the code that indicate not-quite-right code introduced for short-term needs, i.e., technical debt (TD). Previous research has provided evidence of a possible relationship between SATD and the existence of vulnerabilities in the code. In this work, we investigate if multi-task learning could leverage the information shared between SATD and vulnerabilities to improve the automatic detection of these issues. To this aim, we implemented VulSATD, a deep learner that detects vulnerable and SATD code based on CodeBERT, a pre-trained transformers model. We evaluated VulSATD on MADE-WIC, a fused dataset of functions annotated for TD (through SATD) and vulnerability. We compared the results using single and multi-task approaches, obtaining no significant differences even after employing a weighted loss. Our findings indicate the need for further investigation into the relationship between these two aspects of low-quality code. Specifically, it is possible that only a subset of technical debt is directly associated with security concerns. Therefore, the relationship between different types of technical debt and software vulnerabilities deserves future exploration and a deeper understanding.

Developers rely on code comments to document their work, track issues, and understand the source code. As such, comments provide valuable insights into developers' understanding of their code and describe their various intentions in writing the surrounding code. Recent research leverages natural language processing and deep learning to classify comments based on developers' intentions. However, such labelled data are often imbalanced, causing learning models to perform poorly. This work investigates the use of different weighting strategies of the loss function to mitigate the scarcity of certain classes in the dataset. In particular, various RoBERTa-based transformer models are fine-tuned by means of a hyperparameter search to identify their optimal parameter configurations. Additionally, we fine-tuned the transformers with different weighting strategies for the loss function to address class imbalances. Our approach outperforms the STACC baseline by 8.9 per cent on the NLBSE'25 Tool Competition dataset in terms of the average F1$_c$ score, and exceeding the baseline approach in 17 out of 19 cases with a gain ranging from -5.0 to 38.2. The source code is publicly available at https://github.com/moritzmock/NLBSE2025.

In this paper we describe the Fra\"iss\'e limit of finite MV-algebras and then prove that finite MV-algebras verify the Ramsey property. Then we show that MV-algebras are just a special case of a more general situation. In fact, under minumum conditions for the application of the Kechris-Pestov-Todorcevic correspondence, Ramsey property holds for a certain category of finite structures if and only if it holds for a completion subcategory of it., Comment: arXiv admin note: text overlap with arXiv:1102.2000

This study contributes to ongoing research that aims to overcome challenges in predicting the bioapplicability of nanoformulations. It incorporates machine learning and $^{13}$C NMR spectroscopy data. In order to demonstrate the approach, from the PubChem world's largest publicly available chemical database, a bioassay on human dopamine D1 receptor antagonists was downloaded. From the mentioned bioassay the SMILES of compounds were extracted and converted into spectroscopy data by designed for this purpose software. The resulting data was then used for data preprocessing and machine learning, employing scikit-learn algorithms. The ML models were trained by 27,756 samples and tested by 5,466. From the estimators K-Nearest neighbour, Decision Tree Classifier, Random Forest Classifier, Gradient Boosting Classifier, XGBoosting Classifier and Support Vector Classifier, the last performed the best, achieving 71.5 % accuracy, 77.4 % precision, 60.6% recall, 68 % F1, 71.5 % ROC and 0.749 cross-validation score with 0.005 standard deviation. The methodology was designed to be versatile and capable of predicting any functionality of any compound, with or without added nanoformulations, when the appropriate data is available. In addition to the study presented in the article, the time- and cost-efficient CID-SID ML model was developed, giving an opportunity for researchers who have developed a compound and have obtained its PubChem CID and SID to check whether this compound is also a human dopamine D1 receptor antagonist. The metrics of CID_SID ML model were 80.2% accuracy, 86.3% precision, 70.4% recall, 77.6% F1, 79.9% ROC, five-fold cross-validation score 0.8071 with 0.0047 Standard deviation., Comment: 20 pages, 11 figures, 9 tables

We present a quantitative comparison between two different Implicit-Explicit Runge-Kutta (IMEX-RK) approaches for the Euler equations of gas dynamics, specifically tailored for the low Mach limit. In this regime, a classical IMEX-RK approach involves an implicit coupling between the momentum and energy balance so as to avoid the acoustic CFL restriction, while the density can be treated in a fully explicit fashion. This approach leads to a mildly nonlinear equation for the pressure, which can be solved according to a fixed point procedure. An alternative strategy consists of employing a semi-implicit temporal integrator based on IMEX-RK methods (SI-IMEX-RK). The stiff dependence is carefully analyzed, so as to avoid the solution of a nonlinear equation for the pressure also for equations of state (EOS) of non-ideal gases. The spatial discretization is based on a Discontinuous Galerkin (DG) method, which naturally allows high-order accuracy. The asymptotic-preserving (AP) and the asymptotically-accurate (AA) properties of the two approaches are assessed on a number of classical benchmarks for ideal gases and on their extension to non-ideal gases.

Microwave superconducting resonators are extensively studied in fields such as quantum computing and electron spin resonance (ESR) spectroscopy. However, the integration of superconducting resonators with feedback mechanisms to create ultra-low noise oscillators is a relatively unexplored area, and the application of such oscillators in ESR spectroscopy has not yet been demonstrated. In this work, we report the design, fabrication, and application of microwave oscillators based on superconducting resonators for ESR spectroscopy. Specifically, ESR spectra are obtained by measuring the oscillator's frequency shift induced by the ESR effect as a function of the applied static magnetic field. The oscillators are composed of a single heterojunction bipolar transistor (HBT) or high electron mobility transistor (HEMT) coupled with NbTi or YBa$_2$Cu$_3$O$_7$ (YBCO) superconducting resonators. The fabricated oscillators operate at frequencies of 0.6 and 1.7 GHz and temperatures up to 80 K (for YBCO resonators) and 8 K (for NbTi resonators). The lowest measured frequency noise is about 9 mHz/Hz$^{1/2}$, the best spin sensitivity is about $1\times10^{10}$ spins/Hz$^{1/2}$, and the best concentration sensitivity is about $3\times10^{18}$ spins/Hz$^{1/2}$m$^3$. The approach proposed in this work should allow for significantly better spin and concentration sensitivities compared to those achievable with normal conductors, up to operating frequencies, magnetic fields, and temperatures where superconductors exhibit substantially lower effective microwave resistance than normal conductors., Comment: article, 7 pages, 4 figures, supplementary material, 21 pages, 13 figures

Increasingly, recommender systems are tasked with improving users' long-term satisfaction. In this context, we study a content exploration task, which we formalize as a bandit problem with delayed rewards. There is an apparent trade-off in choosing the learning signal: waiting for the full reward to become available might take several weeks, slowing the rate of learning, whereas using short-term proxy rewards reflects the actual long-term goal only imperfectly. First, we develop a predictive model of delayed rewards that incorporates all information obtained to date. Rewards as well as shorter-term surrogate outcomes are combined through a Bayesian filter to obtain a probabilistic belief. Second, we devise a bandit algorithm that quickly learns to identify content aligned with long-term success using this new predictive model. We prove a regret bound for our algorithm that depends on the \textit{Value of Progressive Feedback}, an information theoretic metric that captures the quality of short-term leading indicators that are observed prior to the long-term reward. We apply our approach to a podcast recommendation problem, where we seek to recommend shows that users engage with repeatedly over two months. We empirically validate that our approach significantly outperforms methods that optimize for short-term proxies or rely solely on delayed rewards, as demonstrated by an A/B test in a recommendation system that serves hundreds of millions of users.

We consider coresets for $k$-clustering problems, where the goal is to assign points to centers minimizing powers of distances. A popular example is the $k$-median objective $\sum_{p}\min_{c\in C}dist(p,C)$. Given a point set $P$, a coreset $\Omega$ is a small weighted subset that approximates the cost of $P$ for all candidate solutions $C$ up to a $(1\pm\varepsilon )$ multiplicative factor. In this paper, we give a sharp VC-dimension based analysis for coreset construction. As a consequence, we obtain improved $k$-median coreset bounds for the following metrics: Coresets of size $\tilde{O}\left(k\varepsilon^{-2}\right)$ for shortest path metrics in planar graphs, improving over the bounds $\tilde{O}\left(k\varepsilon^{-6}\right)$ by [Cohen-Addad, Saulpic, Schwiegelshohn, STOC'21] and $\tilde{O}\left(k^2\varepsilon^{-4}\right)$ by [Braverman, Jiang, Krauthgamer, Wu, SODA'21]. Coresets of size $\tilde{O}\left(kd\ell\varepsilon^{-2}\log m\right)$ for clustering $d$-dimensional polygonal curves of length at most $m$ with curves of length at most $\ell$ with respect to Frechet metrics, improving over the bounds $\tilde{O}\left(k^3d\ell\varepsilon^{-3}\log m\right)$ by [Braverman, Cohen-Addad, Jiang, Krauthgamer, Schwiegelshohn, Toftrup, and Wu, FOCS'22] and $\tilde{O}\left(k^2d\ell\varepsilon^{-2}\log m \log |P|\right)$ by [Conradi, Kolbe, Psarros, Rohde, SoCG'24].

In Formula One, teams compete to develop their cars and achieve the highest possible finishing position in each race. During a race, however, teams are unable to alter the car, so they must improve their cars' finishing positions via race strategy, i.e. optimising their selection of which tyre compounds to put on the car and when to do so. In this work, we introduce a reinforcement learning model, RSRL (Race Strategy Reinforcement Learning), to control race strategies in simulations, offering a faster alternative to the industry standard of hard-coded and Monte Carlo-based race strategies. Controlling cars with a pace equating to an expected finishing position of P5.5 (where P1 represents first place and P20 is last place), RSRL achieves an average finishing position of P5.33 on our test race, the 2023 Bahrain Grand Prix, outperforming the best baseline of P5.63. We then demonstrate, in a generalisability study, how performance for one track or multiple tracks can be prioritised via training. Further, we supplement model predictions with feature importance, decision tree-based surrogate models, and decision tree counterfactuals towards improving user trust in the model. Finally, we provide illustrations which exemplify our approach in real-world situations, drawing parallels between simulations and reality., Comment: 9 pages, 6 figures. Copyright ACM 2025. This is the authors' version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record will be published in SAC 2025, http://dx.doi.org/10.1145/3672608.3707766

Formula One (F1) race strategy takes place in a high-pressure and fast-paced environment where split-second decisions can drastically affect race results. Two of the core decisions of race strategy are when to make pit stops (i.e. replace the cars' tyres) and which tyre compounds (hard, medium or soft, in normal conditions) to select. The optimal pit stop decisions can be determined by estimating the tyre degradation of these compounds, which in turn can be computed from the energy applied to each tyre, i.e. the tyre energy. In this work, we trained deep learning models, using the Mercedes-AMG PETRONAS F1 team's historic race data consisting of telemetry, to forecast tyre energies during races. Additionally, we fitted XGBoost, a decision tree-based machine learning algorithm, to the same dataset and compared the results, with both giving impressive performance. Furthermore, we incorporated two different explainable AI methods, namely feature importance and counterfactual explanations, to gain insights into the reasoning behind the forecasts. Our contributions thus result in an explainable, automated method which could assist F1 teams in optimising their race strategy., Comment: 9 pages, 9 figures. Copyright ACM 2025. This is the authors' version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record will be published in SAC 2025, http://dx.doi.org/10.1145/3672608.3707765

We investigate the explainability of Reinforcement Learning (RL) policies from a temporal perspective, focusing on the sequence of future outcomes associated with individual actions. In RL, value functions compress information about rewards collected across multiple trajectories and over an infinite horizon, allowing a compact form of knowledge representation. However, this compression obscures the temporal details inherent in sequential decision-making, presenting a key challenge for interpretability. We present Temporal Policy Decomposition (TPD), a novel explainability approach that explains individual RL actions in terms of their Expected Future Outcome (EFO). These explanations decompose generalized value functions into a sequence of EFOs, one for each time step up to a prediction horizon of interest, revealing insights into when specific outcomes are expected to occur. We leverage fixed-horizon temporal difference learning to devise an off-policy method for learning EFOs for both optimal and suboptimal actions, enabling contrastive explanations consisting of EFOs for different state-action pairs. Our experiments demonstrate that TPD generates accurate explanations that (i) clarify the policy's future strategy and anticipated trajectory for a given action and (ii) improve understanding of the reward composition, facilitating fine-tuning of the reward function to align with human expectations., Comment: 21 pages, 4 figures

Although deep reinforcement learning has been shown to be effective, the model's black-box nature presents barriers to direct policy interpretation. To address this problem, we propose a neuro-symbolic approach called neural DNF-MT for end-to-end policy learning. The differentiable nature of the neural DNF-MT model enables the use of deep actor-critic algorithms for training. At the same time, its architecture is designed so that trained models can be directly translated into interpretable policies expressed as standard (bivalent or probabilistic) logic programs. Moreover, additional layers can be included to extract abstract features from complex observations, acting as a form of predicate invention. The logic representations are highly interpretable, and we show how the bivalent representations of deterministic policies can be edited and incorporated back into a neural model, facilitating manual intervention and adaptation of learned policies. We evaluate our approach on a range of tasks requiring learning deterministic or stochastic behaviours from various forms of observations. Our empirical results show that our neural DNF-MT model performs at the level of competing black-box methods whilst providing interpretable policies., Comment: AAMAS 2025

Clinical trials often collect data on multiple outcomes, such as overall survival (OS), progression-free survival (PFS), and response to treatment (RT). In most cases, however, study designs only use primary outcome data for interim and final decision-making. In several disease settings, clinically relevant outcomes, for example OS, become available years after patient enrollment. Moreover, the effects of experimental treatments on OS might be less pronounced compared to auxiliary outcomes such as RT. We develop a Bayesian decision-theoretic framework that uses both primary and auxiliary outcomes for interim and final decision-making. The framework allows investigators to control standard frequentist operating characteristics, such as the type I error rate, and can be used with auxiliary outcomes from emerging technologies, such as circulating tumor assays. False positive rates and other frequentist operating characteristics are rigorously controlled without any assumption about the concordance between primary and auxiliary outcomes. We discuss algorithms to implement this decision-theoretic approach and show that incorporating auxiliary information into interim and final decision-making can lead to relevant efficiency gains according to established and interpretable metrics.

Significance and Object: The proposed methodology aims to provide time- and cost-effective approach for the early stage in drug discovery. The machine learning models developed in this study used only the identification numbers provided by PubChem. Thus, a drug development researcher who has obtained a PubChem CID and SID can easily identify new functionality of their compound. The approach was demonstrated, using four bioassay which were on (i) the antagonists of human D3 dopamine receptors; (ii) the promoter Rab9 activators; (iii) small molecule inhibitors of CHOP to regulate the unfolded protein response to ER stress; (iv) antagonists of the human M1 muscarinic receptor. Solution: The four bioassays used for demonstration of the approach were provided by PubChem. For each bioassay, the generated by PubChem CIDs, SIDs were extracted together with the corresponding activity. The resulting dataset was sifted with the dataset on a water solubility bioassay, remaining only the compounds common for both bioassays. In this way, the inactive compounds were reduced. Then, all active compounds were added, and the resulted dataset was later used for machine learning based on scikit learn algorithms. Results: The average values of the ML models` metrics for the four bioassays were: 83.82% Accuracy with 5.35 standard deviation; 87.9% Precision with 5.04 standard deviation; 77.1% Recall with 7.65 standard deviation; 82.1% F1 with 6.44 standard deviation; 83.4% ROC with 5.09 standard deviation., Comment: 27 pages, 26 figures, 20 tables

Reward machines (RMs) are an effective approach for addressing non-Markovian rewards in reinforcement learning (RL) through finite-state machines. Traditional RMs, which label edges with propositional logic formulae, inherit the limited expressivity of propositional logic. This limitation hinders the learnability and transferability of RMs since complex tasks will require numerous states and edges. To overcome these challenges, we propose First-Order Reward Machines ($\texttt{FORM}$s), which use first-order logic to label edges, resulting in more compact and transferable RMs. We introduce a novel method for $\textbf{learning}$ $\texttt{FORM}$s and a multi-agent formulation for $\textbf{exploiting}$ them and facilitate their transferability, where multiple agents collaboratively learn policies for a shared $\texttt{FORM}$. Our experimental results demonstrate the scalability of $\texttt{FORM}$s with respect to traditional RMs. Specifically, we show that $\texttt{FORM}$s can be effectively learnt for tasks where traditional RM learning approaches fail. We also show significant improvements in learning speed and task transferability thanks to the multi-agent learning framework and the abstraction provided by the first-order language., Comment: AAMAS'25

We consider the problem of transferring knowledge from a source, or proxy, domain to a new target domain for learning a high-dimensional regression model with possibly different features. Recently, the statistical properties of homogeneous transfer learning have been investigated. However, most homogeneous transfer and multi-task learning methods assume that the target and proxy domains have the same feature space, limiting their practical applicability. In applications, target and proxy feature spaces are frequently inherently different, for example, due to the inability to measure some variables in the target data-poor environments. Conversely, existing heterogeneous transfer learning methods do not provide statistical error guarantees, limiting their utility for scientific discovery. We propose a two-stage method that involves learning the relationship between the missing and observed features through a projection step in the proxy data and then solving a joint penalized regression optimization problem in the target data. We develop an upper bound on the method's parameter estimation risk and prediction risk, assuming that the proxy and the target domain parameters are sparsely different. Our results elucidate how estimation and prediction error depend on the complexity of the model, sample size, the extent of overlap, and correlation between matched and mismatched features.

Natural Language Processing and Generation systems have recently shown the potential to complement and streamline the costly and time-consuming job of professional fact-checkers. In this work, we lift several constraints of current state-of-the-art pipelines for automated fact-checking based on the Retrieval-Augmented Generation (RAG) paradigm. Our goal is to benchmark, under more realistic scenarios, RAG-based methods for the generation of verdicts - i.e., short texts discussing the veracity of a claim - evaluating them on stylistically complex claims and heterogeneous, yet reliable, knowledge bases. Our findings show a complex landscape, where, for example, LLM-based retrievers outperform other retrieval techniques, though they still struggle with heterogeneous knowledge bases; larger models excel in verdict faithfulness, while smaller models provide better context adherence, with human evaluations favouring zero-shot and one-shot approaches for informativeness, and fine-tuned models for emotional alignment., Comment: Code and data at https://github.com/drusso98/face-the-facts

All biological systems are subject to perturbations: due to thermal fluctuations, external environments, or mutations. Yet, while biological systems are composed of thousands of interacting components, recent high-throughput experiments show that their response to perturbations is surprisingly low-dimensional: confined to only a few stereotyped changes out of the many possible. Here, we explore a unifying dynamical systems framework - soft modes - to explain and analyze low-dimensionality in biology, from molecules to eco-systems. We argue that this one framework of soft modes makes non-trivial predictions that generalize classic ideas from developmental biology to disparate systems, namely: phenocopying, dual buffering, and global epistasis. While some of these predictions have been borne out in experiments, we discuss how soft modes allow for a surprisingly far-reaching and unifying framework in which to analyze data from protein biophysics to microbial ecology.

Recent studies in interpretability have explored the inner workings of transformer models trained on tasks across various domains, often discovering that these networks naturally develop surprisingly structured representations. When such representations comprehensively reflect the task domain's structure, they are commonly referred to as ``World Models'' (WMs). In this work, we discover such WMs in transformers trained on maze tasks. In particular, by employing Sparse Autoencoders (SAEs) and analysing attention patterns, we examine the construction of WMs and demonstrate consistency between the circuit analysis and the SAE feature-based analysis. We intervene upon the isolated features to confirm their causal role and, in doing so, find asymmetries between certain types of interventions. Surprisingly, we find that models are able to reason with respect to a greater number of active features than they see during training, even if attempting to specify these in the input token sequence would lead the model to fail. Futhermore, we observe that varying positional encodings can alter how WMs are encoded in a model's residual stream. By analyzing the causal role of these WMs in a toy domain we hope to make progress toward an understanding of emergent structure in the representations acquired by Transformers, leading to the development of more interpretable and controllable AI systems., Comment: Main paper: 9 pages, 9 figures. Supplementary material: 10 pages, 17 additional figures. Code and data will be available upon publication. Corresponding author: A. F. Spies (afspies@imperial.ac.uk)

We study the behavior of perturbations in a compressible one-dimensional inviscid gas with an ambient state consisting of constant pressure and periodically-varying density. We show through asymptotic analysis that long-wavelength perturbations approximately obey a system of dispersive nonlinear wave equations. Computational experiments demonstrate that solutions of the 1D Euler equations agree well with this dispersive model, with solutions consisting mainly of solitary waves. Shock formation seems to be avoided for moderate-amplitude initial data, while shock formation occurs for larger initial data. We investigate the threshold for transition between these behaviors, validating a previously-proposed criterion based on further computational experiments. These results support the existence of large-time non-breaking solutions to the 1D compressible Euler equations, as hypothesized in previous works.

Photoluminescent defects in diamond, like the nitrogen-vacancy (NV) color center, are at the forefront of emerging optical quantum technologies. Most emit in the visible and near-infrared spectral region below 1000 nm (NIR-I), limiting their applications in photonics, fiber communications, and biology. Here, we show that the nitrogen-vacancy-nitrogen ($N_2V$) center, which emits in the visible and near-infrared-II (NIR-II, 1000-1700 nm), is ubiquitous in as-synthesized and processed nitrogen-doped diamond from bulk samples to nanoparticles. We demonstrate that $N_2V$ is also present in commercially available state-of-the-art NV diamond sensing chips made via chemical vapor deposition (CVD). In high-pressure high-temperature (HPHT) diamonds, the photoluminescence (PL) intensity of both $N_2V$ charge states, $N_2V^0$ in the visible and $N_2V^-$ in the NIR-II, increases with increasing substitutional nitrogen concentration. We determine the PL lifetime of $N_2V^-$ to be 0.3 ns and compare a quantum optical and density functional theory model of the $N_2V^-$ with experimental PL spectra. Finally, we show that detonation nanodiamonds (DND) show stable PL in the NIR-II, which we attribute to the $N_2V$ color center, and use this NIR-II PL to image DNDs inside skin cells. Our results will contribute to the scientific and technological exploration and development of the $N_2V$ color center and inspire more research into its effect on other color centers in diamond.

Information processing currently reaches speeds as high as 800 GHz. However, the underlying transistor technology is quickly approaching its fundamental limits and further progress requires a disruptive approach. One such path is to manipulate quantum properties of solids, such as the valley degree of freedom, with ultrashort controlled lightwaves. Here we employ a sequence of few-optical-cycle visible pulses controlled with attosecond precision to excite and switch the valley pseudospin in a 2D semiconductor. We show that a pair of pulses separated in time with linear orthogonal polarizations can induce a valley-selective population. Additionally, exploiting a four-pump excitation protocol, we perform logic operations such as valley de-excitation and re-excitation at room temperature at rates as high as ~10 THz., Comment: 14 pages, 3 figures

We consider the problem of optimally composing a set of primitives to tackle control tasks. To address this problem, we introduce Neo-FREE: a control architecture inspired by the Thousand Brains Theory and Free Energy Principle from cognitive sciences. In accordance with the neocortical (Neo) processes postulated by the Thousand Brains Theory, Neo-FREE consists of functional units returning control primitives. These are linearly combined by a gating mechanism that minimizes the variational free energy (FREE). The problem of finding the optimal primitives' weights is then recast as a finite-horizon optimal control problem, which is convex even when the cost is not and the environment is nonlinear, stochastic, non-stationary. The results yield an algorithm for primitives composition and the effectiveness of Neo-FREE is illustrated via in-silico and hardware experiments on an application involving robot navigation in an environment with obstacles., Comment: 13 pages, 2 figures

Humans can perform exquisite sensorimotor skills, both individually and in teams, from athletes performing rhythmic gymnastics to everyday tasks like carrying a cup of coffee. The "predictive brain" framework suggests that mastering these tasks relies on predictive mechanisms, raising the question of how we deploy such predictions for real-time control and coordination. This review highlights two lines of research: one showing that during the control of complex objects people make the interaction with 'tools' predictable; the second one examines dyadic coordination showing that people make their behavior predictable for their partners. These studies demonstrate that to achieve sophisticated motor skills, we play "prediction tricks": we select subspaces of predictable solutions and make sensorimotor interactions more predictable and legible by and for others. This synthesis underscores the critical role of predictability in optimizing control strategies across various contexts and establishes a link between predictive processing and closed-loop control theories of behavior.

We investigate a scheduling problem arising from a material handling and processing problem in a production line of an Austrian company building prefabricated house walls. The addressed problem is a permutation flow shop with blocking constraints in which the machine of at least one stage can process a number operations of two other stages in the system. This situation is usually referred to as multi-task or inter-stage flexibility. The problem is in general NP-hard, but we derive a number of special cases that can be solved in polynomial time. For the general case, we present a variety of heuristic algorithms, with a focus on matheuristics that are grounded in two different mixed-integer linear programming (MIP) formulations of the problem. These matheuristics leverage the strengths of exact optimization techniques while introducing flexibility to address limits on computation time. To assess the performance of the proposed approaches, we conduct an extensive computational study on randomly generated test cases based on real-world instances., Comment: 33 pages, 9 figures, submitted to "Computers & Operations Research" on Nov. 22, 2024

A deadlock occurs in a network when two or more items prevent each other from moving and are stalled. In a general model, items are stored at vertices and each vertex $v$ has a buffer with $b(v)$ slots. Given a route for each item toward its destination, the Deadlock Safety Problem asks whether the current state is safe, i.e., it is possible to deliver each item at its destination, or is bound to deadlock, i.e., any sequence of moves will end up with a set of items stalled. While when $b \geq 2$ the problem is solvable in polynomial time building upon a nice characterization of YES/NO-instances, it is NP-hard on quite simple graphs as grids when $b=1$ and on trees when $b\leq 3$. We improve on these results by means of two new tools, weak deadlock sets and wise states. We show that for general networks and $b$ a state that is wise and without weak deadlock sets -- this can be recognized in polynomial time -- is safe: this is indeed a strengthening of the result for $b\geq 2$. We sharpen this result for trees, where we show that a wise state is safe if and only if it has no weak deadlock set. That is interesting in particular in the context of rail transportation where networks are often single-tracked and deadlock detection and avoidance focuses on local sub-networks, mostly with a tree-like structure. We pose some research questions for future investigations.

In order to increase rail freight transportation in Italy, Rete Ferroviaria Italiana (RFI) the Italian railway infrastructure manager, is carrying out several investment plans to enhance the Transshipment Yards, that act as an interface between the rail and road networks. The need is to increase their practical capacity, i.e. the maximum number of train services that can be inserted without altering the current timetable while respecting all relevant constraints. Several factors influence the practical capacity of a transshipment yard: physical resources (such as tracks and vehicles for loading/unloading); constraints on the possible time slots of individual operations; constraints on the length of time a train must stay in the yard, that follow from both timetable requirements that are settled by the (prevalent) main line and from administrative and organisational issues in the yard. In this paper, we propose a MILP-based optimization model that is based on the solution of a suitable saturation problem, that deals with all these constraints and that can be used for evaluating the practical capacity of a transshipment yard both in its current configuration and in any plausible future configuration. The model provides operational details, such as routes and schedules, for each train service, and allows to impose periodic timetables and schedules that keep the daily management of the yard easier. Both the model and its solutions are validated on a real Italian transshipment yard, located at Marzaglia, on different scenarios corresponding to different investment plans of RFI. The results show that proper investments allow to get a feasible timetable with a period of 24 hours with doubles the number of current train services.

We study the radiated energy by a scalar particle moving on a circular orbit (smeared in the extra dimension) in the spacetime of a topological star, extending a previous study [Phys. Rev. D {\bf 110}, 084077 (2024)]. We discuss motion in the presence of self-force effects too., Comment: 17 pages, 5 figures, revtex macros

Multiple Sclerosis (MS) is a heterogeneous autoimmune-mediated disorder affecting the central nervous system, commonly manifesting as fatigue and progressive limb impairment. This can significantly impact quality of life due to weakness or paralysis in the upper and lower limbs. A Brain-Computer Interface (BCI) aims to restore quality of life through control of an external device, such as a wheelchair. However, the limited BCI research in people with MS is insufficient. The current study aims to expand on the current MS-BCI literature by highlighting the feasibility of decoding MS imagined movement. We collected electroencephalography (EEG) data from eight participants with various symptoms of MS and ten neurotypical control participants. Participants made imagined movements of the hands and feet as directed by a go no-go protocol. Binary regularised linear discriminant analysis was used to classify imagined movement at individual time-frequency points. The frequency bands which provided the maximal accuracy, and the associated latency, were compared. In all MS participants, the classification algorithm achieved above 70% accuracy in at least one imagined movement vs. rest classification and most movement vs. movement classifications. There was no significant difference between classification of limbs with weakness or paralysis to neurotypical controls. Both the MS and control groups possessed decodable information within the alpha (7-13 Hz) and beta (16-30 Hz) bands at similar latency. This study is the first to demonstrate the feasibility of decoding imagined movements in people with MS. As an alternative to the P300 response, motor imagery-based control of a BCI may also be combined with existing motor imagery therapy to supplement MS rehabilitation. These promising results merit further long term BCI studies to investigate the effect of MS progression on classification performance., Comment: Word count (main text): 5964 Word count (abstract): 299 Page Count: 21 Number of figures: 6 + 1 table, 3 supplementary

This work presents the existence and uniqueness of solution to a free boundary value problem related to biofilm growth. The problem consists of a system of nonlinear hyperbolic partial differential equations governing the microbial species growth, and a system of parabolic partial differential equations describing the substrate dynamics. The free boundary evolution is governed by an ordinary differential equation that accounts for the thickness of the biofilm. We use the method of characteristics and fixed point strategies to prove the existence and uniqueness theorem in small and all times. All the equations are converted into integral equations, in particular this transformation is made for the parabolic equations by using the Green's functions. We consider Dirichlet-Neumann and Neumann-Robin boundary conditions for the substrates equations and their extension to the case with variable

Graph Neural Networks (GNNs) have shown significant promise in various domains, such as recommendation systems, bioinformatics, and network analysis. However, the irregularity of graph data poses unique challenges for efficient computation, leading to the development of specialized GNN accelerator architectures that surpass traditional CPU and GPU performance. Despite this, the structural diversity of input graphs results in varying performance across different GNN accelerators, depending on their dataflows. This variability in performance due to differing dataflows and graph properties remains largely unexplored, limiting the adaptability of GNN accelerators. To address this, we propose a data-driven framework for dataflow-aware latency prediction in GNN inference. Our approach involves training regressors to predict the latency of executing specific graphs on particular dataflows, using simulations on synthetic graphs. Experimental results indicate that our regressors can predict the optimal dataflow for a given graph with up to 91.28% accuracy and a Mean Absolute Percentage Error (MAPE) of 3.78%. Additionally, we introduce an online scheduling algorithm that uses these regressors to enhance scheduling decisions. Our experiments demonstrate that this algorithm achieves up to $3.17\times$ speedup in mean completion time and $6.26\times$ speedup in mean execution time compared to the best feasible baseline across all datasets., Comment: Accepted for ASP-DAC 2025

Agent-Based Modelling (ABM) has emerged as an essential tool for simulating social networks, encompassing diverse phenomena such as information dissemination, influence dynamics, and community formation. However, manually configuring varied agent interactions and information flow dynamics poses challenges, often resulting in oversimplified models that lack real-world generalizability. Integrating modern Large Language Models (LLMs) with ABM presents a promising avenue to address these challenges and enhance simulation fidelity, leveraging LLMs' human-like capabilities in sensing, reasoning, and behavior. In this paper, we propose a novel framework utilizing LLM-empowered agents to simulate social network users based on their interests and personality traits. The framework allows for customizable agent interactions resembling various social network platforms, including mechanisms for content resharing and personalized recommendations. We validate our framework using a comprehensive Twitter dataset from the 2020 US election, demonstrating that LLM-agents accurately replicate real users' behaviors, including linguistic patterns and political inclinations. These agents form homogeneous ideological clusters and retain the main themes of their community. Notably, preference-based recommendations significantly influence agent behavior, promoting increased engagement, network homophily and the formation of echo chambers. Overall, our findings underscore the potential of LLM-agents in advancing social media simulations and unraveling intricate online dynamics., Comment: Accepted at ASONAM2024

We provide non-trivial checks of the recently proposed duality between double-scaled SYK and a 2d dilaton gravity model with sine potential, studying the path integral at one-loop level. Specifically, we compute the logarithmic correction to the free energy of sine-dilaton gravity and, up to potential ordering ambiguities, we find a match with the corresponding quantity in double-scaled SYK. The computation relies on the description of sine-dilaton gravity in terms of a version of the q-Schwarzian theory, the quantum deformation of the standard Schwarzian model dual to JT gravity. A crucial aspect of the calculation is selecting the correct Hartle-Hawking vacuum for the gravitational theory, which implies a specific choice of boundary conditions for the one-loop determinant, computed using a generalization of the Gel'fand-Yaglom's theorem. We also evaluate the gravitational one-loop correction to the boundary to boundary propagator of a non-minimally coupled matter field in the bulk theory, showing a perfect agreement with the corresponding quantum correction of matter correlators in double-scaled SYK.

This paper addresses the problem of stabilization of switched affine systems under dwell-time constraint, giving guarantees on the bound of the quadratic cost associated with the proposed state switching control law. Specifically, two switching rules are presented relying on the solution of differential Lyapunov inequalities and Lyapunov-Metzler inequalities, from which the stability conditions are expressed. The first one allows to regulate the state of linear switched systems to zero, whereas the second one is designed for switched affine systems proving practical stability of the origin. In both cases, the determination of a guaranteed cost associated with each control strategy is shown. In the cases of linear and affine systems, the existence of the solution for the Lyapunov-Metzler condition is discussed and guidelines for the selection of a solution ensuring suitable performance of the system evolution are provided. The theoretical results are finally assessed by means of three examples., Comment: 12 pages, 10 figures

The development of artificial intelligence systems capable of understanding and reasoning about complex real-world scenarios is a significant challenge. In this work we present a novel approach to enhance and exploit LLM reactive capability to address complex problems and interpret deeply contextual real-world meaning. We introduce a method and a tool for creating a multimodal, knowledge-augmented formal representation of meaning that combines the strengths of large language models with structured semantic representations. Our method begins with an image input, utilizing state-of-the-art large language models to generate a natural language description. This description is then transformed into an Abstract Meaning Representation (AMR) graph, which is formalized and enriched with logical design patterns, and layered semantics derived from linguistic and factual knowledge bases. The resulting graph is then fed back into the LLM to be extended with implicit knowledge activated by complex heuristic learning, including semantic implicatures, moral values, embodied cognition, and metaphorical representations. By bridging the gap between unstructured language models and formal semantic structures, our method opens new avenues for tackling intricate problems in natural language understanding and reasoning.

Here we present a technique for using machine learning (ML) for single-qubit gate synthesis on field programmable logic for a superconducting transmon-based quantum computer based on simulated studies. Our approach is multi-stage. We first bootstrap a model based on simulation with access to the full statevector for measuring gate fidelity. We next present an algorithm, named adapted randomized benchmarking (ARB), for fine-tuning the gate on hardware based on measurements of the devices. We also present techniques for deploying the model on programmable devices with care to reduce the required resources. While the techniques here are applied to a transmon-based computer, many of them are portable to other architectures.

We report measurements of production cross sections for $\rho^+$, $\rho^0$, $\omega$, $K^{*+}$, $K^{*0}$, $\phi$, $\eta$, $K_S^0$, $f_0(980)$, $D^+$, $D^0$, $D_s^+$, $D^{*+}$, $D^{*0}$, and $D^{*+}_s$ in $e^+e^-$ collisions at a center-of-mass energy near 10.58 GeV. The data were recorded by the Belle experiment, consisting of 571 fb$^{-1}$ at 10.58 GeV and 74 fb$^{-1}$ at 10.52 GeV. Production cross sections are extracted as a function of the fractional hadron momentum $x_p$ . The measurements are compared to {\sc pythia} Monte Carlo generator predictions with various fragmentation settings, including those that have increased fragmentation into vector mesons over pseudo-scalar mesons. The cross sections measured for light hadrons are consistent with no additional increase of vector over pseudo-scalar mesons. The charmed-meson cross sections are compared to earlier measurements -- when available -- including older Belle results, which they supersede. They are in agreement before application of an improved initial-state radiation correction procedure that causes slight changes in their \xp shapes., Comment: 21 pages, 18 figures, submitted to Phys. Rev. D

Nanoporous carbon materials containing small domains of sp2-carbon with highly disordered structures are promising for supercapacitor applications. Herein, we synthesize amorphous carbon nanofoam with 98% volumetric void fraction and low mass density of around 30 mg/cm3 by pulsed laser deposition at room temperature. With the unavoidable oxygen functional groups on the nanoporous surface, pristine and nitrogen-functionalized carbon nanofoams are directly grown on the desired substrate under background gas of Ar, N2 and N2-H2employed as supercapacitor electrodes. Among the background gases used in synthesis, the use of nitrogen yields nanofoam with higher thickness and more N-content with higher graphitic-N, which is correlated with plasma plume behaviour. From the test of amorphous carbon nanofoam supercapacitor device, nitrogenated amorphous carbon show a higher areal capacitance of 4.1 mF/cm2 at 20 mV/s in aqueous electrolyte, a better capacitance retention at higher current, and excellent cycle stability (98%) over prolonged 10000 charge-discharge cycles are achieved for compared to pristine counterpart prepared under Ar background gas (2.7 mF/cm2 and cycle stability of 88%)., Comment: 14 pages main article, 2 pages supporting material, 5 figures, 3 supporting figures, 1 table

Link adaptation (LA) is an essential function in modern wireless communication systems that dynamically adjusts the transmission rate of a communication link to match time- and frequency-varying radio link conditions. However, factors such as user mobility, fast fading, imperfect channel quality information, and aging of measurements make the modeling of LA challenging. To bypass the need for explicit modeling, recent research has introduced online reinforcement learning (RL) approaches as an alternative to the more commonly used rule-based algorithms. Yet, RL-based approaches face deployment challenges, as training in live networks can potentially degrade real-time performance. To address this challenge, this paper considers offline RL as a candidate to learn LA policies with minimal effects on the network operation. We propose three LA designs based on batch-constrained deep Q-learning, conservative Q-learning, and decision transformer. Our results show that offline RL algorithms can match the performance of state-of-the-art online RL methods when data is collected with a proper behavioral policy.