Search

Bent monolithic active pixel sensors are the basis for the planned fully cylindrical ultra low material budget tracking detector ITS3 of the ALICE experiment. This paper presents results from testbeam campaigns using high-energy particles to verify the performance of 50 um thick bent ALPIDE chips in terms of efficiency and spatial resolution. The sensors were bent to radii of 18, 24 and 30 mm, slightly smaller than the foreseen bending radii of the future ALICE ITS3 layers. An efficiency larger than $99.9\%$ and a spatial resolution of approximately 5 um, in line with the nominal operation of flat ALPIDE sensors, is obtained at nominal operating conditions. These values are found to be independent of the bending radius and thus constitute an additional milestone in the demonstration of the feasibility of the planned ITS3 detector. In addition, a special geometry in which the beam particles graze the chip and traverse it laterally over distances of up to 3 mm is investigated.

The initial conditions in multimode fibers pumped by ultrashort laser pulses strongly determine the following nonlinear optical interactions. In this work we firstly compare the detailed spatial mode content of simple initial conditions, transverse offset and tilt. We then show how those initial conditions can both be used to achieve beam self-cleaning into higher-order spatial modes of a model graded-index fiber, with their own slight differences and advantages. Going beyond purely spatial initial conditions, we introduce self-cleaning results using spatial chirp at the input facet, whereby the different temporal envelopes of the spatial modes allows for tuning the self-cleaning process. Our results open up investigations into higher-dimensional tuning of nonlinear processes in multimode fibers using initial conditions., Comment: 15 pages, 8 figures, comments welcome

Reward-guided text generation (RGTG) has emerged as a viable alternative to offline reinforcement learning from human feedback (RLHF). RGTG methods can align baseline language models to human preferences without further training like in standard RLHF methods. However, they rely on a reward model to score each candidate token generated by the language model at inference, incurring significant test-time overhead. Additionally, the reward model is usually only trained to score full sequences, which can lead to sub-optimal choices for partial sequences. In this work, we present a novel reward model architecture that is trained, using a Bradley-Terry loss, to prefer the optimal expansion of a sequence with just a \emph{single call} to the reward model at each step of the generation process. That is, a score for all possible candidate tokens is generated simultaneously, leading to efficient inference. We theoretically analyze various RGTG reward models and demonstrate that prior techniques prefer sub-optimal sequences compared to our method during inference. Empirically, our reward model leads to significantly faster inference than other RGTG methods. It requires fewer calls to the reward model and performs competitively compared to previous RGTG and offline RLHF methods.

The discovery of new crystalline materials is essential to scientific and technological progress. However, traditional trial-and-error approaches are inefficient due to the vast search space. Recent advancements in machine learning have enabled generative models to predict new stable materials by incorporating structural symmetries and to condition the generation on desired properties. In this work, we introduce SymmBFN, a novel symmetry-aware Bayesian Flow Network (BFN) for crystalline material generation that accurately reproduces the distribution of space groups found in experimentally observed crystals. SymmBFN substantially improves efficiency, generating stable structures at least 50 times faster than the next-best method. Furthermore, we demonstrate its capability for property-conditioned generation, enabling the design of materials with tailored properties. Our findings establish BFNs as an effective tool for accelerating the discovery of crystalline materials.

The Federated Learning paradigm facilitates effective distributed machine learning in settings where training data is decentralized across multiple clients. As the popularity of the strategy grows, increasingly complex real-world problems emerge, many of which require balancing conflicting demands such as fairness, utility, and resource consumption. Recent works have begun to recognise the use of a multi-objective perspective in answer to this challenge. However, this novel approach of combining federated methods with multi-objective optimisation has never been discussed in the broader context of both fields. In this work, we offer a first clear and systematic overview of the different ways the two fields can be integrated. We propose a first taxonomy on the use of multi-objective methods in connection with Federated Learning, providing a targeted survey of the state-of-the-art and proposing unambiguous labels to categorise contributions. Given the developing nature of this field, our taxonomy is designed to provide a solid basis for further research, capturing existing works while anticipating future additions. Finally, we outline open challenges and possible directions for further research.

We present deep ALMA observations targeting the [CII]$_{158\mu m}\,$ line in JADES-GS-z14-0, the most distant known galaxy at z=14.1793. We do not detect the [CII]$_{158\mu m}\,$ line in our deep observations, implying a luminosity of $<$6$\times10^7$ L$_{\odot}$ (3$\sigma$) for the target. Comparing this with the detected [OIII]$_{88\mu m}\,$ line, we constrain the [OIII]/[CII] ratio to be $>$3.5, significantly improving our probe of the ionization parameter $U$. The observed ratio is higher than analogues in the local universe, but consistent with galaxies at $z\approx6$-9. Through ISM modeling, we infer extreme ionizing conditions with log(U)$>-$2.0, likely requiring a young stellar population. Our modeling also indicates a relatively low gas density ($51_{-32}^{+116}$ cm$^{-3}$), significantly lower than expected from lower redshift trends. We infer a relatively high gas-phase metallicity (16$\pm$6\% solar) consistent with previous results and implying a rapid build-up of metals. Finally, using [CII]$_{158\mu m}\,$ as a molecular/cold gas mass tracer, we infer a low gas fraction ($f_\mathrm{gas} < 0.77$), consistent with previous estimates of the dynamical mass from [OIII]$_{88\mu m}\,$. Combined with the low observed gas density, lack of dust and high ionization parameter, this suggests strong feedback processes are playing an important role in the evolution of this galaxy. Our observations show that JADES-GS-z14-0 is a rapidly evolving galaxy with extreme ISM conditions, shedding light on the earliest phases of galaxy formation., Comment: 9 Pages, 4 Figures, 1 Table. To be submitted to The Astrophysical Journal

There is an intricate relationship between the organization of large-scale magnetic fields by a stellar dynamo and the rate of angular momentum loss due to magnetized stellar winds. An essential ingredient for the operation of a large-scale dynamo is the Coriolis force, which imprints organizing flows on the global convective patterns and inhibits the complete cancellation of bipolar magnetic regions. Consequently, it is natural to expect a rotational threshold for large-scale dynamo action and for the efficient angular momentum loss that it mediates through magnetic braking. Here we present new observational constraints on magnetic braking for an evolutionary sequence of six early K-type stars. To determine the wind braking torque for each of our targets, we combine spectropolarimetric constraints on the large-scale magnetic field, Ly-alpha or X-ray constraints on the mass-loss rate, as well as uniform estimates of the stellar rotation period, mass, and radius. As identified previously from similar observations of hotter stars, we find that the wind braking torque decreases abruptly by more than an order of magnitude at a critical value of the stellar Rossby number. Given that all of the stars in our sample exhibit clear activity cycles, we suggest that weakened magnetic braking may coincide with the operation of a subcritical stellar dynamo., Comment: 14 pages including 5 figures and 1 table. Under second review at AAS Journals

Efficient sampling of unnormalized probability densities such as the Boltzmann distribution of molecular systems is a longstanding challenge. Next to conventional approaches like molecular dynamics or Markov chain Monte Carlo, variational approaches, such as training normalizing flows with the reverse Kullback-Leibler divergence, have been introduced. However, such methods are prone to mode collapse and often do not learn to sample the full configurational space. Here, we present temperature-annealed Boltzmann generators (TA-BG) to address this challenge. First, we demonstrate that training a normalizing flow with the reverse Kullback-Leibler divergence at high temperatures is possible without mode collapse. Furthermore, we introduce a reweighting-based training objective to anneal the distribution to lower target temperatures. We apply this methodology to three molecular systems of increasing complexity and, compared to the baseline, achieve better results in almost all metrics while requiring up to three times fewer target energy evaluations. For the largest system, our approach is the only method that accurately resolves the metastable states of the system.

Instruction-based computer control agents (CCAs) execute complex action sequences on personal computers or mobile devices to fulfill tasks using the same graphical user interfaces as a human user would, provided instructions in natural language. This review offers a comprehensive overview of the emerging field of instruction-based computer control, examining available agents -- their taxonomy, development, and respective resources -- and emphasizing the shift from manually designed, specialized agents to leveraging foundation models such as large language models (LLMs) and vision-language models (VLMs). We formalize the problem and establish a taxonomy of the field to analyze agents from three perspectives: (a) the environment perspective, analyzing computer environments; (b) the interaction perspective, describing observations spaces (e.g., screenshots, HTML) and action spaces (e.g., mouse and keyboard actions, executable code); and (c) the agent perspective, focusing on the core principle of how an agent acts and learns to act. Our framework encompasses both specialized and foundation agents, facilitating their comparative analysis and revealing how prior solutions in specialized agents, such as an environment learning step, can guide the development of more capable foundation agents. Additionally, we review current CCA datasets and CCA evaluation methods and outline the challenges to deploying such agents in a productive setting. In total, we review and classify 86 CCAs and 33 related datasets. By highlighting trends, limitations, and future research directions, this work presents a comprehensive foundation to obtain a broad understanding of the field and push its future development.

Under which condition is quantization optimal? We address this question in the context of the additive uniform noise channel under peak amplitude and power constraints. We compute analytically the capacity-achieving input distribution as a function of the noise level, the average power constraint and the exponent of the power constraint. We found that when the cost constraint is tight and the cost function is concave, the capacity-achieving input distribution is discrete, whereas when the cost function is convex, the support of the capacity-achieving input distribution spans the entire interval., Comment: 6 pages, 3 figures

Execution Tickets are currently discussed as a next evolutionary step in Ethereum's block space allocation mechanism, separating consensus rewards from execution rewards and selling execution rights through a dedicated market. We present a theoretical framework identifying three core objectives for this mechanism - Decentralization, MEV capture, and Block Producer Incentive Compatibility - alongside practical metrics for evaluating each objective. To meet these goals, we explore seven key design parameters: ticket quantity, expiry, refundability, resalability, enhanced lookahead, pricing mechanism, and target ticket amount. We then evaluate four pricing mechanisms and construct six concrete mechanism designs from these parameters. To assess trade-offs in real-world conditions, we perform an agent-based simulation with over 300 runs. Our findings suggest that auction-driven formats, particularly second-price, excel in capturing significant MEV. The simulation also indicates that offering a secondary market can help alleviate centralization, since specialized ticket holders can enter and exit the market as needed. Non-expiring tickets show promise in reducing valuation risks, as ticket holders are not influenced by expiry-related discounting. Likewise, removing refundability simplifies the mechanism without notably impairing performance. Extended lookahead periods benefit price predictability and smoothness at a slight cost to price accuracy. Overall, this study provides a theoretical framework on the mechanism design space for Execution Tickets as well as a practical implementation of an agent-based simulation to test mechanism design choices. Further, it provides an exploratory evaluation of Execution Ticket mechanism designs, offering insights into optimal configurations that balance MEV capture, decentralization, and operational efficiency in Ethereum's block space allocation.

Many problems are NP-hard and, unless P = NP, do not admit polynomial-time exact algorithms. The fastest known exact algorithms exactly usually take time exponential in the input size. Much research effort has gone into obtaining faster exact algorithms for instances that are sufficiently well-structured, e.g., through parameterized algorithms with running time $f(k)\cdot n^{\mathcal{O}(1)}$ where n is the input size and k quantifies some structural property such as treewidth. When k is small, this is comparable to a polynomial-time exact algorithm and outperforms the fastest exact exponential-time algorithms for a large range of k. In this work, we are interested instead in leveraging instance structure for polynomial-time approximation algorithms. We aim for polynomial-time algorithms that produce a solution of value at most or at least (depending on minimization vs. maximization) $c\mathrm{OPT}\pm f(k)$ where c is a constant. Unlike for standard parameterized algorithms, we do not assume that structural information is provided with the input. Ideally, we can obtain algorithms with small additive error, i.e., $c=1$ and $f(k)$ is polynomial or even linear in $k$. For small k, this is similarly comparable to a polynomial-time exact algorithm and will beat general case approximation for a large range of k. We study Vertex Cover, Connected Vertex Cover, Chromatic Number, and Triangle Packing. The parameters we consider are the size of minimum modulators to graph classes on which the respective problem is tractable. For most problem-parameter combinations we give algorithms that compute a solution of size at least or at most $\mathrm{OPT}\pm k$. In the case of Vertex Cover, most of our algorithms are tight under the Unique Games Conjecture and provide better approximation guarantees than standard 2-approximations if the modulator is smaller than the optimum solution.

In retail warehouses, returned products are typically placed in an intermediate storage until a decision regarding further shipment to stores is made. The longer products are held in storage, the higher the inefficiency and costs of the returns management process, since enough storage area has to be provided and maintained while the products are not placed for sale. To reduce the average product storage time, we consider an alternative solution where reallocation decisions for products can be made instantly upon their arrival in the warehouse allowing only a limited number of products to still be stored simultaneously. We transfer the problem to an online multiple knapsack problem and propose a novel reinforcement learning approach to pack the items (products) into the knapsacks (stores) such that the overall value (expected revenue) is maximized. Empirical evaluations on simulated data demonstrate that, compared to the usual offline decision procedure, our approach comes with a performance gap of only 3% while significantly reducing the average storage time of a product by 96%.

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 3,000 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai., Comment: 26 pages, 6 figures

Federated Learning (FL) is a distributed machine learning strategy, developed for settings where training data is owned by distributed devices and cannot be shared. FL circumvents this constraint by carrying out model training in distribution. The parameters of these local models are shared intermittently among participants and aggregated to enhance model accuracy. This strategy has been rapidly adopted by the industry in efforts to overcome privacy and resource constraints in model training. However, the application of FL to real-world settings brings additional challenges associated with heterogeneity between participants. Research into mitigating these difficulties in FL has largely focused on only two types of heterogeneity: the unbalanced distribution of training data, and differences in client resources. Yet more types of heterogeneity are becoming relevant as the capability of FL expands to cover more complex problems, from the tuning of LLMs to enabling machine learning on edge devices. In this work, we discuss a novel type of heterogeneity that is likely to become increasingly relevant in future applications: this is preference heterogeneity, emerging when clients learn under multiple objectives, with different importance assigned to each objective on different clients. In this work, we discuss the implications of this type of heterogeneity and propose FedPref, a first algorithm designed to facilitate personalised FL in this setting. We demonstrate the effectiveness of the algorithm across different problems, preference distributions and model architectures. In addition, we introduce a new analytical point of view, based on multi-objective metrics, for evaluating the performance of FL algorithms in this setting beyond the traditional client-focused metrics. We perform a second experimental analysis based in this view, and show that FedPref outperforms compared algorithms., Comment: Accepted to ACM ToMPECS journal

Active Inference, grounded in the Free Energy Principle, provides a powerful lens for understanding how agents balance exploration and goal-directed behavior in uncertain environments. Here, we propose a new planning framework, that integrates Monte Carlo Tree Search (MCTS) with active inference objectives to systematically reduce epistemic uncertainty while pursuing extrinsic rewards. Our key insight is that MCTS already renowned for its search efficiency can be naturally extended to incorporate free energy minimization by blending expected rewards with information gain. Concretely, the Cross-Entropy Method (CEM) is used to optimize action proposals at the root node, while tree expansions leverage reward modeling alongside intrinsic exploration bonuses. This synergy allows our planner to maintain coherent estimates of value and uncertainty throughout planning, without sacrificing computational tractability. Empirically, we benchmark our planner on a diverse set of continuous control tasks, where it demonstrates performance gains over both standalone CEM and MCTS with random rollouts.

Self-supervised learning (SSL) aims to find meaningful representations from unlabeled data by encoding semantic similarities through data augmentations. Despite its current popularity, theoretical insights about SSL are still scarce. For example, it is not yet known whether commonly used SSL loss functions can be related to a statistical model, much in the same as OLS, generalized linear models or PCA naturally emerge as maximum likelihood estimates of an underlying generative process. In this short paper, we consider a latent variable statistical model for SSL that exhibits an interesting property: Depending on the informativeness of the data augmentations, the MLE of the model either reduces to PCA, or approaches a simple non-contrastive loss. We analyze the model and also empirically illustrate our findings.

This study analyzes proposer-builder data and MEV-Boost payment data following the Ethereum merge in September 2022 to identify patterns of multi-block MEV. Our findings reveal fewer multi-slot sequences of builders than predicted by a random Monte Carlo simulation, with the longest observed sequence spanning 25 slots. Additionally, we observe that average MEV-Boost payments increase with the length of consecutive sequences, from approximately 0.05 ETH for single slots to 0.08 ETH for nine consecutive slots. Within longer sequences, payments per slot show a slight increase, suggesting that builders bid higher for longer sequences or the first slot after a longer sequence. A weak positive autocorrelation is found between subsequent MEV-Boost payments, challenging the hypothesis of alternating periods of low and high MEV. Finally, our comparison of builders during periods of low and high base fee volatility reveals minimal correlation, indicating the absence of builder specialization based on base fee volatility.

Automated Urban Air Mobility (UAM) can improve passenger transportation and reduce congestion, but its success depends on passenger trust. While initial research addresses passengers' information needs, questions remain about how to simulate air taxi flights and how these simulations impact users and interface requirements. We conducted a between-subjects study (N=40), examining the influence of motion fidelity in Virtual-Reality-simulated air taxi flights on user effects and interface design. Our study compared simulations with and without motion cues using a 3-Degrees-of-Freedom motion chair. Optimizing the interface design across six objectives, such as trust and mental demand, we used multi-objective Bayesian optimization to determine the most effective design trade-offs. Our results indicate that motion fidelity decreases users' trust, understanding, and acceptance, highlighting the need to consider motion fidelity in future UAM studies to approach realism. However, minimal evidence was found for differences or equality in the optimized interface designs, suggesting personalized interface designs.

Detecting the first generation of stars, Population III (PopIII), has been a long-standing goal in astrophysics, yet they remain elusive even in the JWST era. Here we present a novel NIRCam-based selection method for PopIII galaxies, and carefully validate it through completeness and contamination simulations. We systematically search ~500 arcmin$^{2}$ across JWST legacy fields for PopIII candidates, including GLIMPSE which, assisted by gravitational lensing, has produced JWST's deepest NIRCam imaging thus far. We discover one promising PopIII galaxy candidate (GLIMPSE-16043) at $z=6.50^{+0.03}_{-0.24}$, a moderately lensed galaxy (mu=2.9) with an intrinsic UV magnitude of $M_{UV}$=-15.89. It exhibits key PopIII features: strong H$\alpha$ emission (rest-frame EW $2810\pm550$\AA); a Balmer jump; no dust (UV slope $\beta=-2.34\pm0.36$); and undetectable metal lines (e.g., [OIII]; [OIII]/H$\beta$<0.44) implying a gas-phase metallicity of Zgas/Zsun<0.5%. These properties indicate the presence of a nascent, metal-deficient young stellar population (<5Myr) with a stellar mass of $\simeq10^{5}M_{\odot}$. Intriguingly, this source deviates significantly from the extrapolated UV-metallicity relation derived from recent JWST observations at $z=4-10$, consistent with UV enhancement by a top-heavy PopIII initial mass function or the presence of an extremely metal-poor AGN. We also derive the first observational constraints on the PopIII UV luminosity function at z~6-7. The volume density of GLIMPSE-16043 ($\approx10^{-4}$ cMpc$^{-3}$) is in excellent agreement with theoretical predictions, independently reinforcing its plausibility. This study demonstrates the power of our novel NIRCam method to finally reveal distant galaxies even more pristine than the Milky Way's most metal-poor satellites, thereby promising to bring us closer to the first generation of stars than we have ever been before., Comment: 41 pages, 20 figures, and 10 Tables. Submitted to ApJ. Comments are welcome!

Systemic drug administration often causes off-target effects limiting the efficacy of advanced therapies. Targeted drug delivery approaches increase local drug concentrations at the diseased site while minimizing systemic drug exposure. We present a magnetically guided microrobotic drug delivery system capable of precise navigation under physiological conditions. This platform integrates a clinical electromagnetic navigation system, a custom-designed release catheter, and a dissolvable capsule for accurate therapeutic delivery. In vitro tests showed precise navigation in human vasculature models, and in vivo experiments confirmed tracking under fluoroscopy and successful navigation in large animal models. The microrobot balances magnetic material concentration, contrast agent loading, and therapeutic drug capacity, enabling effective hosting of therapeutics despite the integration complexity of its components, offering a promising solution for precise targeted drug delivery.

The absence of a human operator in automated vehicles (AVs) may require external Human-Machine Interfaces (eHMIs) to facilitate communication with other road users in uncertain scenarios, for example, regarding the right of way. Given the plethora of adjustable parameters, balancing visual and auditory elements is crucial for effective communication with other road users. With N=37 participants, this study employed multi-objective Bayesian optimization to enhance eHMI designs and improve trust, safety perception, and mental demand. By reporting the Pareto front, we identify optimal design trade-offs. This research contributes to the ongoing standardization efforts of eHMIs, supporting broader adoption., Comment: Accepted at CHI 2025

Metallicity is a crucial tracer of galaxy evolution, providing insights into gas accretion, star formation, and feedback. At high redshift, these processes reveal how early galaxies assembled and enriched their interstellar medium. In this work, we present rest-frame optical spectroscopy of 12 massive ($\log(M_*/\mathrm{M_{\odot}})>9$) galaxies at $z\sim 6$-$8$ from the REBELS ALMA large program, observed with JWST NIRSpec/IFU in the prism mode. These observations span emission lines from [OII]$\lambda$3727,9 to [SII]$\lambda$6716,31, providing key information on nebular dust attenuation, ionisation states, and chemical abundances. We find lower O32 ratios (average $\sim3.7$) and [OIII]$\lambda$5007 equivalent widths (average ${EW_{[OIII]}}\sim390$\AA) than are generally found in existing large spectroscopic surveys at $z>6$, indicating less extreme ionising conditions. Strong-line diagnostics suggest that these systems are some of the most metal-rich galaxies observed at $z>6$ (average $Z_{\mathrm{gas}}\sim 0.4 Z_{\odot}$), including sources with near-solar oxygen abundances, in line with their high stellar masses (average $\log{M_*/\mathrm{M_{\odot}}}\sim9.5$). Supplementing with literature sources at lower masses, we investigate the mass-metallicity and fundamental metallicity relations (MZR and FMR, respectively) over a 4 dex stellar mass range at $63$ studies., Comment: Submitted to MNRAS, 22 pages

The theory of Leonard triples is applied to the derivation of normalized scalar products of on-shell and off-shell Bethe states generated from a Leonard pair. The scalar products take the form of linear combinations of $q$-Racah polynomials with coefficients depending on the off-shell parameters. Upon specializations, explicit solutions for the corresponding Belliard-Slavnov linear systems are obtained. It implies the existence of a determinant formula in terms of inhomogeneous Bethe roots for the $q$-Racah polynomials. Also, a set of relations that determines solutions (Bethe roots) of the corresponding Bethe equations of inhomogeneous type in terms of solutions of Bethe equations of homogenous type is obtained., Comment: 26pp

Future multifunction RF systems will be able to not only perform various different radar, communication and electronic warfare functionalities but also to perform them simultaneously on the same aperture. This ability of concurrent operations requires new, cognitive approaches of resource management compared to classical methods. This paper presents such a new approach using a combination of quality of service based resource management and Monte Carlo tree search., Comment: 6 pages, 9 figures, presented at 2024 International Radar Conference (RADAR2024)

Recent studies suggest that the candidate Kitaev magnet Na$_2$Co$_2$TeO$_6$ possesses novel triple-$\mathbf{q}$ magnetic order instead of conventional single-$\mathbf{q}$ zigzag order. Here we present dedicated experiments in search for distinct properties expected of the triple-$\mathbf{q}$ order, namely, insensitivity of the magnetic domains to weak $C_3$ symmetry-breaking fields and fictitious magnetic fields generated by the spin vorticity. In structurally pristine single crystals, we show that $C_3$ symmetry-breaking in-plane uniaxial strains do not affect the order's magnetic neutron diffraction signals. We further show that $\mathbf{c}$-axis propagating light exhibits large Faraday rotations in the ordered state due to the spin vorticity, the sign of which can be trained via the system's ferrimagnetic moment. These results are in favor of the triple-$\mathbf{q}$ order in Na$_2$Co$_2$TeO$_6$ and reveal its unique emerging behavior., Comment: 4 pages, 4 figures, plus Supplemental Material

Generative modeling aims to transform random noise into structured outputs. In this work, we enhance video diffusion models by allowing motion control via structured latent noise sampling. This is achieved by just a change in data: we pre-process training videos to yield structured noise. Consequently, our method is agnostic to diffusion model design, requiring no changes to model architectures or training pipelines. Specifically, we propose a novel noise warping algorithm, fast enough to run in real time, that replaces random temporal Gaussianity with correlated warped noise derived from optical flow fields, while preserving the spatial Gaussianity. The efficiency of our algorithm enables us to fine-tune modern video diffusion base models using warped noise with minimal overhead, and provide a one-stop solution for a wide range of user-friendly motion control: local object motion control, global camera movement control, and motion transfer. The harmonization between temporal coherence and spatial Gaussianity in our warped noise leads to effective motion control while maintaining per-frame pixel quality. Extensive experiments and user studies demonstrate the advantages of our method, making it a robust and scalable approach for controlling motion in video diffusion models. Video results are available on our webpage: https://eyeline-research.github.io/Go-with-the-Flow. Source code and model checkpoints are available on GitHub: https://github.com/Eyeline-Research/Go-with-the-Flow.

We propose a novel data-driven method called QENDy (Quadratic Embedding of Nonlinear Dynamics) that not only allows us to learn quadratic representations of highly nonlinear dynamical systems, but also to identify the governing equations. The approach is based on an embedding of the system into a higher-dimensional feature space in which the dynamics become quadratic. Just like SINDy (Sparse Identification of Nonlinear Dynamics), our method requires trajectory data, time derivatives for the training data points, which can also be estimated using finite difference approximations, and a set of preselected basis functions, called dictionary. We illustrate the efficacy and accuracy of QENDy with the aid of various benchmark problems and compare its performance with SINDy and a deep learning method for identifying quadratic embeddings. Furthermore, we analyze the convergence of QENDy and SINDy in the infinite data limit, highlight their similarities and main differences, and compare the quadratic embedding with linearization techniques based on the Koopman operator.

Understanding how molecular changes caused by genetic variation drive disease risk is crucial for deciphering disease mechanisms. However, interpreting genome sequences is challenging because of the vast size of the human genome, and because its consequences manifest across a wide range of cells, tissues and scales -- spanning from molecular to whole organism level. Here, we present Phenformer, a multi-scale genetic language model that learns to generate mechanistic hypotheses as to how differences in genome sequence lead to disease-relevant changes in expression across cell types and tissues directly from DNA sequences of up to 88 million base pairs. Using whole genome sequencing data from more than 150 000 individuals, we show that Phenformer generates mechanistic hypotheses about disease-relevant cell and tissue types that match literature better than existing state-of-the-art methods, while using only sequence data. Furthermore, disease risk predictors enriched by Phenformer show improved prediction performance and generalisation to diverse populations. Accurate multi-megabase scale interpretation of whole genomes without additional experimental data enables both a deeper understanding of molecular mechanisms involved in disease and improved disease risk prediction at the level of individuals.

We study the vectorial stationary Schr\"odinger equation $-\Delta u+a\,U+b\,u=F,$ with a saturated nonlinearity $U=u/|u|$ and with some complex coefficients $(a,b)\in\mathbb{C}^2$. Besides the existence and uniqueness of solutions for the Dirichlet and Neumann problems, we prove the compactness of the support of the solution, under suitable conditions on $(a,b)$ and even when the source in the right hand side $F(x)$ is not vanishing for large values of $|x|.$ The proof of the compactness of the support uses a local energy method, given the impossibility of applying the maximum principle. We also consider the associate Schr\"{o}dinger-Poisson system when coupling with a simple magnetic field. Among other consequences, our results give a rigorous proof of the existence of ``solitons with compact support" claimed, without any proof, by several previous authors.

Automated vehicle (AV) acceptance relies on their understanding via feedback. While visualizations aim to enhance user understanding of AV's detection, prediction, and planning functionalities, establishing an optimal design is challenging. Traditional "one-size-fits-all" designs might be unsuitable, stemming from resource-intensive empirical evaluations. This paper introduces OptiCarVis, a set of Human-in-the-Loop (HITL) approaches using Multi-Objective Bayesian Optimization (MOBO) to optimize AV feedback visualizations. We compare conditions using eight expert and user-customized designs for a Warm-Start HITL MOBO. An online study (N=117) demonstrates OptiCarVis's efficacy in significantly improving trust, acceptance, perceived safety, and predictability without increasing cognitive load. OptiCarVis facilitates a comprehensive design space exploration, enhancing in-vehicle interfaces for optimal passenger experiences and broader applicability.

Laser light with squeezed quantum uncertainty is a powerful tool for interferometric sensing. A routine application can be found in gravitational wave observatories. A significant quantum advantage is only achievable if a large fraction of the photons are actually measured. For this reason, quantum-enhanced vibrational measurements of strongly absorbing or scattering surfaces have not been considered so far. Here we demonstrate the strongly quantum-enhanced measurement of the frequency characteristics of surface vibrations in air by measuring the air pressure wave instead. Our squeezed laser beam, which simply passes the vibrating surface, delivers a sensitivity that an ultra-stable conventional light beam in the same configuration can only achieve with ten times the power. The pressure amplitude of a ultrasonic wave of just 0.12 mPa/ Hz was clearly visible with a spatial resolution in the millimetre range and a 1 kHz resolution bandwidth. We envision applications in sensor technology where distant, highly absorbing or optically inaccessible surface vibrations in air are to be measured with limited, e.g. eye-safe, light powers.

Over the past decade, autonomous stabilization of bosonic qubits has emerged as a promising approach for hardware-efficient protection of quantum information. However, applying these techniques to more complex encodings than the Schr\"odinger cat code requires exquisite control of high-order wave mixing processes. The challenge is to enable specific multiphotonic dissipation channels while avoiding unintended non-linear interactions. In this work, we leverage a genuine six-wave mixing process enabled by a near Kerr-free Josephson element to enforce dissipation of quartets of excitations in a high-impedance superconducting resonator. Owing to residual non-linearities stemming from stray inductances in our circuit, this dissipation channel is only effective when the resonator holds a specific number of photons. Applying it to the fourth excited state of the resonator, we show an order of magnitude enhancement of the state decay rate while only marginally impacting the relaxation and coherence of lower energy states. Given that stray inductances could be strongly reduced through simple modifications in circuit design and that our methods can be adapted to activate even higher-order dissipation channels, these results pave the way toward the dynamical stabilization of four-component Schr\"odinger cat qubits and even more complex bosonic qubits.

We consider inner functions $\Theta$ with the zero set $\mathcal Z(\Theta)$ such that the quotient algebra $H^\infty / \Theta H^\infty$ satisfies the Strong Invertibility Property (SIP), that is for every $\varepsilon>0$ there exists $\delta>0$ such that the conditions $f \in H^\infty$, $\|[f]\|_{H^\infty/ \Theta H^\infty}=1$, $\inf_{\mathcal Z(\Theta)} |f| \ge 1-\delta$ imply that $[f]$ is invertible in $H^\infty / \Theta H^\infty$ and $\| 1/ [f] \|_{H^\infty/ \Theta H^\infty}\le 1+\varepsilon$. We prove that the SIP is equivalent to the maximal asymptotic growth of $\Theta $ away from its zero set. We also describe inner functions satisfying the SIP in terms of the narrowness of their sublevel sets and relate the SIP to the Weak Embedding Property introduced by P.Gorkin, R.Mortini, and N.Nikolski as well as to inner functions whose Frostman shifts are Carleson--Newman Blaschke products. We finally study divisors of inner functions satisfying the SIP. We describe geometrically the zero set of inner functions such that all its divisors satisfy the SIP. We also prove that a closed subset $E$ of the unit circle is of finite entropy if and only if any singular inner function associated to a singular measure supported on $E$ is a divisor of an inner function satisfying the SIP.

Whether deep neural networks can exhibit emergent behaviour is not only relevant for understanding how deep learning works, it is also pivotal for estimating potential security risks of increasingly capable artificial intelligence systems. Here, we show that training deep neural networks gives rise to emergent weight morphologies independent of the training data. Specifically, in analogy to condensed matter physics, we derive a theory that predict that the homogeneous state of deep neural networks is unstable in a way that leads to the emergence of periodic channel structures. We verified these structures by performing numerical experiments on a variety of data sets. Our work demonstrates emergence in the training of deep neural networks, which impacts the achievable performance of deep neural networks.

We study strategic location choice by customers and sellers, termed the Bakers and Millers Game in the literature. In our generalized setting, each miller can freely choose any location for setting up a mill, while each baker is restricted in the choice of location for setting up a bakery. For optimal bargaining power, a baker would like to select a location with many millers to buy flour from and with little competition from other bakers. Likewise, a miller aims for a location with many bakers and few competing millers. Thus, both types of agents choose locations to optimize the ratio of agents of opposite type divided by agents of the same type at their chosen location. Originally raised in the context of Fractional Hedonic Games, the Bakers and Millers Game has applications that range from commerce to product design. We study the impact of location restrictions on the properties of the game. While pure Nash equilibria trivially exist in the setting without location restrictions, we show via a sophisticated, efficient algorithm that even the more challenging restricted setting admits equilibria. Moreover, the computed equilibrium approximates the optimal social welfare by a factor of at most $2\left(\frac{e}{e-1}\right)$. Furthermore, we give tight bounds on the price of anarchy/stability. On the conceptual side, the location choice feature adds a new layer to the standard setting of Hedonic Games, in the sense that agents that select the same location form a coalition. This allows to naturally restrict the possible coalitions that can be formed. With this, our model generalizes simple symmetric Fractional Hedonic Games on complete bipartite valuation graphs and also Hedonic Diversity Games with utilities single-peaked at 0. We believe that this generalization is also a very interesting direction for other types of Hedonic Games., Comment: To appear at the 24th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2025)

It is established that the linear spectral statistics (LSS) of the smoothed periodogram estimate of the spectral coherence matrix of a complex Gaussian high-dimensional times series (yn) n$\in$Z with independent components satisfy at each frequency a central limit theorem in the asymptotic regime where the sample size N , the dimension M of the observation, and the smoothing span B both converge towards +$\infty$ in such a way that M = O(N $\alpha$ ) for $\alpha$ \< 1 and M B $\rightarrow$ c, c $\in$ (0, 1). It is deduced that two recentered and renormalized versions of the LSS, one based on an average in the frequency domain and the other one based on a sum of squares also in the frequency domain, and both evaluated over a well-chosen frequency grid, also verify a central limit theorem. These two statistics are proposed to test with controlled asymptotic level the hypothesis that the components of y are independent. Numerical simulations assess the performance of the two tests.

Fluorescence microscopes can record the dynamics of living cells with high spatio-temporal resolution in a single plane. However, monitoring rapid and dim fluorescence fluctuations, e.g induced by neuronal activity in the brain, remains challenging for 3D-distributed emitters due to out-of-focus fluorescence background, a restricted photon budget, and the speed limit of conventional scanning systems. Here, we introduce a Thermally Adaptive Surface strategy, capable of simultaneously recording, at camera framerate, the activity of 3D-distributed objects. This innovative microscope leverages on an array of thermally tuneable microlenses that offer low chromatic aberration and high transmission, and can be combined with patterned illumination to provide optical sectioning. We demonstrate its potential in vivo, by simultaneously monitoring fast fluorescent dynamics at different depths in the zebrafish larval brain, at a rate of 0.5 kHz and over a large field of view (360um x 360um).

Most observed exoplanets have high equilibrium temperatures. Understanding the chemistry of their atmospheres and interpreting their observations requires the use of chemical kinetic models including photochemistry. The thermal dependence of the vacuum ultraviolet (VUV) absorption cross sections of molecules used in these models is poorly known at high temperatures, leading to uncertainties in the resulting abundance profiles. The aim of our work is to study experimentally the thermal dependence of VUV absorption cross sections of molecules of interest for exoplanet atmospheres and provide accurate data for use in atmospheric models. This study focuses on acetylene (C2H2). We measured absorption cross sections of C2H2 at seven temperatures ranging from 296 to 773 K recorded in the 115-230 nm spectral domain using VUV spectroscopy and synchrotron radiation. These data were used in our 1D thermo-photochemical model, to assess their impact on the predicted composition of a generic hot Jupiter-like exoplanet atmosphere. The absolute absorption cross sections of C2H2 increase with temperature. This increase is relatively constant from 115 to 185 nm and rises sharply from 185 to 230 nm. The abundance profile of C2H2 calculated using the model shows a slight variation, with a maximum decrease of 40% near 5 x 10-5 bar, when using C2H2 absorption cross sections measured at 773 K compared to those at 296 K. This is explained by the absorption, higher in the atmosphere, of the actinic flux from 150 to 230 nm due to the increase in the C2H2 absorption in this spectral range. This change also impacts the abundance profiles of other by-products such as methane (CH4) and ethylene (C2H4). We present the first experimental measurements of the VUV absorption cross sections of C2H2 at high temperatures. Similar studies of other major species are needed to improve our understanding of exoplanet atmospheres., Comment: 12 pages, 1 table, 11 figures

This article introduces the concept of optimization learning, a methodology to design optimization proxies that learn the input/output mapping of parametric optimization problems. These optimization proxies are trustworthy by design: they compute feasible solutions to the underlying optimization problems, provide quality guarantees on the returned solutions, and scale to large instances. Optimization proxies are differentiable programs that combine traditional deep learning technology with repair or completion layers to produce feasible solutions. The article shows that optimization proxies can be trained end-to-end in a self-supervised way. It presents methodologies to provide performance guarantees and to scale optimization proxies to large-scale optimization problems. The potential of optimization proxies is highlighted through applications in power systems and, in particular, real-time risk assessment and security-constrained optimal power flow.

LiDAR scenes constitute a fundamental source for several autonomous driving applications. Despite the existence of several datasets, scenes from adverse weather conditions are rarely available. This limits the robustness of downstream machine learning models, and restrains the reliability of autonomous driving systems in particular locations and seasons. Collecting feature-diverse scenes under adverse weather conditions is challenging due to seasonal limitations. Generative models are therefore essentials, especially for generating adverse weather conditions for specific driving scenarios. In our work, we propose a latent diffusion process constituted by autoencoder and latent diffusion models. Moreover, we leverage the clear condition LiDAR scenes with a postprocessing step to improve the realism of the generated adverse weather condition scenes., Comment: This is an intermediate version of our work

Transformer-based models for time series forecasting (TSF) have attracted significant attention in recent years due to their effectiveness and versatility. However, these models often require extensive hyperparameter optimization (HPO) to achieve the best possible performance, and a unified pipeline for HPO in transformer-based TSF remains lacking. In this paper, we present one such pipeline and conduct extensive experiments on several state-of-the-art (SOTA) transformer-based TSF models. These experiments are conducted on standard benchmark datasets to evaluate and compare the performance of different models, generating practical insights and examples. Our pipeline is generalizable beyond transformer-based architectures and can be applied to other SOTA models, such as Mamba and TimeMixer, as demonstrated in our experiments. The goal of this work is to provide valuable guidance to both industry practitioners and academic researchers in efficiently identifying optimal hyperparameters suited to their specific domain applications. The code and complete experimental results are available on GitHub.

For the first time the neutron-rich hydrogen isotope $\rm ^{6}H$ was produced in an electron scattering experiment in the reaction $\rm ^{7}Li(e,~e'p\pi^{+})^{6}H$ using the spectrometer facility of the A1 Collaboration at the Mainz Microtron accelerator. By measuring the triple coincidence between the scattered electron, the produced proton, and $\pi^{+}$, the missing mass spectrum of $\rm ^{6}H$ was obtained. A clear peak above $^3$H+n+n+n energy threshold was seen resulting in a ground state energy of $\rm ^{6}H$ at $2.3\pm0.5({\rm stat.})\pm0.4({\rm syst.})$ MeV with a width of $1.9\pm1.0({\rm stat.})\pm0.4({\rm syst.})$ MeV. This work challenges the understandings of multi-nucleon interactions and presents a new method to study light neutron-rich nuclei with electron scattering experiments., Comment: 6 pages, 4 figures

The newly accessible mid-infrared (MIR) window offered by the James Webb Space Telescope (JWST) for exoplanet imaging is expected to provide valuable information to characterize their atmospheres. In particular, coronagraphs on board the JWST Mid-InfraRed instrument (MIRI) are capable of imaging the coldest directly imaged giant planets at the wavelengths where they emit most of their flux. The MIRI coronagraphs have been specially designed to detect the NH3 absorption around 10.5 microns, which has been predicted by atmospheric models. We aim to assess the presence of NH3 while refining the atmospheric parameters of one of the coldest companions detected by directly imaging GJ 504 b. Its mass is still a matter of debate and depending on the host star age estimate, the companion could either be placed in the brown dwarf regime or in the young Jovian planet regime. We present an analysis of MIRI coronagraphic observations of the GJ 504 system. We took advantage of previous observations of reference stars to build a library of images and to perform a more efficient subtraction of the stellar diffraction pattern. We detected the presence of NH3 at 12.5 sigma in the atmosphere, in line with atmospheric model expectations for a planetary-mass object and observed in brown dwarfs within a similar temperature range. The best-fit model with Exo-REM provides updated values of its atmospheric parameters, yielding a temperature of Teff = 512 K and radius of R = 1.08 RJup. These observations demonstrate the capability of MIRI coronagraphs to detect NH3 and to provide the first MIR observations of one of the coldest directly imaged companions. Overall, NH3 is a key molecule for characterizing the atmospheres of cold planets, offering valuable insights into their surface gravity. These observations provide valuable information for spectroscopic observations planned with JWST., Comment: Published in A&A

In this paper, we develop a universal, conceptually simple and systematic method to prove well-posedness to Cauchy problems for weak solutions of parabolic equations with non-smooth, time-dependent, elliptic part having a variational definition. Our classes of weak solutions are taken with minimal assumptions. We prove the existence and uniqueness of a fundamental solution which seems new in this generality: it is shown to always coincide with the associated evolution family for the initial value problem with zero source and it yields representation of all weak solutions. Our strategy is a variational approach avoiding density arguments, a priori regularity of weak solutions or regularization by smooth operators. One of our main tools are embedding results which yield time continuity of our weak solutions going beyond the celebrated Lions regularity theorem and that is addressing a variety of source terms. We illustrate our results with three concrete applications : second order uniformly elliptic part with Dirichlet boundary condition on domains, integro-differential elliptic part, and second order degenerate elliptic part., Comment: 39 pages. Comments are welcome

This article deals with the stationary Gross-Pitaevskii non-linear eigenvalue problem in the presence of a rotating magnetic field that is used to model macroscopic quantum effects such as Bose-Einstein condensates (BECs). In this regime, the ground-state wave-function can exhibit an a priori unknown number of quantum vortices at unknown locations, which necessitates the exploitation of adaptive numerical strategies. To this end, we consider the conforming finite element method and introduce a combination of a Sobolev gradient descent that respects the energy-topology of the problem to solve the non-linearity and an $hp$-adaptive strategy that is solely based on energy decay rather than a posteriori error estimators for the refinement process. Numerical results demonstrate that the $hp$-adaptive strategy is highly efficient in terms of accuracy to compute the ground-state wave function and energy for several test problems where we observe exponential convergence.

In the context of kilonova (KN) modeling, the present work focusses on large-scale atomic data and opacity computations for all heavy elements from Ca to Lr, with a special effort on lanthanides and actinides, for a grid of typical KN ejecta conditions between one day and one week after the merger (corresponding to the LTE photosphere phase of the KN ejecta). In order to do so, we used the pseudo-relativistic Hartree-Fock (HFR) method, in which the choice of the interaction configuration model is of crucial importance. In this paper, HFR atomic data and opacities for all elements between Ca (Z = 20) and Lr (Z = 103) are presented, with a special focus on lanthanides and actinides. Besides, we also discuss the contribution of every single element to the total KN ejecta opacity for a given neutron star merger model, depending on their Planck mean opacities and elemental abundances. An important result is that lanthanides are found to not be the dominant sources of opacity, at least on average. The impact on KN light curves of considering such atomic-physics based opacity data instead of typical crude approximation formulae is also evaluated. In addition, the importance of taking the ejecta composition into account directly in the expansion opacity determination (instead of estimating single-element opacities) is highlighted. A database containing all the relevant atomic data and opacity tables has also been created and published online along with this work.

The electrocaloric effect refers to the temperature change in a material when an electric field is applied or removed. Significant breakthroughs revealed its potential for solid-state cooling technologies in past decades. These devices offer a sustainable alternative to traditional vapor compression refrigeration, with advantages such as compactness, silent operation, and the absence of moving parts or refrigerants. Electrocaloric effects are typically studied using indirect methods using polarization data, and which suffer from inaccuracies related to assumptions about heat capacity. Direct methods, although more precise, require device fabrication and face challenges in studying meso- or nanoscale systems, like 2D materials, and materials with non-uniform polarization textures where high spatial resolution is required. In this study, a novel technique, Scanning Electrocaloric Thermometry, is introduced for characterizing the local electrocaloric effect in nanomaterials. This approach achieves high spatial resolution by locally applying electric fields and by simultaneously measuring the resulting temperature change. By employing AC excitation, the measurement sensitivity is further enhanced and the electrocaloric effect is disentangled from other heating mechanisms such as Joule heating and dielectric losses. The effectiveness of the method is demonstrated by examining electrocaloric and heat dissipation phenomena in two-dimensional In${}_2$Se${}_3$ micrometer-sized flakes.

Adaptation of behavior requires the brain to change goals in a changing environment. Synaptic learning has demonstrated its effectiveness in changing the probability of selecting actions based on their outcome. In the extreme case, it is vital not to repeat an action to a given goal that led to harmful punishment. The present model proposes a simple neural mechanism of gain modulation that makes possible immediate changes in the probability of selecting a goal after punishment of variable intensity. Results show how gain modulation determine the type of elementary navigation process within the state space of a network of neuronal populations of excitatory neurons regulated by inhibition. Immediately after punishment, the system can avoid the punished populations by going back or by jumping to unpunished populations. This does not require particular credit assignment at the `choice' population but only gain modulation of neurons active at the time of punishment. Gain modulation does not require statistical relearning that may lead to further errors, but can encode memories of past experiences without modification of synaptic efficacies. Therefore, gain modulation can complements synaptic plasticity.

Although distributed machine learning (distributed ML) is gaining considerable attention in the community, prior works have independently looked at instances of distributed ML in either the training or the inference phase. No prior work has examined the combined robustness stemming from distributing both the learning and the inference process. In this work, we explore, for the first time, the robustness of distributed ML models that are fully heterogeneous in training data, architecture, scheduler, optimizer, and other model parameters. Supported by theory and extensive experimental validation using CIFAR10 and FashionMNIST, we show that such properly distributed ML instantiations achieve across-the-board improvements in accuracy-robustness tradeoffs against state-of-the-art transfer-based attacks that could otherwise not be realized by current ensemble or federated learning instantiations. For instance, our experiments on CIFAR10 show that for the Common Weakness attack, one of the most powerful state-of-the-art transfer-based attacks, our method improves robust accuracy by up to 40%, with a minimal impact on clean task accuracy., Comment: To appear in the Proceedings of the AAAI Conference on Artificial Intelligence (AAAI) 2025