Search

This paper introduces madupite, a high-performance distributed solver for large-scale Markov Decision Processes (MDPs). MDPs are widely used to model complex dynamical systems in various fields, including finance, epidemiology, and traffic control. However, real-world applications often result in extremely high-dimensional MDPs, leading to the curse of dimensionality, which is typically addressed through function approximators like neural networks. While existing solvers such as pymdptoolbox and mdpsolver provide tools for solving MDPs, they either lack scalability, support for distributed computing, or flexibility in solution methods. madupite is designed to overcome these limitations by leveraging modern high-performance computing resources. It efficiently distributes memory load and computation across multiple nodes, supports a diverse set of solution methods, and offers a user-friendly Python API while maintaining a C++ core for optimal performance. With the ability to solve MDPs with millions of states, madupite provides researchers and engineers with a powerful tool to tackle large-scale decision-making problems with greater efficiency and flexibility.

This paper proposes a stochastic model predictive control method for linear systems affected by additive Gaussian disturbances. Closed-loop satisfaction of probabilistic constraints and recursive feasibility of the underlying convex optimization problem is guaranteed. Optimization over feedback policies online increases performance and reduces conservatism compared to fixed-feedback approaches. The central mechanism is a finitely determined maximal admissible set for probabilistic constraints, together with the reconditioning of the predicted probabilistic constraints on the current knowledge at every time step. The proposed method's reduced conservatism and improved performance in terms of the achieved closed-loop cost is demonstrated in a numerical example.

Thermal light dark matter (LDM) with particle masses in the 1 MeV - 1 GeV range could successfully explain the observed dark matter abundance as a relic from the primordial Universe. In this picture, a new feeble interaction acts as a "portal" between the Standard Model and LDM particles, allowing for the exploration of this paradigm at accelerator experiments. In the last years, the "missing energy" experiment NA64e at CERN SPS (Super Proton Synchrotron) has set world-leading constraints in the vector-mediated LDM parameter space, by exploiting a 100 GeV electron beam impinging on an electromagnetic calorimeter, acting as an active target. In this paper, we report a detailed description of the analysis of a preliminary measurement with a 70 GeV positron beam at NA64e, performed during summer 2023 with an accumulated statistic of 1.6 x 10^10 positrons on target. This data set was analyzed with the primary aim of evaluating the performance of the NA64e detector with a lower energy positron beam, towards the realization of the post-LS3 program. The analysis results, other than additionally probing unexplored regions in the LDM parameter space, provide valuable information towards the future NA64e positron campaign.

This paper presents a novel dataset of public broadcast interviews featuring high-ranking German politicians. The interviews were sourced from YouTube, transcribed, processed for speaker identification, and stored in a tidy and open format. The dataset comprises 99 interviews with 33 different German politicians across five major interview formats, containing a total of 28,146 sentences. As the first of its kind, this dataset offers valuable opportunities for research on various aspects of political communication in the (German) political contexts, such as agenda-setting, interviewer dynamics, or politicians' self-presentation.

The Jansen-Rit model of a cortical column in the cerebral cortex is widely used to simulate spontaneous brain activity (EEG) and event-related potentials. It couples a pyramidal cell population with two interneuron populations, of which one is fast and excitatory and the other slow and inhibitory. Our paper studies the transition between alpha and delta oscillations produced by the model. Delta oscillations are slower than alpha oscillations and have a more complex relaxation-type time profile. In the context of neuronal population activation dynamics, a small threshold means that neurons begin to activate with small input or stimulus, indicating high sensitivity to incoming signals. A steep slope signifies that activation increases sharply as input crosses the threshold. Accordingly in the model the excitatory activation thresholds are small and the slopes are steep. Hence, a singular limit replacing the excitatory activation function with all-or-nothing switches, eg. a Heaviside function, is appropriate. In this limit we identify the transition between alpha and delta oscillations as a discontinuity-induced grazing bifurcation. At the grazing the minimum of the pyramidal-cell output equals the threshold for switching off the excitatory interneuron population, leading to a collapse in excitatory feedback.

We derive a Gutzwiller-type trace formula for quantum chaotic systems that accounts for both particle spin precession and discrete geometrical symmetries. This formula generalises previous results that were obtained either for systems with spin [1,2] or for systems with symmetries [3,4], but not for a combination of both. The derivation requires not only a combination of methodologies for these two settings, but also the treatment of new effects in the form of double groups and spin components of symmetry operations. The resulting trace formula expresses the level density of subspectra associated to irreducible representations of the group of unitary symmetries in terms of periodic orbits in the system's fundamental domain. We also derive a corresponding expression for the spectral determinant. In a follow-up paper [5] we will show that our formula allows to study the impact of geometrical symmetries and spin on spectral statistics., Comment: 23 pages

Bifurcation analysis collects techniques for characterizing the dependence of certain classes of solutions of a dynamical system on variations in problem parameters. Common solution classes of interest include equilibria and periodic orbits, the number and stability of which may vary as parameters vary. Continuation techniques generate continuous families of such solutions in the combined state and parameter space, e.g., curves (branches) of periodic orbits or surfaces of equilibria. Their advantage over simulation-based approaches is the ability to map out such families independently of the dynamic stability of the equilibria or periodic orbits. Bifurcation diagrams represent families of equilibria and periodic orbits as curves or surfaces in appropriate coordinate systems. Special points, such as bifurcations, are often highlighted in such diagrams. This article provides an illustration of this paradigm of synergy between theoretical derivations and computational analysis for several characteristic examples of bifurcation analysis in commonly encountered classes of problems. General theoretical principles are deduced from these illustrations and collected for the reader's subsequent reference., Comment: 31 pages, 12 figures, code for reproducing computational data for figures available at https://github.com/jansieber/DS-handbook24-odebifurcations-resources

Rank collapse, a phenomenon where embedding vectors in sequence models rapidly converge to a uniform token or equilibrium state, has recently gained attention in the deep learning literature. This phenomenon leads to reduced expressivity and potential training instabilities due to vanishing gradients. Empirical evidence suggests that architectural components like skip connections, LayerNorm, and MultiLayer Perceptrons (MLPs) play critical roles in mitigating rank collapse. While this issue is well-documented for transformers, alternative sequence models, such as State Space Models (SSMs), which have recently gained prominence, have not been thoroughly examined for similar vulnerabilities. This paper extends the theory of rank collapse from transformers to SSMs using a unifying framework that captures both architectures. We study how a parametrized version of the classic skip connection component, which we call \emph{lambda-skip connections}, provides guarantees for rank collapse prevention. Through analytical results, we present a sufficient condition to guarantee prevention of rank collapse across all the aforementioned architectures. We also study the necessity of this condition via ablation studies and analytical examples. To our knowledge, this is the first study that provides a general guarantee to prevent rank collapse, and that investigates rank collapse in the context of SSMs, offering valuable understanding for both theoreticians and practitioners. Finally, we validate our findings with experiments demonstrating the crucial role of architectural components such as skip connections and gating mechanisms in preventing rank collapse.

We prove convergence of piecewise polynomial collocation methods applied to periodic boundary value problems for functional differential equations with state-dependent delays. The state dependence of the delays leads to nonlinearities that are not locally Lipschitz continuous preventing the direct application of general abstract discretization theoretic frameworks. We employ a weaker form of differentiability, which we call mild differentiability, to prove that a locally unique solution of the functional differential equation is approximated by the solution of the discretized problem with the expected order., Comment: 26 pages, 1 figure

A search for Dark Sectors is performed using the unique M2 beam line at the CERN Super Proton Synchrotron. New particles ($X$) could be produced in the bremsstrahlung-like reaction of high energy 160 GeV muons impinging on an active target, $\mu N\rightarrow\mu NX$, followed by their decays, $X\rightarrow\text{invisible}$. The experimental signature would be a scattered single muon from the target, with about less than half of its initial energy and no activity in the sub-detectors located downstream the interaction point. The full sample of the 2022 run is analyzed through the missing energy/momentum channel, with a total statistics of $(1.98\pm0.02)\times10^{10}$ muons on target. We demonstrate that various muon-philic scenarios involving different types of mediators, such as scalar or vector particles, can be probed simultaneously with such a technique. For the vector-case, besides a $L_\mu-L_\tau$ $Z'$ vector boson, we also consider an invisibly decaying dark photon ($A'\rightarrow\text{invisible}$). This search is complementary to NA64 running with electrons and positrons, thus, opening the possibility to expand the exploration of the thermal light dark matter parameter space by combining the results obtained with the three beams., Comment: 24 pages, 22 figures

We consider the solution of an additive fractional stochastic differential equation (SDE) and, leveraging continuous observations of the process, introduce a methodology for estimating its stationary density $\pi$. Initially, employing a tailored martingale decomposition specifically designed for the statistical challenge at hand, we establish convergence rates surpassing those found in existing literature. Subsequently, we refine the attained rate for the case where $H < \frac{1}{2}$ by incorporating bounds on the density of the semi-group. This enhancement outperforms previous rates. Finally, our results weaken the usual convexity assumptions on the drift component, allowing to consider settings where strong convexity only holds outside a compact set.

Accurate prediction of drug-target interactions is critical for advancing drug discovery. By reducing time and cost, machine learning and deep learning can accelerate this laborious discovery process. In a novel approach, BarlowDTI, we utilise the powerful Barlow Twins architecture for feature-extraction while considering the structure of the target protein. Our method achieves state-of-the-art predictive performance against multiple established benchmarks using only one-dimensional input. The use of gradient boosting machine as the underlying predictor ensures fast and efficient predictions without the need for substantial computational resources. We also investigate how the model reaches its decision based on individual training samples. By comparing co-crystal structures, we find that BarlowDTI effectively exploits catalytically active and stabilising residues, highlighting the model's ability to generalise from one-dimensional input data. In addition, we further benchmark new baselines against existing methods. Together, these innovations improve the efficiency and effectiveness of drug-target interaction predictions, providing robust tools for accelerating drug development and deepening the understanding of molecular interactions. Therefore, we provide an easy-to-use web interface that can be freely accessed at https://www.bio.nat.tum.de/oc2/barlowdti ., Comment: Refined model architecture, additional results added

Mobile robots, becoming increasingly autonomous, are capable of operating in diverse and unknown environments. This flexibility allows them to fulfill goals independently and adapting their actions dynamically without rigidly predefined control codes. However, their autonomous behavior complicates guaranteeing safety and reliability due to the limited influence of a human operator to accurately supervise and verify each robot's actions. To ensure autonomous mobile robot's safety and reliability, which are aspects of dependability, methods are needed both in the planning and execution of missions for autonomous mobile robots. In this article, a twofold approach is presented that ensures fault removal in the context of mission planning and fault prevention during mission execution for autonomous mobile robots. First, the approach consists of a concept based on formal verification applied during the planning phase of missions. Second, the approach consists of a rule-based concept applied during mission execution. A use case applying the approach is presented, discussing how the two concepts complement each other and what contribution they make to certain aspects of dependability., Comment: Language: German. Dieser Beitrag wird eingereicht in: "dtec.bw-Beitr\"age der Helmut-Schmidt-Universit\"at/Universit\"at der Bundeswehr Hamburg: Forschungsaktivit\"aten im Zentrum f\"ur Digitalisierungs- und Technologieforschung der Bundeswehr dtec.bw"

Tube-based model predictive control (MPC) is the principal robust control technique for constrained linear systems affected by additive disturbances. While tube-based methods that compute the tubes online have been successfully applied to systems with additive disturbances, their application to systems affected by additional model uncertainties is challenging. This paper introduces a new tube-based MPC method - named filter-based system level tube-MPC (SLTMPC) - which overapproximates both uncertainties with an online optimized disturbance set, while simultaneously computing the tube controller online. Extending prior work, we generalize the method to polytopic disturbance sets and for the first time provide rigorous closed-loop guarantees for the receding horizon controller. These guarantees are obtained by virtue of a new terminal controller design and an online optimized terminal set. To reduce the computational complexity of the proposed method, we additionally introduce an asynchronous computation scheme that separates the optimization of the tube controller and the nominal trajectory. Finally, we provide a comprehensive numerical evaluation of the proposed methods to demonstrate their effectiveness.

We present the first results from a proof-of-concept search for dark sectors via invisible decays of pseudoscalar $\eta$ and $\eta'$ mesons in the NA64h experiment at the CERN SPS. Our novel technique uses the charge-exchange reaction of 50 GeV $\pi^-$ on nuclei of an active target as the source of neutral mesons. The $\eta, \eta' \to invisible$ events would exhibit themselves via a striking signature - the complete disappearance of the incoming beam energy in the detector. No evidence for such events has been found with $2.9\times10^{9}$ pions on target accumulated during one day of data taking. This allows us to set a stringent limit on the branching ratio ${\rm Br}(\eta' \to invisible) < 2.1 \times 10^{-4}$ improving the current bound by a factor of $\simeq3$. We also set a limit on ${\rm Br}(\eta \to invisible) < 1.1 \times 10^{-4}$ comparable with the existing one. These results demonstrate the great potential of our approach and provide clear guidance on how to enhance and extend the sensitivity for dark sector physics from future searches for invisible neutral meson decays., Comment: 5 pages, 3 figures

Softmax attention is the principle backbone of foundation models for various artificial intelligence applications, yet its quadratic complexity in sequence length can limit its inference throughput in long-context settings. To address this challenge, alternative architectures such as linear attention, State Space Models (SSMs), and Recurrent Neural Networks (RNNs) have been considered as more efficient alternatives. While connections between these approaches exist, such models are commonly developed in isolation and there is a lack of theoretical understanding of the shared principles underpinning these architectures and their subtle differences, greatly influencing performance and scalability. In this paper, we introduce the Dynamical Systems Framework (DSF), which allows a principled investigation of all these architectures in a common representation. Our framework facilitates rigorous comparisons, providing new insights on the distinctive characteristics of each model class. For instance, we compare linear attention and selective SSMs, detailing their differences and conditions under which both are equivalent. We also provide principled comparisons between softmax attention and other model classes, discussing the theoretical conditions under which softmax attention can be approximated. Additionally, we substantiate these new insights with empirical validations and mathematical arguments. This shows the DSF's potential to guide the systematic development of future more efficient and scalable foundation models., Comment: NeurIPS 2024

In this study, we demonstrate the potential of the NA64 experiment at CERN SPS to search for New Physics processes involving $e\rightarrow\mu$ transitions after the collision of 100 GeV electrons with target nuclei. A new Dark Sector leptonic portal in which a scalar boson $\varphi$ could be produced in the lepton-flavor-changing bremsstrahlung-like reaction, $eN\rightarrow \mu N\varphi$, is used as benchmark process. In this work, we develop a realistic Monte Carlo simulation of the NA64 experimental setup implementing the differential and total production cross-section computed at exact tree-level and applying the Weisz\"{a}cker-Williams phase space approximation. Using this framework, we investigate the main background sources and calculate the expected sensitivity of the experiment. The results indicate that with minor setup optimization, NA64 can probe a large fraction of the available parameter space compatible with the muon $g-2$ anomaly and the Dark Matter relic predictions in the context of a new Dark Sector leptonic portal with $10^{11}$ EOT. This result paves the way to the exploration of lepton-flavour-changing transitions in NA64., Comment: 11 pages, 8 figures

The inclusion of an additional $U(1)$ gauge $L_\mu-L_\tau$ symmetry would release the tension between the measured and the predicted value of the anomalous muon magnetic moment: this paradigm assumes the existence of a new, light $Z^\prime$ vector boson, with dominant coupling to $\mu$ and $\tau$ leptons and interacting with electrons via a loop mechanism. The $L_\mu-L_\tau$ model can also explain the Dark Matter relic abundance, by assuming that the $Z'$ boson acts as a "portal" to a new Dark Sector of particles in Nature, not charged under known interactions. In this work we present the results of the $Z'$ search performed by the NA64-$e$ experiment at CERN SPS, that collected $\sim 9\times10^{11}$ 100 GeV electrons impinging on an active thick target. Despite the suppressed $Z'$ production yield with an electron beam, NA64-$e$ provides the first accelerator-based results excluding the $g-2$ preferred band of the $Z'$ parameter space in the 1 keV $ < m_{Z'} \lesssim 2$ MeV range, in complementarity with the limits recently obtained by the NA64-$\mu$ experiment with a muon beam., Comment: 15 pages, 5 figures. New version with extended exclusion limits down to 1 keV in Z' mass

We consider inexact policy iteration methods for large-scale infinite-horizon discounted MDPs with finite spaces, a variant of policy iteration where the policy evaluation step is implemented inexactly using an iterative solver for linear systems. In the classical dynamic programming literature, a similar principle is deployed in optimistic policy iteration, where an a-priori fixed-number of iterations of value iteration is used to inexactly solve the policy evaluation step. Inspired by the connection between policy iteration and semismooth Newton's method, we investigate a class of iPI methods that mimic the inexact variants of semismooth Newton's method by adopting a parametric stopping condition to regulate the level of inexactness of the policy evaluation step. For this class of methods we discuss local and global convergence properties and derive a practical range of values for the stopping-condition parameter that provide contraction guarantees. Our analysis is general and therefore encompasses a variety of iterative solvers for policy evaluation, including the standard value iteration as well as more sophisticated ones such as GMRES. As underlined by our analysis, the selection of the inner solver is of fundamental importance for the performance of the overall method. We therefore consider different iterative methods to solve the policy evaluation step and analyze their applicability and contraction properties when used for policy evaluation. We show that the contraction properties of these methods tend to be enhanced by the specific structure of policy evaluation and that there is margin for substantial improvement in terms of convergence rate. Finally, we study the numerical performance of different instances of inexact policy iteration on large-scale MDPs for the design of health policies to control the spread of infectious diseases in epidemiology.

In recent years, there has been a growing interest in integrating linear state-space models (SSM) in deep neural network architectures of foundation models. This is exemplified by the recent success of Mamba, showing better performance than the state-of-the-art Transformer architectures in language tasks. Foundation models, like e.g. GPT-4, aim to encode sequential data into a latent space in order to learn a compressed representation of the data. The same goal has been pursued by control theorists using SSMs to efficiently model dynamical systems. Therefore, SSMs can be naturally connected to deep sequence modeling, offering the opportunity to create synergies between the corresponding research areas. This paper is intended as a gentle introduction to SSM-based architectures for control theorists and summarizes the latest research developments. It provides a systematic review of the most successful SSM proposals and highlights their main features from a control theoretic perspective. Additionally, we present a comparative analysis of these models, evaluating their performance on a standardized benchmark designed for assessing a model's efficiency at learning long sequences.

Machine Learning (ML) techniques are encountered nowadays across disciplines, from social sciences, through natural sciences to engineering. The broad application of ML and the accelerated pace of its evolution lead to an increasing need for dedicated teaching concepts aimed at making the application of this technology more reliable and responsible. However, teaching ML is a daunting task. Aside from the methodological complexity of ML algorithms, both with respect to theory and implementation, the interdisciplinary and empirical nature of the field need to be taken into consideration. This paper introduces the MachineLearnAthon format, an innovative didactic concept designed to be inclusive for students of different disciplines with heterogeneous levels of mathematics, programming and domain expertise. At the heart of the concept lie ML challenges, which make use of industrial data sets to solve real-world problems. These cover the entire ML pipeline, promoting data literacy and practical skills, from data preparation, through deployment, to evaluation.

This study presents experimental observations of single laser-induced cavitation bubbles collapsing in water with different levels of air saturation. The average trends of the bubble size reveal a clear yet little dependence of the energy dissipation at collapse on the water gas content. Similarly, the observed trend of luminescence energy at collapse varies within the investigated range of air saturation levels, in agreement with findings in similar works. We argue that perturbations in bubble shape may amplify with decreasing air saturation, consequently affecting light emission.

The composition of the gaseous phase of cavitation bubbles and its role on the collapse remains to date poorly understood. In this work, experiments of single cavitation bubbles in aqueous ammonia serve as a novel approach to investigate the effect of the vapor contained in a bubble on its collapse. We find that the higher vapor pressure of more concentrated aqueous ammonia acts as a resistance to the collapse, reducing the total energy dissipation. In line with visual observation, acoustic measurements, and luminescence recordings, it is also observed that higher vapor pressures contribute to a more spherical collapse, likely hindering the growth of interface instabilities by decreasing the collapse velocities and accelerations. Remarkably, we evidence a strong difference between the effective damping and the energy of the shock emission, suggesting that the latter is not the dominant dissipation mechanism at collapse as predicted from classical correction models accounting for slightly compressible liquids. Furthermore, our results suggest that the vapor inside collapsing bubbles gets compressed, consistently with previous studies performed in the context of single bubble sonoluminescence, addressing the question about the ability of vapors to readily condense during a bubble collapse in similar regimes. These findings provide insights into the identification of the influence of the bubble content and the energy exchanges of the bubble with its surrounding media, eventually paving the way to a more efficient use of cavitation in engineering and biomedical applications.

The search for new comparably light (well below the electroweak scale) feebly interacting particles is an exciting possibility to explain some mysterious phenomena in physics, among them the origin of Dark Matter. The sensitivity study through detailed simulation of projected experiments is a key point in estimating their potential for discovery. Several years ago we created the DMG4 package for the simulation of DM (Dark Matter) particles in fixed target experiments. The natural approach is to integrate this simulation into the same program that performs the full simulation of particles in the experiment setup. The Geant4 toolkit framework was chosen as the most popular and versatile solution nowadays. The simulation of DM particles production by this package accommodates several possible scenarios, employing electron, muon or photon beams and involving various mediators, such as vector, axial vector, scalar, pseudoscalar, or spin 2 particles. The bremsstrahlung, annihilation or Primakoff processes can be simulated. The package DMG4 contains a subpackage DarkMatter with cross section methods weakly connected to Geant4. It can be used in different frameworks. In this paper, we present the latest developments of the package, such as extending the list of possible mediator particle types, refining formulas for the simulation and extending the mediator mass range. The user interface is also made more flexible and convenient. In this work, we also demonstrate the usage of the package, the improvements in the simulation accuracy and some cross check validations., Comment: 17 pages, 11 figures, 1 table

The success of drug discovery and development relies on the precise prediction of molecular activities and properties. While in silico molecular property prediction has shown remarkable potential, its use has been limited so far to assays for which large amounts of data are available. In this study, we use a fine-tuned large language model to integrate biological assays based on their textual information, coupled with Barlow Twins, a Siamese neural network using a novel self-supervised learning approach. This architecture uses both assay information and molecular fingerprints to extract the true molecular information. TwinBooster enables the prediction of properties of unseen bioassays and molecules by providing state-of-the-art zero-shot learning tasks. Remarkably, our artificial intelligence pipeline shows excellent performance on the FS-Mol benchmark. This breakthrough demonstrates the application of deep learning to critical property prediction tasks where data is typically scarce. By accelerating the early identification of active molecules in drug discovery and development, this method has the potential to help streamline the identification of novel therapeutics., Comment: 13(+9) pages(+appendix), 5 figures, 11 tables

We report on a search for a new $Z'$ ($L_\mu-L_\tau$) vector boson performed at the NA64 experiment employing a high energy muon beam and a missing energy-momentum technique. Muons from the M2 beamline at the CERN Super Proton Synchrotron with a momentum of 160 GeV/c are directed to an active target. A signal event is a single scattered muon with momentum $<$ 80 GeV/c in the final state, accompanied by missing energy, i.e. no detectable activity in the downstream calorimeters. For a total statistic of $(1.98\pm0.02)\times10^{10}$ muons on target, no event is observed in the expected signal region. This allows us to set new limits on part of the remaining $(m_{Z'},\ g_{Z'})$ parameter space which could provide an explanation for the muon $(g-2)_\mu$ anomaly. Additionally, our study excludes part of the parameter space suggested by the thermal Dark Matter relic abundance. Our results pave the way to explore Dark Sectors and light Dark Matter with muon beams in a unique and complementary way to other experiments., Comment: 5 pages, 4 figures

This paper proposes a novel method for demand forecasting in a pricing context. Here, modeling the causal relationship between price as an input variable to demand is crucial because retailers aim to set prices in a (profit) optimal manner in a downstream decision making problem. Our methods bring together the Double Machine Learning methodology for causal inference and state-of-the-art transformer-based forecasting models. In extensive empirical experiments, we show on the one hand that our method estimates the causal effect better in a fully controlled setting via synthetic, yet realistic data. On the other hand, we demonstrate on real-world data that our method outperforms forecasting methods in off-policy settings (i.e., when there's a change in the pricing policy) while only slightly trailing in the on-policy setting.

As language models are adopted by a more sophisticated and diverse set of users, the importance of guaranteeing that they provide factually correct information supported by verifiable sources is critical across fields of study. This is especially the case for high-stakes fields, such as medicine and law, where the risk of propagating false information is high and can lead to undesirable societal consequences. Previous work studying attribution and factuality has not focused on analyzing these characteristics of language model outputs in domain-specific scenarios. In this work, we conduct human evaluation of responses from a few representative systems along various axes of attribution and factuality, by bringing domain experts in the loop. Specifically, we collect expert-curated questions from 484 participants across 32 fields of study, and then ask the same experts to evaluate generated responses to their own questions. In addition, we ask experts to improve upon responses from language models. The output of our analysis is ExpertQA, a high-quality long-form QA dataset with 2177 questions spanning 32 fields, along with verified answers and attributions for claims in the answers., Comment: Accepted to NAACL 2024. Dataset & code is available at https://github.com/chaitanyamalaviya/expertqa

We present the results of a missing-energy search for Light Dark Matter which has a new interaction with ordinary matter transmitted by a vector boson, called dark photon $A^\prime$. For the first time, this search is performed with a positron beam by using the significantly enhanced production of $A^\prime$ in the resonant annihilation of positrons with atomic electrons of the target nuclei, followed by the invisible decay of $A^\prime$ into dark matter. No events were found in the signal region with $(10.1 \pm 0.1)~\times~10^{9}$ positrons on target with 100 GeV energy. This allowed us to set new exclusion limits that, relative to the collected statistics, prove the power of this experimental technique. This measurement is a crucial first step toward a future exploration program with positron beams, whose estimated sensitivity is here presented., Comment: Version submitted to PRL

We construct systematically experimental steady-state bifurcation diagrams for entorhinal cortex neurons. A slowly ramped voltage-clamp electrophysiology protocol serves as closed-loop feedback controlled experiment for the subsequent current-clamp open-loop protocol on the same cell. In this way, the voltage-clamped experiment determines dynamically stable and unstable (hidden) steady states of the current-clamp experiment. The transitions between observable steady states and observable spiking states in the current-clamp experiment reveal stability and bifurcations of the steady states, completing the steady-state bifurcation diagram.