Search

Machine learning (ML) has the potential to revolutionize various domains, but its adoption is often hindered by the disconnect between the needs of domain experts and translating these needs into robust and valid ML tools. Despite recent advances in LLM-based co-pilots to democratize ML for non-technical domain experts, these systems remain predominantly focused on model-centric aspects while overlooking critical data-centric challenges. This limitation is problematic in complex real-world settings where raw data often contains complex issues, such as missing values, label noise, and domain-specific nuances requiring tailored handling. To address this we introduce CliMB-DC, a human-guided, data-centric framework for LLM co-pilots that combines advanced data-centric tools with LLM-driven reasoning to enable robust, context-aware data processing. At its core, CliMB-DC introduces a novel, multi-agent reasoning system that combines a strategic coordinator for dynamic planning and adaptation with a specialized worker agent for precise execution. Domain expertise is then systematically incorporated to guide the reasoning process using a human-in-the-loop approach. To guide development, we formalize a taxonomy of key data-centric challenges that co-pilots must address. Thereafter, to address the dimensions of the taxonomy, we integrate state-of-the-art data-centric tools into an extensible, open-source architecture, facilitating the addition of new tools from the research community. Empirically, using real-world healthcare datasets we demonstrate CliMB-DC's ability to transform uncurated datasets into ML-ready formats, significantly outperforming existing co-pilot baselines for handling data-centric challenges. CliMB-DC promises to empower domain experts from diverse domains -- healthcare, finance, social sciences and more -- to actively participate in driving real-world impact using ML., Comment: Saveliev, Liu & Seedat contributed equally

Lorentz invariance violation is a feature of several quantum gravity models in which Lorentz symmetry is broken at high energies, leading to potential changes in particle behavior and interactions. In this study, we investigate vacuum Cherenkov radiation, a reaction in which an electron spontaneously emits a photon. This process, forbidden when considering unbroken Lorentz symmetry, is a phenomenological consequence of some quantum gravity models. We derive, for the first time, the spectra for the vacuum Cherenkov reaction, and confirm our results numerically. These results can be used to derive limits on Lorentz invariance violation., Comment: 12 pages, 6 figures

In the early days of the Floer theory, Atiyah asked if there is a Milnor fiber description of the Floer homology of the links of singularities. We answer this question for the Brieskorn-Hamm complete intersection singularities. The resulting combinatorial formulas lead to an independent proof of the equality of the Casson invariants in the Donaldson and Seiberg-Witten theories. We use similar techniques to express the Heegaard Floer d-invariant of torus knots in terms of their Milnor fibers., Comment: 22 pages

Despite its significant promise and continuous technical advances, real-world applications of artificial intelligence (AI) remain limited. We attribute this to the "domain expert-AI-conundrum": while domain experts, such as clinician scientists, should be able to build predictive models such as risk scores, they face substantial barriers in accessing state-of-the-art (SOTA) tools. While automated machine learning (AutoML) has been proposed as a partner in clinical predictive modeling, many additional requirements need to be fulfilled to make machine learning accessible for clinician scientists. To address this gap, we introduce CliMB, a no-code AI-enabled partner designed to empower clinician scientists to create predictive models using natural language. CliMB guides clinician scientists through the entire medical data science pipeline, thus empowering them to create predictive models from real-world data in just one conversation. CliMB also creates structured reports and interpretable visuals. In evaluations involving clinician scientists and systematic comparisons against a baseline GPT-4, CliMB consistently demonstrated superior performance in key areas such as planning, error prevention, code execution, and model performance. Moreover, in blinded assessments involving 45 clinicians from diverse specialties and career stages, more than 80% preferred CliMB over GPT-4. Overall, by providing a no-code interface with clear guidance and access to SOTA methods in the fields of data-centric AI, AutoML, and interpretable ML, CliMB empowers clinician scientists to build robust predictive models. The proof-of-concept version of CliMB is available as open-source software on GitHub: https://github.com/vanderschaarlab/climb., Comment: * Evgeny Saveliev and Tim Schubert contributed equally to this work

The Benard-Conway invariant of links in the 3-sphere is a Casson-Lin type invariant defined by counting irreducible SU(2) representations of the link group with fixed meridional traces. For two-component links with linking number one, the invariant has been shown to equal a symmetrized multivariable link signature. We extend this result to all two-component links with non-zero linking number. A key ingredient in the proof is an explicit calculation of the Benard-Conway invariant for (2, 2n)-torus links with the help of the Chebyshev polynomials., Comment: Added a paragraph relating our work to that of Daemi and Scaduto on irreducible instanton homology of links. 32 pages, 10 figures

We present the comprehensive Russian primary and secondary legislation corpus covering 1991 to 2023. The corpus collects all 281,413 texts (176,523,268 tokens) of non-secret federal regulations and acts, along with their metadata. The corpus has two versions the original text with minimal preprocessing and a version prepared for linguistic analysis with morphosyntactic markup., Comment: 6 pages, 2 figures, 2 tables

We theoretically study an artificial neuron circuit containing a quantum memristor in the presence of relaxation and dephasing. The charge transport in the quantum element is realized via tunneling of a charge through a quantum particle which shuttles between two terminals -- a functionality reminiscent of classical diffusive memristors. We demonstrate that this physical principle enables hysteretic behavior of the current-voltage characteristics of the quantum device. In addition, being used in artificial neural circuit, the quantum switcher is able to generate self-sustained current oscillations. Our analysis reveals that these self-oscillations are triggered only in quantum regime with a moderate rate of relaxation, and cannot exist either in a purely coherent regime or at a very high decoherence. We investigate the hysteresis and instability leading to the onset of current self-oscillations and analyze their properties depending on the circuit parameters. Our results provide a generic approach to the use of quantum regimes for controlling hysteresis and generating self-oscillations., Comment: 14 pages, 7 figures, 6 appendices

Lorentz invariance violation (LIV) is a phenomenon featuring in various quantum gravity models whereby Lorentz symmetry is broken at high energies, potentially impacting the behaviour of particles and their interactions. Here we investigate the phenomenology of LIV within the context of gamma-ray-induced electromagnetic cascades. We conduct detailed numerical simulations to explore the expected manifestations of LIV on gamma-ray fluxes, taking into account relevant effects such as pair production and inverse Compton scattering. Additionally, we consider processes forbidden in the Standard Model, namely vacuum Cherenkov emission and photon decay. Our analysis reveals that these modifications result in distinct characteristics within the measured particle fluxes at Earth, which have the potential to be observed in high-energy gamma-ray observations., Comment: 18 pages, 9 figures; various changes to match published version

We consider a certain class of infinitary rules of inference, called here restriction rules, using of which allows us to deduce complete theories of given models. The first instance of such rules was the $\omega$-rule introduced by Hilbert, and generalizations of the $\omega$-rule were first considered by Henkin. Later on Barwise showed that within countable languages, for any countable model $\mathfrak M$, first-order logic expanded by the corresponding $\mathfrak M$-rule deduces from the diagram of $\mathfrak M$ all formulas that are true in all models that include $\mathfrak M$, in particular, it deduces the (relativized) complete theory of $\mathfrak M$. We show that, if the aim is only deducing the complete theory of a given model, these countability assumptions can be omitted. Moreover, similar facts hold for infinitary and higher-order logics, even if these logics are highly incomplete. Finally, we show that Barwise's theorem in its stronger form, for vocabularies of arbitrary cardinality, holds for infinitary logics $\mathscr L_{\kappa,\lambda}$ whenever $\kappa$ is supercompact. The note also contains brief historical remarks.

The Compact Linear Accelerator for Research and Applications (CLARA) is a 250 MeV ultrabright electron beam test facility at STFC Daresbury Laboratory. A user beam line has been designed to maximise exploitation of CLARA in a variety of fields, including novel acceleration and new modalities of radiotherapy. In this paper we present the specification and design of this beam line for Full Energy Beam Exploitation (FEBE). We outline the key elements which provide users to access ultrashort, low emittance electron bunches in two large experiment chambers. The results of start-to-end simulations are reported which verify the expected beam parameters delivered to these chambers. Key technical systems are detailed, including those which facilitate combination of electron bunches with high power laser pulses., Comment: 13 pages, 12 figures

CRPropa is a Monte Carlo framework for simulating the propagation of (ultra-) high-energy particles in the Universe, including cosmic rays, gamma rays, electrons, and neutrinos. It covers energies from ZeV down to GeV for gamma rays and electrons, and TeV for cosmic rays and neutrinos, supporting various astrophysical environments such as the surroundings of astrophysical sources, galactic, and extragalactic environments. The newest version, CRPropa 3.2, represents a significant leap forward towards a universal multi-messenger framework, opening up the possibility for many more astrophysical applications. This includes extensions to simulate cosmic-ray acceleration and particle interactions within astrophysical source environments, a full Monte Carlo treatment of electromagnetic cascades, improved ensemble-averaged Galactic propagation, significant performance improvements for cosmic-ray tracking through magnetic fields, and a user-friendly implementation of custom photon fields, among many more enhancements. This contribution will give an overview of the new features and present several applications to cosmic-ray and gamma-ray propagation., Comment: PoS 444 (38th ICRC) (2023) 1471

Lorentz invariance violation (LIV) is a proposed phenomenon where Lorentz symmetry is violated at high energies, potentially affecting particle dynamics and interactions. We use numerical simulations with the CRPropa framework to investigate LIV in gamma-ray-induced electromagnetic cascades, specifically studying how it impacts cascading electrons and photons undergoing pair production and inverse Compton scattering. Our detailed analysis of the simulation results, compared with existing theoretical models, reveals that LIV can significantly alter the behavior of both components of the cascade, photons and electrons, resulting in specific signatures in measured fluxes that could be observed in high-energy gamma-ray observations. These insights are crucial for ongoing searches for LIV and for the development of theoretical models incorporating LIV effects., Comment: 8 pages, 2 figures; preliminary version of the proceedings for ICRC2023

Diffusive memristors owing to their ability to produce current spiking when a constant or slowly changing voltage is applied are competitive candidates for the development of artificial electronic neurons. These artificial neurons can be integrated into various prospective autonomous and robotic systems as sensors, e.g. ones implementing object grasping and classification. We report here Ag nanoparticle-based diffusive memristor prepared on a flexible polyethylene terephthalate (PET) substrate in which the electric spiking behaviour was induced by the electric voltage under an additional stimulus of external mechanical impact. By changing the magnitude and frequency of the mechanical impact, we are able to manipulate the spiking response of our artificial neuron. This functionality to control the spiking characterstics paves a pathway for the development of touch-perception sensors that can convert local pressure into electrical spikes for further processing in neural networks. We have proposed a mathematical model which captures the operation principle of the fabricated memristive sensors and qualitatively describes the measured spiking behaviour., Comment: 18 pages, 7 figures

Diffusive memristors continue to receive tremendous interest due to their ability to emulate biological neurons and thus aid the development of bio-inspired computation technology. A major issue with the diffusive memristor is the inability to reliably control the formation of the conduction filaments which affects both the device functionality and reproducibility of regimes after each application of voltage. Here we investigate the effect of gamma radiation on the electrical properties of the diffusive memristors based on metallic nanoparticles in dielectric matrix. Our experiments show that after exposing to radiation, the memristors demonstrate much sharper (and less noisy) hysteresis in the current-voltage characteristics while preserving the same low- and high-resistive states as in the pristine samples. Additionally, the radiation lowers both threshold and hold voltages that correspond to onset of low- and high- resistive states, respectively. The proposed mechanism involves radiation-induced defects in the silica matrix which help to establish dominant pathways for nanoparticles to form conduction filaments. Our findings suggest an efficient way to enhance working characteristics of diffusive memristors and to improve their reproducibility., Comment: 23 pages, 6 figures

After an explosive eruption of the Hunga Tonga volcano on January 15, 2022, disturbances were observed at a distance of about 12000km in Northern Tien Shan among the variations of the atmosphere pressure, of telluric current, and of the Doppler frequency shift of ionospheric signal. At 16:00:55UTC a pulse of atmospheric pressure was detected there with a peak amplitude of 1.3hPa and propagation speed of 0.3056km/s, equal to the velocity of Lamb wave. In the variations of the Doppler frequency shift, the disturbances of two types were registered on the 3212km and 2969km long inclined radio-paths, one of which arose as a response to the passage of a Lamb wave (0.3059km/s) through the reflection point of radio wave, and another as reaction to the acoustic-gravity wave (0.2602km/s). Two successive perturbations were also detected in the records of telluric current at the arrival times of the Lamb and acoustic-gravity waves into the registration point. According to the parameters of the Lamb wave, an energy transfer into the atmosphere at the explosion of the Hunga Tonga volcano was roughly estimated as 2000Mt of TNT equivalent.

TemporAI is an open source Python software library for machine learning (ML) tasks involving data with a time component, focused on medicine and healthcare use cases. It supports data in time series, static, and eventmodalities and provides an interface for prediction, causal inference, and time-to-event analysis, as well as common preprocessing utilities and model interpretability methods. The library aims to facilitate innovation in the medical ML space by offering a standardized temporal setting toolkit for model development, prototyping and benchmarking, bridging the gaps in the ML research, healthcare professional, medical/pharmacological industry, and data science communities. TemporAI is available on GitHub (https://github.com/vanderschaarlab/temporai) and we welcome community engagement through use, feedback, and code contributions.

The landscape of high- and ultra-high-energy astrophysics has changed in the last decade, largely due to the inflow of data collected by large-scale cosmic-ray, gamma-ray, and neutrino observatories. At the dawn of the multimessenger era, the interpretation of these observations within a consistent framework is important to elucidate the open questions in this field. CRPropa 3.2 is a Monte Carlo code for simulating the propagation of high-energy particles in the Universe. This version represents a major leap forward, significantly expanding the simulation framework and opening up the possibility for many more astrophysical applications. This includes, among others: efficient simulation of high-energy particles in diffusion-dominated domains, self-consistent and fast modelling of electromagnetic cascades with an extended set of channels for photon production, and studies of cosmic-ray diffusion tensors based on updated coherent and turbulent magnetic-field models. Furthermore, several technical updates and improvements are introduced with the new version, such as: enhanced interpolation, targeted emission of sources, and a new propagation algorithm (Boris push). The detailed description of all novel features is accompanied by a discussion and a selected number of example applications., Comment: 32 pages, 9 figures

Given a matrix pseudodifferential operator on a smooth manifold, one may be interested in diagonalising it by choosing eigenvectors of its principal symbol in a smooth manner. We show that diagonalisation is not always possible, on the whole cotangent bundle or even in a single fibre. We identify global and local topological obstructions to diagonalisation and examine physically meaningful examples demonstrating that all possible scenarios can occur., Comment: v3: final version

Structure wakefield acceleration (SWFA) is one of the most promising AAC schemes in several recent strategic reports, including DOE's 2016 AAC Roadmap, report on the Advanced and Novel Accelerators for High Energy Physics Roadmap (ANAR), and report on Accelerator and Beam Physics Research Goals and Opportunities. SWFA aims to raise the gradient beyond the limits of conventional radiofrequency (RF) accelerator technology, and thus the RF to beam energy efficiency, by reducing RF breakdowns from confining the microwave energy in a short (on the order of about 10 ns) and intense pulse excited by a drive beam. We envision that the following research topics, within the scope of AF7, are of great interest in the next decade: advanced wakefield structures, terahertz and sub-terahertz (THz) structures, and RF breakdown physics. Research on SWFA in the above directions would directly contribute to long-term large-scale applications, including AAC-based linear colliders and compact light sources. There is also potentially a strong synergy between SWFA and other AAC concepts, when structures are combined with plasmas into hybrid AAC schemes. Research on novel structures is at the core of advancing SWFA, and is critical to future AAC-based linear colliders; at the same, it has a strong synergy with other directions, such as cavity designs, high-power microwave systems and sources, and compact light sources., Comment: contribution to Snowmass 2021