Search

Hallucinations are one of the major issues affecting LLMs, hindering their wide adoption in production systems. While current research solutions for detecting hallucinations are mainly based on heuristics, in this paper we introduce a mathematically sound methodology to reason about hallucination, and leverage it to build a tool to detect hallucinations. To the best of our knowledge, we are the first to show that hallucinated content has structural differences with respect to correct content. To prove this result, we resort to the Minkowski distances in the embedding space. Our findings demonstrate statistically significant differences in the embedding distance distributions, that are also scale free -- they qualitatively hold regardless of the distance norm used and the number of keywords, questions, or responses. We leverage these structural differences to develop a tool to detect hallucinated responses, achieving an accuracy of 66\% for a specific configuration of system parameters -- comparable with the best results in the field. In conclusion, the suggested methodology is promising and novel, possibly paving the way for further research in the domain, also along the directions highlighted in our future work.

The fate of the Fermi surface in bulk electron-doped Sr$_{2}$IrO$_{4}$ remains elusive, as does the origin and extension of its pseudogap phase. Here, we use high-resolution angle-resolved photoelectron spectroscopy (ARPES) to investigate the electronic structure of Sr$_{2-x}$La$_{x}$IrO$_{4}$ up to $x=0.2$, a factor of two higher than in previous work. Our findings reveal that the Fermi surface evolves smoothly with doping. Notably, the antinodal pseudogap persists up to the highest doping level, while nodal quasiparticle coherence increases monotonously. This demonstrates that the sharp increase in Hall carrier density recently observed above $x^{*}=0.16$ [Y.-T. Hsu et al., Nature Physics 20, 1596 (2024)] cannot be attributed to the closure of the pseudogap. Further, we determine a temperature boundary of the pseudogap of $T^{*}\simeq~200~\textrm{K}$ for $x=0.2$, comparable to cuprates. Our results suggest that pseudogaps are a generic feature of doped quasi-2D antiferromagnetic Mott insulators, likely related to short range magnetic correlations., Comment: 8 pages, 3 figures. Supplementary Information: 3 pages, 3 figures

We study the generation, nonlinear development and secondary instability of unsteady G\"ortler vortices and streaks in compressible boundary layers exposed to free-stream vortical disturbances and evolving over concave, flat and convex walls. The formation and evolution of the disturbances are governed by the compressible nonlinear boundary-region equations, supplemented by initial and boundary conditions that characterise the impact of the free-stream disturbances on the boundary layer. Computations are performed for parameters typical of flows over high-pressure turbine blades, where the G\"ortler number, a measure of the curvature effects, and the disturbance Reynolds number, a measure of the nonlinear effects, are order-one quantities. At moderate intensities of the free-stream disturbances, increasing the G\"ortler number renders the boundary layer more unstable, while increasing the Mach number or the frequency stabilises the flow. As the free-stream disturbances become more intense, vortices over concave surfaces no longer develop into the characteristic mushroom-shaped structures, while the flow over convex surfaces is destabilised. An occurrence map identifies G\"ortler vortices or streaks for different levels of free-stream disturbances and G\"ortler numbers. Our calculations capture well the experimental measurements of the enhanced skin friction and wall-heat transfer over turbine-blade pressure surfaces. The time-averaged wall-heat transfer modulations, termed hot fingers, are elongated in the streamwise direction and their spanwise wavelength is half of the characteristic wavelength of the free-stream disturbances. Nonlinearly saturated disturbances are unstable to secondary high-frequencymodes, whose growth rate increases with the G\"ortler number. A new varicose even mode is reported, which may promote transition to turbulence at the stem of nonlinear streaks.

The magnetised near detector (ND280) of the T2K long-baseline neutrino oscillation experiment has been recently upgraded aiming to satisfy the requirement of reducing the systematic uncertainty from measuring the neutrinonucleus interaction cross section, which is the largest systematic uncertainty in the search for leptonic charge-parity symmetry violation. A key component of the upgrade is SuperFGD, a 3D segmented plastic scintillator detector made of approximately 2,000,000 optically-isolated 1 cm3 cubes. It will provide a 3D image of GeV neutrino interactions by combining tracking and stopping power measurements of final state particles with sub-nanosecond time resolution. The performance of SuperFGD is characterized by the precision of its response to charged particles as well as the systematic effects that might affect the physics measurements. Hence, a detailed Geant4 based optical simulation of the SuperFGD building block, i.e. a plastic scintillating cube read out by three wavelength shifting fibers, has been developed and validated with the different datasets collected in various beam tests. In this manuscript the description of the optical model as well as the comparison with data are reported., Comment: 31 pages, 15 figures

We explore the concept of negative pressure and its relevance in a variety of physical contexts: the expansion of the universe, mixture theory, cavitation, and the capillary effect in plants. Using thermodynamic arguments, we discuss the intricate connection between negative pressure and negative thermal expansion. We highlight the fact that metastable states and competing phases are often associated with the emergence of negative pressure. We also propose a new link between the effective Gr\"uneisen parameter and nucleation theory., Comment: 15 pages, 5 figs, comments are welcome

Protein compartmentalization in the frame of a liquid-liquid phase separation is a key mechanism to optimize spatiotemporal control of biological systems. Such a compartmentalization process reduces the intrinsic noise in protein concentration due to stochasticity in gene expression. Employing Flory-Huggins solution theory, Avramov/Casalini's model, and the Gr\"uneisen parameter, we unprecedentedly propose a cellular Griffiths-like phase (CGLP), which can impact its functionality and self-organization. The here-proposed CGLP is key ranging from the understanding of primary organisms' evolution to the treatment of diseases. Our findings pave the way for an alternative Biophysics approach to investigate coacervation processes., Comment: 13 pages, 2 figures, suppl. material upon request

The "Poor Man's Majorana" [Phys. Rev. B 86, 134528 (2012)] devoid of topological protection has been theoretically predicted to rely on the minimal Kitaev chain. Afterward, a pair of superconducting and spinless quantum dots turned the proposal practicable and differential conductance pinpointed consistent fingerprints with such a scenario [Nature 614, 445 (2023) and Nature 630, 329 (2024)]. In this work, we propose a model wherein the "Poor Man's Majorana" presents protection when one of the dots is exchange coupled to a quantum spin. If this quantum dot is perturbed by tuning the exchange coupling, the well-known spill over-like behavior of this Majorana surprisingly remains unchanged, and solely half of the fine structure is unexpectedly viewed. As a matter of fact, the "Poor Man's Majorana" zero mode consists in squeezing of the other half at zero frequency, which imposes its pinning there and prevents the mixing of the mode with the explicit fine structure. We claim that if the supposed unavoidable split of the zero mode by the fine structure is unexpectedly absent, then the "Poor Man's Majorana" can be considered robust against the quantum spin. In this way, it becomes protected and the lack of topological protection paradigm of the "Poor Man's Majorana" has been revisited, pushing this seemingly well-established issue into a new direction.

We introduce Imagen 3, a latent diffusion model that generates high quality images from text prompts. We describe our quality and responsibility evaluations. Imagen 3 is preferred over other state-of-the-art (SOTA) models at the time of evaluation. In addition, we discuss issues around safety and representation, as well as methods we used to minimize the potential harm of our models.

The promise of multi-messenger astronomy relies on the rapid detection of gravitational waves at very low latencies ($\mathcal{O}$(1\,s)) in order to maximize the amount of time available for follow-up observations. In recent years, neural-networks have demonstrated robust non-linear modeling capabilities and millisecond-scale inference at a comparatively small computational footprint, making them an attractive family of algorithms in this context. However, integration of these algorithms into the gravitational-wave astrophysics research ecosystem has proven non-trivial. Here, we present the first fully machine learning-based pipeline for the detection of gravitational waves from compact binary coalescences (CBCs) running in low-latency. We demonstrate this pipeline to have a fraction of the latency of traditional matched filtering search pipelines while achieving state-of-the-art sensitivity to higher-mass stellar binary black holes.

Lung cancer (LC) remains the primary cause of cancer-related mortality, largely due to late-stage diagnoses. Effective strategies for early detection are therefore of paramount importance. In recent years, machine learning (ML) has demonstrated considerable potential in healthcare by facilitating the detection of various diseases. In this retrospective development and validation study, we developed an ML model based on dynamic ensemble selection (DES) for LC detection. The model leverages standard blood sample analysis and smoking history data from a large population at risk in Denmark. The study includes all patients examined on suspicion of LC in the Region of Southern Denmark from 2009 to 2018. We validated and compared the predictions by the DES model with diagnoses provided by five pulmonologists. Among the 38,944 patients, 9,940 had complete data of which 2,505 (25\%) had LC. The DES model achieved an area under the roc curve of 0.77$\pm$0.01, sensitivity of 76.2\%$\pm$2.4\%, specificity of 63.8\%$\pm$2.3\%, positive predictive value of 41.6\%$\pm$1.2\%, and F\textsubscript{1}-score of 53.8\%$\pm$1.1\%. The DES model outperformed all five pulmonologists, achieving a sensitivity 9\% higher than their average. The model identified smoking status, age, total calcium levels, neutrophil count, and lactate dehydrogenase as the most important factors for the detection of LC. The results highlight the successful application of the ML approach in detecting LC, surpassing pulmonologists' performance. Incorporating clinical and laboratory data in future risk assessment models can improve decision-making and facilitate timely referrals., Comment: 9 pages, 4 figures

Capturing the diversity of people in images is challenging: recent literature tends to focus on diversifying one or two attributes, requiring expensive attribute labels or building classifiers. We introduce a diverse people image ranking method which more flexibly aligns with human notions of people diversity in a less prescriptive, label-free manner. The Perception-Aligned Text-derived Human representation Space (PATHS) aims to capture all or many relevant features of people-related diversity, and, when used as the representation space in the standard Maximal Marginal Relevance (MMR) ranking algorithm, is better able to surface a range of types of people-related diversity (e.g. disability, cultural attire). PATHS is created in two stages. First, a text-guided approach is used to extract a person-diversity representation from a pre-trained image-text model. Then this representation is fine-tuned on perception judgments from human annotators so that it captures the aspects of people-related similarity that humans find most salient. Empirical results show that the PATHS method achieves diversity better than baseline methods, according to side-by-side ratings from human annotators.

We consider bosonic condensates of exciton-polaritons optically confined in elliptical traps. A superposition of two non-degenerated \textit{p}-type states of the condensate oriented along the two main axes of the trap is represented by a point on a Bloch sphere, being considered as an optically tunable qubit. We describe a set of universal single-qubit gates resulting in a controllable shift of the Bloch vector by means of an auxiliary laser beam. Moreover, we consider interaction mechanisms between two neighboring traps that enable designing two-qubit operations such as CPHASE, \textit{i}SWAP, and CNOT gates. Both the single- and two-qubit gates are analyzed in the presence of error sources in the context of polariton traps, such as pure dephasing and spontaneous relaxation mechanisms, leading to a fidelity reduction of the final qubit states and quantum concurrence, as well as the increase of Von Neumann entropy. We also discuss the applicability of our qubit proposal in the context of DiVincenzo's criteria for the realization of local quantum computing processes. Altogether, the developed set of quantum operations would pave the way to the realization of a variety of quantum algorithms in a planar microcavity with a set of optically induced elliptical traps.

The COVID-19 pandemic presented numerous challenges to healthcare systems worldwide. Given that lung infections are prevalent among COVID-19 patients, chest Computer Tomography (CT) scans have frequently been utilized as an alternative method for identifying COVID-19 conditions and various other types of pulmonary diseases. Deep learning architectures have emerged to automate the identification of pulmonary disease types by leveraging CT scan slices as inputs for classification models. This paper introduces COVID-VR, a novel approach for classifying pulmonary diseases based on volume rendering images of the lungs captured from multiple angles, thereby providing a comprehensive view of the entire lung in each image. To assess the effectiveness of our proposal, we compared it against competing strategies utilizing both private data obtained from partner hospitals and a publicly available dataset. The results demonstrate that our approach effectively identifies pulmonary lesions and performs competitively when compared to slice-based methods.

We theoretically investigate the possibility of obtaining Majorana zero modes (MZMs) in penta-silicene nanoribbons (p-SiNRs) with induced \textit{p}-wave superconductivity. The model explicitly considers an external magnetic field perpendicularly applied to the nanoribbon plane, as well as an extrinsic Rashba spin-orbit coupling (RSOC), in addition to the first nearest neighbor hopping term and \textit{p}-wave superconducting pairing. By analyzing the dispersion relation profiles, we observe the successive closing and reopening of the induced superconducting gap with a single spin component, indicating a spin-polarized topological phase transition (TPT). Correspondingly, the plots of the energy spectrum versus the chemical potential reveal the existence of zero-energy states with a preferential spin orientation characterized by nonoverlapping wave functions localized at opposite ends of the superconducting p-SiNRs. These findings strongly suggest the emergence of topologically protected, spin-polarized MZMs at the ends of the p-SiNRs with induced \textit{p}-wave superconducting pairing, which can be realized by proximitizing the nanoribbon with an \textit{s}-wave superconductor, such as lead. The proposal paves the way for silicene-based Majorana devices hosting multiple MZMs with a well-defined spin orientation, with possible applications in fault-tolerant quantum computing platforms and Majorana spintronics., Comment: 12 pages, 9 figures

Supersonic pre-transitional boundary layers flowing over porous flat and concave surfaces are studied using numerical and asymptotic methods. The porous wall is composed of thin equally-spaced cylindrical microcavities. The flow is perturbed by small-amplitude, free-stream vortical disturbances of the convected gust type. From the proximity of the leading edge, these external agents generate the compressible Klebanoff modes, i.e. low-frequency disturbances of the kinematic and thermal kind that grow algebraically downstream. For Klebanoff modes with a spanwise wavelength comparable with the boundary-layer thickness, the porous surface has a negligible effect on their growth. When the spanwise wavelength is instead larger than the boundary-layer thickness, these disturbances are effectively attenuated by the porous surface. For a specified set of frequency and wavelengths, the Klebanoff modes evolve into oblique Tollmien-Schlichting waves through a leading-edge-adjustment receptivity mechanism. The wavenumber of these waves is only slightly modified over the porous surface, while the growth rate increases, thus confirming previous experimental results. An asymptotic analysis based on the triple-deck theory confirms these numerical findings. When the wall is concave, the amplitude of the Klebanoff modes is enhanced by the wall curvature and is attenuated by the wall porosity during the initial development., Comment: Accepted for publication in Physical Review Fluids. 43 pages, 15 figures

We present one of the first cosmic ray muon flux-angle variation experiments on the QuarkNet Cosmic Ray Detector (QNCRD). We first describe QNCRD and its calibration. The main focus is then quantifying muon flux decrease as a function of angle from the zenith. The angle of counters of QNCRD were incremented 15 degrees on average every $3.1$ days over the range of 0 degrees to 90 degrees for a period of approximately one month. Results showed that as the angle of the detector increased from the zenith, muon flux decreased, which agrees with previous studies. An estimate for the flux based on the model $I(\theta)=I_0cos(\theta)^n$ had an exponent value of $n=1.39 \pm 0.01$ for $\theta < 75$ degrees, an underestimate of values in other literature. These findings provided a reasonable, although not entirely accurate, estimate for the value of $n$ considering the duration of the study and sensitivity of the instrument. Our results constrain the accuracy of QNCRD and provide a source for future long-term experiments. This study also demonstrates the feasibility of conducting science experiments in high school classrooms, increasing science accessibility., Comment: 6 pages, 5 figures, 4 tables

We present fluxes and light curves for a population of asteroids at millimeter (mm) wavelengths, detected by the Atacama Cosmology Telescope (ACT) over 18, 000 deg2 of the sky using data from 2017 to 2021. We utilize high cadence maps, which can be used in searching for moving objects such as asteroids and trans-Neptunian Objects (TNOs), as well as for studying transients. We detect 160 asteroids with a signal-to-noise of at least 5 in at least one of the ACT observing bands, which are centered near 90, 150, and 220 GHz. For each asteroid, we compare the ACT measured flux to predicted fluxes from the Near Earth Asteroid Thermal Model (NEATM) fit to WISE data. We confirm previous results that detected a deficit of flux at millimeter wavelengths. Moreover, we report a spectral characteristic to this deficit, such that the flux is relatively lower at 150 and 220 GHz than at 90 GHz. Additionally, we find that the deficit in flux is greater for S-type asteroids than for C-type., Comment: 15 pages, 9 Figures, 4 Tables

The Gr\"uneisen ratio $\Gamma$, i.e., the singular part of the ratio of thermal expansion to the specific heat, has been broadly employed to explore both finite-$T$ and quantum critical points (QCPs). For a genuine quantum phase transition (QPT), thermal fluctuations are absent and thus the thermodynamic $\Gamma$ cannot be employed. We propose a quantum analogue to $\Gamma$ that computes entanglement as a function of a tuning parameter $\lambda$ and show that QPTs take place only for systems in which the ground-state energy depends on $\lambda$ non-linearly. Furthermore, we demonstrate the breakdown of the Hellmann-Feynman theorem in the thermodynamic limit at any QCP. We showcase our approach using the quantum 1D Ising model with transverse field and Kane's quantum computer. The slowing down of the dynamics and thus the "creation of mass" close to any QCP/QPT is also discussed., Comment: 5 pages, 3 figures, comments are wellcome!

Understanding different human attributes and how they affect model behavior may become a standard need for all model creation and usage, from traditional computer vision tasks to the newest multimodal generative AI systems. In computer vision specifically, we have relied on datasets augmented with perceived attribute signals (e.g., gender presentation, skin tone, and age) and benchmarks enabled by these datasets. Typically labels for these tasks come from human annotators. However, annotating attribute signals, especially skin tone, is a difficult and subjective task. Perceived skin tone is affected by technical factors, like lighting conditions, and social factors that shape an annotator's lived experience. This paper examines the subjectivity of skin tone annotation through a series of annotation experiments using the Monk Skin Tone (MST) scale, a small pool of professional photographers, and a much larger pool of trained crowdsourced annotators. Along with this study we release the Monk Skin Tone Examples (MST-E) dataset, containing 1515 images and 31 videos spread across the full MST scale. MST-E is designed to help train human annotators to annotate MST effectively. Our study shows that annotators can reliably annotate skin tone in a way that aligns with an expert in the MST scale, even under challenging environmental conditions. We also find evidence that annotators from different geographic regions rely on different mental models of MST categories resulting in annotations that systematically vary across regions. Given this, we advise practitioners to use a diverse set of annotators and a higher replication count for each image when annotating skin tone for fairness research.

Photo-excitation at terahertz and mid-infrared frequencies has emerged as a new way to manipulate functionalities in quantum materials, in some cases creating non-equilibrium phases that have no equilibrium analogue. In K$_3$C$_{60}$, a metastable zero-resistance phase was documented with optical properties and pressure dependences compatible with non-equilibrium high temperature superconductivity. Here, we report the discovery of a dominant energy scale for this phenomenon, along with the demonstration of a giant increase in photo-susceptibility near 10 THz excitation frequency. At these drive frequencies a metastable superconducting-like phase is observed up to room temperature for fluences as low as ~400 $\mu J/cm^2$. These findings shed light on the microscopic mechanism underlying photo-induced superconductivity. They also trace a path towards steady state operation, currently limited by the availability of a suitable high-repetition rate optical source at these frequencies., Comment: 35 pages, 13 figures, including supplementary

We theoretically investigate the spectral properties of a quantum impurity (QI) hosting the here proposed {Majorana-Ising-type quasiparticle (MIQ) excitation}. It arises from the coupling between a finite topological superconductor (TSC) based on a chain of magnetic adatoms-superconducting hybrid system and an integer large spin $S$ flanking the QI. Noteworthy, the spin $S$ couples to the QI via the Ising-type exchange interaction. As the Majorana zero-modes (MZMs) at the edges of the TSC chain are overlapped, we counterintuitively find a regime wherein the Ising term modulates the localization of a fractionalized and resonant MZM at the QI site. Interestingly enough, the fermionic nature of this state is revealed as purely of electron tunneling-type and most astonishingly, it has the Andreev conductance completely null in its birth. Therefore, we find that a resonant edge state appears as a zero-mode and discuss it in terms of a poor man's Majorana[Nature 614, 445 (2023)].