Search

Terahertz (THz) generation via photomixing on photoconductive antenna using twin delayed chirped pulses provides a long THz pulse with a narrow bandwidth. To generate a long pulse with broad bandwidth, we propose a new method that combines two long optical pulses with opposite chirps. The pulses exhibit temporal distributions of their instantaneous frequencies with opposite slopes. As a result, interaction between the beat frequency evolving over time and a photoconductor produces a broad THz spectrum with temporal variations. In our experimental setup, we generate a 12 ps-long pulse with a 1 THz bandwidth spectrum, featuring a frequency ramp of 90 GHz/ps, resembling a chirped THz pulse. This approach signifies a major advancement toward integrating photomixer technology, particularly in THz ranging applications.

The Micro Vertex Detector (MVD) is the most upstream detector of the fixed-target Compressed Baryonic Matter Experiment (CBM) at the future Facility for Antiproton and Ion Research (FAIR). It enables high-precision low-momentum tracking in direct proximity of the target. Reaching the stringent requirements for the MVD, a material budget of~$0.3\,-\,0.5\%\,X_0$ per layer, operating the dedicated CMOS MAPS~(`MIMOSIS') in the target vacuum, the strong magnetic dipole field, and a harsh radiation environment~(5\,Mrad, $7\times10^{13}\,n_{\text{eq}}/\text{cm}^2$ per CBM year), poses an unprecedented integration challenge. In this paper, the integration concept of the detector is be outlined, elaborating on the selection and preparation of materials, assembly procedures, and quality assessment steps in the ongoing preparation of pre-series production and detector commissioning in 2028., Comment: Titlepage + 5 pages, 9 figures, proceedings for Pixel 2024

Spin-orbit interaction couples the spin of a particle to its motion and leads to spin-induced transport phenomena such as spin-Hall effects and Chern insulators. In this work, we extend the concept of internal-external state coupling to higher internal symmetry, exploring features beyond the established spin-orbit regime. We couple suitable resonant laser beams to a gas of ultracold atoms, thereby inducing artificial SU(3) non-Abelian gauge fields that act on a degenerate ground state manifold comprised of three dark states. We demonstrate the inherent all-state connectivity of SU(3) systems by performing targeted geometric transformations. Then, we investigate color-orbit coupling, an extension of SU(2) spin-orbit coupling to SU(3) systems. We reveal a rich dynamical interplay between three distinct oscillation frequencies, which possesses interesting analogies with neutrino oscillations and quark mixing mechanisms. In the future, the system should provide a testbed for exploring topological properties of SU(3) systems., Comment: 6 pages, 3 figures, 1 table. SM is available upon request

In the recent years, high energy physics discoveries have been driven by the increasing of luminosity and/or detector granularity. This evolution gives access to bigger statistics and data samples, but can make it hard to process results with current methods and algorithms. Graph convolution networks, have been shown to be powerful tools to address these challenges. We present our graph convolution framework for particle identification and energy regression in high granularity calorimeters. In particular, we introduce our algorithm for optimised graph construction in resource constrained environments. We also introduce our implementation of graph convolution and pooling layers. We observe satisfying accuracies, and discuss possible application to other high granularity particle detector challenges., Comment: Submitted for proceedings of the 2024 conference on Computing in High Energy Physics

This thesis is split up into two parts: The first one concerns (pseudo)-holomorphic Hamiltonian systems, while the second part is about K\"ahler structures of complex coadjoint orbits. We begin the first part by investigating basic properties of holomorphic Hamiltonian systems (HHSs) like maximal holomorphic trajectories and holomorphic Hamiltonian foliations. Afterwards, we use these notions to combine HHSs with two structures frequently studied in geometry, namely Lefschetz and almost toric fibrations, leading us to the notion of a holomorphic symplectic Lefschetz fibration. Following this examination, we formulate action functionals for HHSs. These action functionals are well-suited to find periodic orbits. However, it turns out that HHSs rarely exhibit periodic orbits. One possible obstruction for a HHS to possess periodic orbits is the integrability of the underlying complex structure. To circumvent this obstruction, we introduce the notion of pseudo-holomorphic Hamiltonian systems (PHHSs) which allow us to describe classical mechanics on almost complex manifolds. We show that PHHSs satisfy almost the same properties as HHSs. In the second part, we study coadjoint orbits of complex Lie groups and show that they also exhibit, next to their Hyperk\"ahler structure which was introduced by Kronheimer and Kovalev in the 90s, a holomorphic K\"ahler structure. A holomorphic K\"ahler structure can be thought of as a complexification of a usual K\"ahler structure. We call the fact that a space admits both Hyperk\"ahler and holomorphic K\"ahler structures ''K\"ahler duality''. In this thesis, we suspect that the K\"ahler duality of complex coadjoint orbits can be traced back to double cotangent bundles. Precisely speaking, we conjecture that double cotangent bundles naturally exhibit K\"ahler duality and that this K\"ahler duality transfers via reduction to complex coadjoint orbits., Comment: 136 + 100 (appendix) pages; Doctoral Thesis; Already published by the University of Augsburg (https://opus.bibliothek.uni-augsburg.de/opus4/frontdoor/index/index/docId/117987)

The LHCb experiment at CERN has been upgraded for the Run 3 operation of the Large Hadron Collider (LHC). A new concept of tracking detector based on Scintillating Fibres (SciFi) read out with multichannel silicon photomultipliers (SiPMs) was installed during its upgrade. One of the main challenges the SciFi tracker will face during the Run 4 operation of the LHC is the higher radiation environment due to fast neutrons, where the SiPMs are located. To cope with the increase in radiation, cryogenic cooling with liquid nitrogen is being investigated as a possible solution to mitigate the performance degradation of the SiPMs induced by radiation damage. Thus, a detailed performance study of different layouts of SiPM arrays produced by Fondazione Bruno Kessler (FBK) and Hamamatsu Photonics K.K. is being carried out. These SiPMs have been designed to operate at cryogenic temperatures. Several SiPMs have been tested in a dedicated cryogenic setup down to 100 K. Key performance parameters such as breakdown voltage, dark count rate, photon detection efficiency, gain and direct cross-talk are characterized as a function of the temperature. The main results of this study are going to be presented here., Comment: Prepared for submission to JINST Proceedings 6th International Workshop on New Photon-Detector

In a digraph, a kernel is a subset of vertices that is both independent and absorbing. Kernels have important applications in combinatorics and outside. Kernels do not always exist and finding sufficient conditions ensuring their existence is a key theoretical challenge. In this work, we revisit and generalize a few classical results of this sort, especially the Sands--Sauer--Woodrow theorem and the Galeana-S\'anchez--Neumann-Lara theorem.

This paper explores the temporal robustness of language models (LMs) in handling factual knowledge. While LMs can often complete simple factual statements, their ability to manage temporal facts (those valid only within specific timeframes) remains uncertain. We design a controlled experiment to test the robustness of temporal factual knowledge inside LMs, which we use to evaluate several pretrained and instruction-tuned models using prompts on popular Wikidata facts, assessing their performance across different temporal granularities (Day, Month, and Year). Our findings indicate that even very large state-of-the-art models, such as Llama-3.1-70B, vastly lack robust knowledge of temporal facts. In addition, they are incapable of generalizing their knowledge from one granularity to another. These results highlight the inherent limitations of using LMs as temporal knowledge bases. The source code and data to reproduce our experiments will be released.

The SPIRAL2 superconducting linear accelerator (LINAC), which has been operational since 2019, employs superconducting, independently phased RF resonators to deliver a wide range of particle beams. Designed for flexibility in particle types, intensities, and energies, it utilizes superconducting quarter-wave resonators (QWRs), whose performances are critically dependent on the reliability of the cryogenic operation. This paper reviews the evolution from commissioning to the routine operation of the SPIRAL2 cryogenic system, initially commissioned in 2017. It highlights the key challenges encountered, including thermo-acoustic oscillations, thermal management, and abnormal behavior of cavities. Furthermore, it explores the integration of thermodynamic modeling and machine learning techniques to enhance system control and diagnose issues. This work serves as a comprehensive resource for advancing the cryogenic operation and performance of superconducting LINACs.

We identify a chain of galaxies along an almost straight line in the nearby Universe with a projected length of ~5 Mpc. The galaxies are distributed within projected distances of only 7-105 kpc from the axis of the identified filament. They have redshifts in a very small range of z=0.0361-0.0370 so that their radial velocities are consistent with galaxy proper motions. The filament galaxies are mainly star-forming and have stellar masses in a range of $\rm 10^{9.1}-10^{10.7}\,M_{\odot}$. We search for systems with similar geometrical properties in the full-sky mock galaxy catalogue of the MillenniumTNG simulations and find that although such straight filaments are unusual and rare, they are predicted by $\Lambda$CDM simulations (4% incidence). We study the cold HI gas in a 1.3 Mpc section of the filament through HI-21cm emission line observations and detect eleven HI sources, many more than expected from the HI mass function in a similar volume. They have HI masses $\rm 10^{8.5}-10^{9.5}\,M_{\odot}$ and are mostly within ~120 kpc projected distance from the filament axis. None of these HI sources has a confirmed optical counterpart. Their darkness together with their large HI-21cm line-widths indicate that they contain gas that might not yet be virialized. These clouds must be marking the peaks of the dark matter and HI distributions over large scales within the filament. The presence of such gas clouds around the filament spines is predicted by simulations, but this is the first time that the existence of such clouds in a filament is observationally confirmed., Comment: Published in ApJL. Main figures: Fig. 1 and Fig. 3

Low-Rank Adaptation (LoRA) and its variants have shown impressive results in reducing the number of trainable parameters and memory requirements of large transformer networks while maintaining fine-tuning performance. However, the low-rank nature of the weight update inherently limits the representation power of fine-tuned models, potentially compromising performance on complex tasks. This raises a critical question: when a performance gap between LoRA and standard fine-tuning is observed, is it due to the reduced number of trainable parameters or the rank deficiency? This paper aims to answer this question by introducing RandLoRA, a parameter-efficient method that performs full-rank updates using a learned linear combinations of low-rank, non-trainable random matrices. Our method limits the number of trainable parameters by restricting optimization to diagonal scaling matrices applied to the fixed random matrices. This allows us to effectively overcome the low-rank limitations while maintaining parameter and memory efficiency during training. Through extensive experimentation across vision, language, and vision-language benchmarks, we systematically evaluate the limitations of LoRA and existing random basis methods. Our findings reveal that full-rank updates are beneficial across vision and language tasks individually, and even more so for vision-language tasks, where RandLoRA significantly reduces -- and sometimes eliminates -- the performance gap between standard fine-tuning and LoRA, demonstrating its efficacy., Comment: To appear at the International Conference on Learning Representations (ICLR) 2025

This paper presents a novel approach to the joint optimization of job scheduling and data allocation in grid computing environments. We formulate this joint optimization problem as a mixed integer quadratically constrained program. To tackle the nonlinearity in the constraint, we alternatively fix a subset of decision variables and optimize the remaining ones via Mixed Integer Linear Programming (MILP). We solve the MILP problem at each iteration via an off-the-shelf MILP solver. Our experimental results show that our method significantly outperforms existing heuristic methods, employing either independent optimization or joint optimization strategies. We have also verified the generalization ability of our method over grid environments with various sizes and its high robustness to the algorithm hyper-parameters.

We consider radially symmetric solutions of the degenerate Keller-Segel system \begin{align*} \begin{cases} \partial_t u=\nabla\cdot (u^{m-1}\nabla u - u\nabla v),\\ 0=\Delta v -\mu +u,\quad\mu =\frac{1}{|\Omega|}\int_\Omega u, \end{cases} \end{align*} in balls $\Omega\subset\mathbb R^n$, $n\ge 1$, where $m>1$ is arbitrary. Our main result states that the initial evolution of the positivity set of $u$ is essentially determined by the shape of the (nonnegative, radially symmetric, H\"older continuous) initial data $u_0$ near the boundary of its support $\overline{B_{r_1}(0)}\subsetneq\Omega$: It shrinks for sufficiently flat and expands for sufficiently steep $u_0$. More precisely, there exists an explicit constant $A_{\mathrm{crit}} \in (0, \infty)$ (depending only on $m, n, R, r_1$ and $\int_\Omega u_0$) such that if \begin{align*} u_0(x)\le A(r_1-|x|)^\frac{1}{m-1} \qquad \text{for all $|x|\in(r_0, r_1)$ and some $r_0\in(0,r_1)$ and $A0$ and $\zeta>0$ such that $\sup\{\, |x| \mid x \in \operatorname{supp} u(\cdot, t)\,\}\le r_1 -\zeta t$ for all $t\in(0, T)$, while if \begin{align*} u_0(x)\ge A(r_1-|x|)^\frac{1}{m-1} \qquad \text{for all $|x|\in(r_0, r_1)$ and some $r_0 \in (0, r_1)$ and $A>A_{\mathrm{crit}}$}, \end{align*} then we can find $T>0$ and $\zeta>0$ such that $\sup\{\, |x| \mid x \in \operatorname{supp} u(\cdot, t)\,\}\ge r_1 +\zeta t$ for all $t\in(0, T)$.

A longstanding open question in sub-Riemannian geometry is the following: are sub-Riemannian length minimizers smooth? We give a negative answer to this question, exhibiting an example of a $C^2$ but not $C^3$ length-minimizer of a real-analytic (even polynomial) sub-Riemannian structure.

This report provides an overview of the workshop titled Autonomy and Safety Assurance in the Early Development of Robotics and Autonomous Systems, hosted by the Centre for Robotic Autonomy in Demanding and Long-Lasting Environments (CRADLE) on September 2, 2024, at The University of Manchester, UK. The event brought together representatives from six regulatory and assurance bodies across diverse sectors to discuss challenges and evidence for ensuring the safety of autonomous and robotic systems, particularly autonomous inspection robots (AIR). The workshop featured six invited talks by the regulatory and assurance bodies. CRADLE aims to make assurance an integral part of engineering reliable, transparent, and trustworthy autonomous systems. Key discussions revolved around three research questions: (i) challenges in assuring safety for AIR; (ii) evidence for safety assurance; and (iii) how assurance cases need to differ for autonomous systems. Following the invited talks, the breakout groups further discussed the research questions using case studies from ground (rail), nuclear, underwater, and drone-based AIR. This workshop offered a valuable opportunity for representatives from industry, academia, and regulatory bodies to discuss challenges related to assured autonomy. Feedback from participants indicated a strong willingness to adopt a design-for-assurance process to ensure that robots are developed and verified to meet regulatory expectations., Comment: 7 pages, 2 figures

We introduce an evaluation methodology for reading comprehension tasks based on the intuition that certain examples, by the virtue of their linguistic complexity, consistently yield lower scores regardless of model size or architecture. We capitalize on semantic frame annotation for characterizing this complexity, and study seven complexity factors that may account for model's difficulty. We first deploy this methodology on a carefully annotated French reading comprehension benchmark showing that two of those complexity factors are indeed good predictors of models' failure, while others are less so. We further deploy our methodology on a well studied English benchmark by using Chat-GPT as a proxy for semantic annotation. Our study reveals that fine-grained linguisticallymotivated automatic evaluation of a reading comprehension task is not only possible, but helps understand models' abilities to handle specific linguistic characteristics of input examples. It also shows that current state-of-the-art models fail with some for those characteristics which suggests that adequately handling them requires more than merely increasing model size.

Polarized K3 surfaces of genus sixteen have a Mukai vector bundle of rank two. We study the geometry of the projectivization of this bundle. We prove that it has an embedding in $\mathbb{P}_9$ with an ideal generated by quadrics. We give an effective method to compute these quadrics from a general choice in Mukai's unirationalization of the moduli space. This linear system gives a double cover of $\mathbb{P}_9$ ramified on a degree $10$ hypersurface. It gives relative Weddle/Kummer surfaces over a Peskine variety associated to an explicit trivector. This work is also motivated by hyperk\"ahler geometry and Debarre-Voisin varieties. Oberdieck showed that the Hilbert square of a general K3-surface of genus $16$ is a Debarre-Voisin variety for some trivector. We start to investigate the relationship between these two trivectors.

Deep learning algorithms -- typically consisting of a class of deep artificial neural networks (ANNs) trained by a stochastic gradient descent (SGD) optimization method -- are nowadays an integral part in many areas of science, industry, and also our day to day life. Roughly speaking, in their most basic form, ANNs can be regarded as functions that consist of a series of compositions of affine-linear functions with multidimensional versions of so-called activation functions. One of the most popular of such activation functions is the rectified linear unit (ReLU) function $\mathbb{R} \ni x \mapsto \max\{ x, 0 \} \in \mathbb{R}$. The ReLU function is, however, not differentiable and, typically, this lack of regularity transfers to the cost function of the supervised learning problem under consideration. Regardless of this lack of differentiability issue, deep learning practioners apply SGD methods based on suitably generalized gradients in standard deep learning libraries like {\sc TensorFlow} or {\sc Pytorch}. In this work we reveal an accurate and concise mathematical description of such generalized gradients in the training of deep fully-connected feedforward ANNs and we also study the resulting generalized gradient function analytically. Specifically, we provide an appropriate approximation procedure that uniquely describes the generalized gradient function, we prove that the generalized gradients are limiting Fr\'echet subgradients of the cost functional, and we conclude that the generalized gradients must coincide with the standard gradient of the cost functional on every open sets on which the cost functional is continuously differentiable., Comment: 38 pages

Advances in computer vision, particularly in optical image-based 3D reconstruction and feature matching, enable applications like marker-less surgical navigation and digitization of surgery. However, their development is hindered by a lack of suitable datasets with 3D ground truth. This work explores an approach to generating realistic and accurate ex vivo datasets tailored for 3D reconstruction and feature matching in open orthopedic surgery. A set of posed images and an accurately registered ground truth surface mesh of the scene are required to develop vision-based 3D reconstruction and matching methods suitable for surgery. We propose a framework consisting of three core steps and compare different methods for each step: 3D scanning, calibration of viewpoints for a set of high-resolution RGB images, and an optical-based method for scene registration. We evaluate each step of this framework on an ex vivo scoliosis surgery using a pig spine, conducted under real operating room conditions. A mean 3D Euclidean error of 0.35 mm is achieved with respect to the 3D ground truth. The proposed method results in submillimeter accurate 3D ground truths and surgical images with a spatial resolution of 0.1 mm. This opens the door to acquiring future surgical datasets for high-precision applications., Comment: 18 pages, 12 figures. Submitted to the 16th International Conference on Information Processing in Computer-Assisted Interventions (IPCAI 2025)

We investigate the effect of the microstructure on the mechanical response of lattice structures made of a defect tolerant Ti-14 Mo binary alloy fabricated by Electron Powder Bed Fusion. Some of the lattice structures were subjected to different heat treatments to compare different microstructures: a two-phase $\alpha$+$\beta$ microstructure inherited from the fabrication (as-built), a full $\beta$-metastable microstructure, and a $\beta$-metastable microstructure strengthened via nanoprecipitation of the $\omega$iso phase. We demonstrate that the macroscopic mechanical behavior of one of the most popular stretching-dominated lattice structure, namely the octet-truss, can be tailored through microstructural optimization by increasing the tolerance to plastic buckling of the constitutive struts. Defect-tolerant materials pave the way towards lattice structures fabricated by additive manufacturing with optimized energy absorption performances.

Multi-Agent Path Finding (MAPF) focuses on determining conflict-free paths for multiple agents navigating through a shared space to reach specified goal locations. This problem becomes computationally challenging, particularly when handling large numbers of agents, as frequently encountered in practical applications like coordinating autonomous vehicles. Quantum computing (QC) is a promising candidate in overcoming such limits. However, current quantum hardware is still in its infancy and thus limited in terms of computing power and error robustness. In this work, we present the first optimal hybrid quantum-classical MAPF algorithm which is based on branch-and-cut-and-prize. QC is integrated by iteratively solving QUBO problems, based on conflict graphs. Experiments on actual quantum hardware and results on benchmark data suggest that our approach dominates previous QUBO formulations and baseline MAPF solvers.

Determining a beam's full trajectory requires tracking both its position and momentum (angular) information. However, the product of position and momentum uncertainty in a simultaneous measurement of the two parameters is bound by the Heisenberg uncertainty limit (HUL). In this work, we present a proof-of-principle demonstration of a quantum-enhanced beam tracking technique, leveraging the inherent position and momentum entanglement between photons produced via spontaneous parametric down-conversion (SPDC). We show that quantum entanglement can be exploited to achieve a beam tracking accuracy beyond the HUL in a simultaneous measurement. Moreover, with existing detection technologies, it is already possible to achieve near real-time beam tracking capabilities at the single-photon level. The technique also exhibits high resilience to background influences, with negligible reduction in tracking accuracy even when subjected to a disruptive beam that is significantly brighter than SPDC., Comment: 7 pages, 5 figures

Using a time-dependent density matrix renormalization group approach, we study the dynamical structure factor of the $J_1 - J_2$ spin-1 chain. As $J_2$ increases, the magnon mode develops incommensurability. The system undergoes a first-order transition at $J_2 = 0.76 J_1$, and at that point, domain walls lead to a continuum of fractional quasi-particles or spinons. By studying small variations in $J_2$ around the transition point, we observe the confinement of spinons into bound states in the spectral function and find a smooth evolution of the spectrum into magnon modes away from the phase transition. We employ the single-mode approximation to accurately account for the dispersion of the magnon mode away from the phase transition and describe the associated continua and bound states. We extend the single-mode approximation to describe the dispersion of a spinon at the phase transition point and obtain its dispersion throughout the Brillouin zone. This allows us to relate the incommensurability at and around the transition point to the competition between a negative nearest-neighbour hopping amplitude and a positive next-nearest-neighbour one for the domain wall., Comment: 16 pages, 13 Figures

Improving the efficiency of photothermal (PT) therapies and photoacoustic (PA) imaging at the microscopic scale hinges on developing multiphoton-absorbing photothermal molecules or contrast agents that operate in the near-infrared (NIR) range. These advanced agents or molecules will enable excitation with NIR lasers, in an improved transparency range for biological tissues, while enabling minute, highly localized spatial control of the excitation area. However, progress in this field requires innovative experimental techniques to characterize photothermal and photoacoustic effects under multiphoton excitation. In this article, we showcase a study of a model organometallic molecular compound excited via two-photon absorption (2PA) using femtosecond laser pulses. Based on a time-domain Brillouin scattering technique, well adapted for investigating ultrafast nonlinear optical absorption processes in ultrathin films on substrates, we determine the effective nonlinear absorption coefficients of the compound directly linked to PT/PA. Our findings provide a practical approach for exploring and optimizing nonlinear PT/PA absorbers and contrast agents.

We present an all-silicon source of near-infrared linearly-polarized single photons, fabricated by nanoscale positioning of a color center in a silicon-on-insulator microcavity. The color center consists of a single W center, created at a well-defined position by Si$^{+}$ ion implantation through a 150 nm-diameter nanohole in a mask. A circular Bragg grating cavity resonant with the W's zero-phonon line at 1217 nm is fabricated at the same location as the nanohole. Under above-gap continuous-wave excitation, a very clean photon antibunching behavior ($g{^2} \leq 0.06$) is observed over the entire power range, which highlights the absence of parasitic emitters. Purcell-enhancement of W's zero-phonon emission provides both a record-high photoluminescence count rate among Si color centers (ca $1.2 \times 10^{6}$ counts/s) and apparent Debye-Waller factor around 99%. We also demonstrate the triggered emission of single photons with 93% purity under weak pulsed laser excitation. At high pulsed laser power, we reveal a detrimental effect of repumping processes, that could be mitigated using selective pumping schemes in the future. These results represent a major step towards on-demand sources of indistinguishable near-infrared single photons within silicon photonics chips.

Training task-oriented dialogue systems is both costly and time-consuming, due to the need for high-quality datasets encompassing diverse intents. Traditional methods depend on extensive human annotation, while recent advancements leverage large language models (LLMs) to generate synthetic data. However, these approaches often require custom prompts or code, limiting accessibility for non-technical users. We introduce GraphTOD, an end-to-end framework that simplifies the generation of task-oriented dialogues. Users can create dialogues by specifying transition graphs in JSON format. Our evaluation demonstrates that GraphTOD generates high-quality dialogues across various domains, significantly lowering the cost and complexity of dataset creation.

We extend to soft repulsive interaction potentials a recently proposed irreversible swap algorithm originally designed for polydisperse hard spheres. The original algorithm performs rejection-free, irreversible, collective swap moves. We show that event-driven cluster updates of particle diameters can also be performed in continuous potentials by introducing a factorised Metropolis probability. However, the Metropolis factorisation needed to deal with continuous potentials decreases the efficiency of the algorithm and mitigates the benefits of breaking detailed balance. This leads us to propose another irreversible swap algorithm using the standard Metropolis probability that accelerates the relaxation of soft sphere glasses at low temperatures, compared to the original swap algorithm. We apply these efficient swap algorithms to produce very stable inherent structures with vibrational density of states lacking the quasi-localised excitations observed in conventional glasses., Comment: 11 pages, 6 figures

Merging binary black holes (BBHs) formed dynamically in dense star clusters are expected to have uncorrelated spin--orbit orientations since they are assembled through many random interactions. However, measured effective spins in BBHs detected by LIGO/Virgo/KAGRA hint at additional physical processes that may introduce anisotropy. Here we address this question by exploring the impact of stellar collisions, and accretion of collision debris, on the spin--orbit alignment in merging BBHs formed in dense star clusters. Through hydrodynamic simulations, we study the regime where the disruption of a massive star by a BBH causes the stellar debris to form individual accretion disks bound to each black hole. We show that these disks, which are randomly oriented relative to the binary orbital plane after the initial disruption of the star, can be reoriented by strong tidal torques in the binary near pericenter passages. Following accretion by the BHs on longer timescales, BBHs with small but preferentially positive effective spin parameters ($\chi_{\rm eff} \lesssim 0.2$) are formed. Our results indicate that BBH collisions in young massive star clusters could contribute to the observed trend toward small positive $\chi_{\rm eff}$, and we suggest that the standard assumption often made that dynamically assembled BBHs should have isotropically distributed BH spins is not always justified., Comment: 13 pages, 6 figures, V1

We present PhyDeformer, a new deformation method for high-quality garment mesh registration. It operates in two phases: In the first phase, a garment grading is performed to achieve a coarse 3D alignment between the mesh template and the target mesh, accounting for proportional scaling and fit (e.g. length, size). Then, the graded mesh is refined to align with the fine-grained details of the 3D target through an optimization coupled with the Jacobian-based deformation framework. Both quantitative and qualitative evaluations on synthetic and real garments highlight the effectiveness of our method.

The giant, star forming clumps in gas-rich, high redshift disks are commonly assumed to form due to gravitational instabilities, in which protoclumps have a Toomre-$Q$ parameter less than unity. However, some cosmological simulations show that clumps can form in regions where $Q\gg1$. In these simulations, there is an excess of compressive modes of turbulence that lead to gravitational collapse of regions that were not supposed to gravitationally collapse, according to linear theory. In contrast, sites of clump formation in isolated simulations do not show this excess, hinting that the origin may be external. We explore two external mechanisms that can induce compressive modes of disk turbulence in protoclumps, namely, compressive tides exerted by the cosmological environment and the direct driving by inflowing streams. We correlate the local strength of compressive tides and the amount of fresh stream material with protoclump regions in zoom-in cosmological simulations. The local strength of compressive tides is derived from the tidal tensor. The local strength of incoming streams is derived from the fractional presence of the stream compared to the average. We find that the tidal field in protoclumps tends to be over-compressive while random patches in the disk show diverging tides. In particular, in $25\%$ of the protoclumps, the tidal field is fully compressive, while no random patch resides in regions of fully compressive tides. In addition, protoclumps tend to reside in regions where the fraction of incoming stream mass is 2-10 times larger than the average at the same galactocentric radius. Both compressive tides and inflowing streams are correlated with the protoclumps and can thus serve as the drivers of excessive compressive turbulence that can initiate clump formation. This constitutes a new, non-linear mode of violent disk instabilities in high-$z$ galaxies., Comment: 15 pages, submitted to A&A

Weak supervision (WS) is a popular approach for label-efficient learning, leveraging diverse sources of noisy but inexpensive weak labels to automatically annotate training data. Despite its wide usage, WS and its practical value are challenging to benchmark due to the many knobs in its setup, including: data sources, labeling functions (LFs), aggregation techniques (called label models), and end model pipelines. Existing evaluation suites tend to be limited, focusing on particular components or specialized use cases. Moreover, they often involve simplistic benchmark tasks or de-facto LF sets that are suboptimally written, producing insights that may not generalize to real-world settings. We address these limitations by introducing a new benchmark, BOXWRENCH, designed to more accurately reflect real-world usages of WS. This benchmark features tasks with (1) higher class cardinality and imbalance, (2) notable domain expertise requirements, and (3) opportunities to re-use LFs across parallel multilingual corpora. For all tasks, LFs are written using a careful procedure aimed at mimicking real-world settings. In contrast to existing WS benchmarks, we show that supervised learning requires substantial amounts (1000+) of labeled examples to match WS in many settings., Comment: NeurIPS 2024 Datasets and Benchmarks Track

Foundation models are trained on large-scale web-crawled datasets, which often contain noise, biases, and irrelevant information. This motivates the use of data selection techniques, which can be divided into model-free variants -- relying on heuristic rules and downstream datasets -- and model-based, e.g., using influence functions. The former can be expensive to design and risk introducing unwanted dependencies, while the latter are often computationally prohibitive. Instead, we propose an efficient, model-based approach using the Mimic Score, a new data quality metric that leverages the weights of a reference model to assess the usefulness of individual samples for training a new model. It relies on the alignment between gradients and a target direction induced by the reference model. Using the derived Mimic Scores, we develop Grad-Mimic, a framework that prioritizes samples for learning, creates effective filters, and automates data selection. Empirically, using Mimic Scores to guide training improves data efficiency, results in consistent performance gains across six image datasets, and includes enhancements to CLIP models. Moreover, Mimic Score-based filters improve upon existing filtering methods, e.g., cutting 4.7 million samples to train better CLIP models while offering accurate estimation of training dataset quality.

This paper explores a novel approach aimed at overcoming existing challenges in the realm of local search algorithms. Our aim is to improve the decision process that takes place within a local search algorithm so as to make the best possible transitions in the neighborhood at each iteration. To improve this process, we propose to use a neural network that has the same input information as conventional local search algorithms. In this paper, which is an extension of the work [Goudet et al. 2024] presented at EvoCOP2024, we investigate different ways of representing this information so as to make the algorithm as efficient as possible but also robust to monotonic transformations of the problem objective function. To assess the efficiency of this approach, we develop an experimental setup centered around NK landscape problems, offering the flexibility to adjust problem size and ruggedness. This approach offers a promising avenue for the emergence of new local search algorithms and the improvement of their problem-solving capabilities for black-box problems.

In the unit tangent bundle of noncompact finite volume negatively curved Riemannian manifolds, we prove the equidistribution towards the measure of maximal entropy for the geodesic flow of the Lebesgue measure along the divergent geodesic flow orbits, as their complexity tends to infinity. We prove the analogous result for geometrically finite tree quotients, where the equidistribution takes place in the quotient space of geodesic lines towards the Bowen-Margulis measure., Comment: 27 pages

Motivated by recent experiments on the quantum magnet K$_{2}$Co(SeO$_{3}$)$_{2}$, we study theoretically the excitation spectrum of the nearest-neighbour triangular XXZ model in the limit of strong easy-axis anisotropy, within the up-up-down (1/3-plateau) phase. We make an expansion in $\alpha =J_{xy}/J_{zz}$ instead of $1/S$ and calculate the magnon dispersion for any value of the spin $S$ at second order in $\alpha$, with two important conclusions: (i) the 1/S expansion converges very slowly for S=1/2, making spin-wave theory quantitatively inaccurate up to very large orders; (ii) compared to the linear spin-wave predictions, our magnon dispersion presents a much better agreement with experimental results on K$_{2}$Co(SeO$_{3}$)$_{2}$, for which $\alpha \simeq 0.08$., Comment: 6 pages, 3 figures

We consider the small-time local controllability in the vicinity of the ground state of a bilinear Schr\"odinger equation with Neumann boundary conditions. We prove that, when the linearized system is not controllable, the nonlinear system is not controllable, due to a quadratic obstruction involving the squared norm of the control's primitive. This obstruction has been known since 1983 for ODEs and observed for some PDEs since 2006. However, our situation is more intricate since the kernel describing the quadratic expansion of the solution is not twice differentiable. We thus follow a Fourier-based approach, closer to the one used for quadratic obstructions of fractional Sobolev regularity. In this Fourier-based approach, a challenge is to formulate a necessary and sufficient condition on the convolution kernel, for the quadratic form to be coercive. In previous studies, the coercivity was ensured by a signed asymptotic equivalent for the Fourier transform of the convolution kernel of the form $\widehat{K}(\omega) \sim \omega^{-2}$ as $|\omega| \to \infty$. In our case, $\widehat{K}$ is a distribution which has singularities and changes sign up to infinity. We still prove coercivity because one of the signs appears too infrequently.

Precision spectroscopic measurements in atoms and molecules play an increasingly important role in chemistry and physics, e.g., to characterize structure and dynamics at long timescales, to determine physical constants, or to search for physics beyond the standard model of particle physics. In this article, we demonstrate the combination of Zeeman deceleration and transverse laser cooling to generate slow (mean velocity of 175 m/s) and transversely ultracold ($T_\perp \approx 135 \,\mu$K) supersonic beams of metastable $(1s)(2s)\,^3S_1$ He (He$^*$) for precision spectroscopy. The curved-wavefront laser-cooling approach is used to achieve large capture velocities and high He$^*$ number densities. The beam properties are characterized by imaging, time-of-flight and high-resolution spectroscopic methods, and the factors limiting the Doppler widths in single-photon spectroscopic measurements of the $(1 s)(40 p) \,^3 P_J \, \leftarrow (1 s)(2 s) \,^3 S_1$ transition at UV frequencies around $1.15\times 10^{15}$ Hz are analyzed. In particular, the use of skimmers to geometrically confine the beam in the transverse directions is examined and shown to not always lead to a reduction of the Doppler width. Linewidths as narrow as 5 MHz could be obtained, enabling the determination of line centers with a precision of $\Delta \nu/\nu$ of $4\times 10^{-11}$ limited by the signal-to-noise ratio. Numerical particle-trajectory simulations are used to interpret the experimental observations and validate the conclusions.

Helium (He) is the ideal atom to perform tests of ab-initio calculations in two-electron systems that consider all known effects, including quantum-electrodynamics and nuclear-size contributions. Recent state-of-the-art calculations and measurements of energy intervals involving the He $2\;^3S_1$ metastable state reveal discrepancies at the level of $7\,\sigma$ that require clarification both from the experimental and theoretical sides. We report on a new determination, with unprecedented accuracy, of the ionization energy $E_\mathrm{I}\,(2\;^3S_1)$ of the $(1s)(2s)\;^3S_1$ metastable state of He. The measurements rely on a new approach combining interferometric laser-alignment control, SI-traceable frequency calibration and imaging-assisted Doppler-free spectroscopy. With this approach we record spectra of the $np$ Rydberg series in a highly-collimated cold supersonic beam of metastable He generated by a cryogenic valve and an electric discharge. Extrapolation of the Rydberg series yields a new value of the ionization energy ($E_\mathrm{I}\,(2\;^3S_1)/h= 1\,152\,842\,742.7082(55)_\mathrm{stat}(25)_\mathrm{sys}\,\mathrm{MHz}$) that deviates by $9\,\sigma$ from the most precise theoretical result ($1\,152\,842\,742.231(52)\;\mathrm{MHz}$), reported by Patk\'o\v{s}, Yerokhin and Pachucki [Phys. Rev. A. 103, 042809 (2021)], confirming earlier discrepancies between experiment and theory in this fundamental system.

Protein design with desirable properties has been a significant challenge for many decades. Generative artificial intelligence is a promising approach and has achieved great success in various protein generation tasks. Notably, diffusion models stand out for their robust mathematical foundations and impressive generative capabilities, offering unique advantages in certain applications such as protein design. In this review, we first give the definition and characteristics of diffusion models and then focus on two strategies: Denoising Diffusion Probabilistic Models and Score-based Generative Models, where DDPM is the discrete form of SGM. Furthermore, we discuss their applications in protein design, peptide generation, drug discovery, and protein-ligand interaction. Finally, we outline the future perspectives of diffusion models to advance autonomous protein design and engineering. The E(3) group consists of all rotations, reflections, and translations in three-dimensions. The equivariance on the E(3) group can keep the physical stability of the frame of each amino acid as much as possible, and we reflect on how to keep the diffusion model E(3) equivariant for protein generation.

We analyze the asymptotic pointwise behavior of families of solutions to the high-order critical equation $$P_\alpha u_\alpha=\Delta_g^k u_\alpha+\hbox{lot}=|u_\alpha|^{2^\star-2-\epsilon_\alpha} u_\alpha\hbox{ in }M$$ that behave like $$u_\alpha=u_0+B_\alpha+o(1)\hbox{ in }H_k^2(M)$$ where $B=(B_\alpha)_\alpha$ is a Bubble, also called a Peak. We give obstructions for such a concentration to occur: depending on the dimension, they involve the mass of the associated Green's function or the difference between $P_\alpha$ and the conformally invariant GJMS operator. The bulk of this analysis is the proof of the pointwise control \begin{equation*} |u_\alpha(x)|\leq C\Vert u_0\Vert_\infty^{(2^\star-1)^2}+C\left(\frac{\mu_\alpha^{2}}{\mu_\alpha^{2 }+d_g(x,x_\alpha)^{2 }}\right)^{\frac{n-2k}{2}}\hbox{ for all }x\in M\hbox{ and }\alpha\in\mathbb{N}, \end{equation*} where $|u_\alpha(x_\alpha)|=\max_M|u_\alpha|\to +\infty$ and $\mu_\alpha:=|u_\alpha(x_\alpha)|^{-\frac{2}{n-2k}}$. The key to obtain this estimate is a sharp control of the Green's function for elliptic operators involving a Hardy potential., Comment: Pour Mathilde et Daniel. arXiv admin note: text overlap with arXiv:2407.14893

We review observations of solar activity, geomagnetic variation, and auroral visibility for the extreme geomagnetic storm on 1872 February 4. The extreme storm (referred to here as the Chapman-Silverman storm) apparently originated from a complex active region of moderate area (\approx 500 {\mu}sh) that was favorably situated near disk center (S19{\deg} E05{\deg}). There is circumstantial evidence for an eruption from this region at 9--10 UT on 1872 February 3, based on the location, complexity, and evolution of the region, and on reports of prominence activations, which yields a plausible transit time of \approx29 hr to Earth. Magnetograms show that the storm began with a sudden commencement at \approx14:27 UT and allow a minimum Dst estimate of {\pounds} -834 nT. Overhead aurorae were credibly reported at Jacobabad (British India) and Shanghai (China), both at 19{\deg}.9 in magnetic latitude (MLAT) and 24{\deg}. 2 in invariant latitude (ILAT). Auroral visibility was reported from 13 locations with MLAT below |20|{\deg} for the 1872 storm (ranging from |10{\deg}. 0|--|19{\deg}. 9| MLAT) versus one each for the 1859 storm (|17{\deg}. 3| MLAT) and the 1921 storm (|16.{\deg}2| MLAT). The auroral extension and conservative storm intensity indicate a magnetic storm of comparable strength to the extreme storms of 1859 September (25{\deg}.1 \pm 0{\deg}.5 ILAT and -949 \pm 31 nT) and 1921 May (27{\deg}.1 ILAT and -907 \pm 132 nT), which places the 1872 storm among the three largest magnetic storms yet observed., Comment: 20 pages, 13 figures, 2 tables, published

When submerged in a chiral active bath, a passive object becomes a spinning ratchet imbued with odd transport properties. In the adiabatic limit of a massive object, we derive the most general Langevin dynamics for a rigid body in a chiral active bath, with odd diffusion and odd mobility connected by an Einstein relation, and numerically explore the breakdown of these predictions beyond the adiabatic limit. We show that the irreversibility of a massive object increases as its symmetry decreases: a disk exhibits an effective equilibrium dynamics, while a rod admits distinct translational and rotational temperatures, and a wedge is fully out of equilibrium. Conversely, this departure from equilibrium can be read in the universal far-field currents and density modulations of the bath, which we measure numerically and derive analytically.

Characterizing the local voltage distribution within nanophysiological domains, driven by ionic currents through membrane channels, is crucial for studying cellular activity in modern biophysics, yet it presents significant experimental and theoretical challenges. Theoretically, the complexity arises from the difficulty of solving electro-diffusion equations in three-dimensional domains. Currently, there are no methods available for obtaining asymptotic computations or approximated solutions of nonlinear equations, and numerically, it is challenging to explore solutions across both small and large spatial scales. In this work, we develop a method to solve the Poisson-Nernst-Planck equations with ionic currents entering and exiting through two narrow, circular window channels located on the boundary. The inflow through the first window is composed of a single cation, while the outflow maintains a constant ionic density satisfying local electro-neutrality conditions. Employing regular expansions and Green's function representations, we derive the ionic profiles and voltage drops in both small and large charge regimes. We explore how local surface curvature and window channels size influence voltage dynamics and validate our theoretical predictions through numerical simulations, assessing the accuracy of our asymptotic computations. These novel relationships between current, voltage, concentrations and geometry can enhance the characterization of physiological behaviors of nanodomains., Comment: 30 pages, 7 figures, 3 tables

We propose constructions of codes over quotient rings of Eisenstein integers equipped with the Euclidean, square Euclidean, and hexagonal distances as a generalization of codes over Eisenstein integer fields. By set partitioning, we effectively divide the ring of Eisenstein integers into equal-sized subsets for distinct encoding. Unlike in Eisenstein integer fields of prime size, where partitioning is not feasible due to structural limitations, we partition the quotient rings into additive subgroups in such a way that the minimum square Euclidean and hexagonal distances of each subgroup are strictly larger than in the original set. This technique facilitates multilevel coding and enhances signal constellation efficiency., Comment: 17 Pages, 4 figures

Building upon recent progress in Lindblad engineering for quantum Gibbs state preparation algorithms, we propose a simplified protocol that is shown to be efficient under the eigenstate thermalization hypothesis (ETH). The ETH reduces circuit overhead of the Lindblad simulation algorithm and ensures a fast convergence toward the target Gibbs state. Moreover, we show that the realized Lindblad dynamics exhibits resilience against depolarizing noise, opening up the path to a first demonstration on quantum computers. We complement our claims with numerical studies of the algorithm's convergence in various regimes of the mixed-field Ising model. In line with our predictions, we observe a mixing time scaling polynomially with system size when the ETH is satisfied. In addition, we assess the impact of algorithmic and hardware-induced errors on the algorithm's performance by carrying out quantum circuit simulations of our Lindblad simulation protocol with realistic noise models. This work bridges the gap between recent theoretical advances in Gibbs state preparation algorithms and their eventual quantum hardware implementation., Comment: 52 pages, 16 figures

Turn-taking prediction is the task of anticipating when the speaker in a conversation will yield their turn to another speaker to begin speaking. This project expands on existing strategies for turn-taking prediction by employing a multi-modal ensemble approach that integrates large language models (LLMs) and voice activity projection (VAP) models. By combining the linguistic capabilities of LLMs with the temporal precision of VAP models, we aim to improve the accuracy and efficiency of identifying TRPs in both scripted and unscripted conversational scenarios. Our methods are evaluated on the In-Conversation Corpus (ICC) and Coached Conversational Preference Elicitation (CCPE) datasets, highlighting the strengths and limitations of current models while proposing a potentially more robust framework for enhanced prediction.

In this companion paper for the DAGECC (Domain Adaptation and GEneralization for Character Classification) competition organized within the frame of the ICPR 2024 conference, we present the general context of the tasks we proposed to the community, we introduce the data that were prepared for the competition and we provide a summary of the results along with a description of the top three winning entries. The competition was centered around domain adaptation and generalization, and our core aim is to foster interest and facilitate advancement on these topics by providing a high-quality, lightweight, real world dataset able to support fast prototyping and validation of novel ideas., Comment: Companion paper for the ICPR 2024 Competition on Domain Adaptation and GEneralization for Character Classification (DAGECC)