Search

3D human motion generation has seen substantial advancement in recent years. While state-of-the-art approaches have improved performance significantly, they still struggle with complex and detailed motions unseen in training data, largely due to the scarcity of motion datasets and the prohibitive cost of generating new training examples. To address these challenges, we introduce CoMA, an agent-based solution for complex human motion generation, editing, and comprehension. CoMA leverages multiple collaborative agents powered by large language and vision models, alongside a mask transformer-based motion generator featuring body part-specific encoders and codebooks for fine-grained control. Our framework enables generation of both short and long motion sequences with detailed instructions, text-guided motion editing, and self-correction for improved quality. Evaluations on the HumanML3D dataset demonstrate competitive performance against state-of-the-art methods. Additionally, we create a set of context-rich, compositional, and long text prompts, where user studies show our method significantly outperforms existing approaches., Comment: Project Page: https://gabrie-l.github.io/coma-page/

The channel estimation problem has been widely discussed in traditional reconfigurable intelligent surface assisted multiple-input multiple-output. However, solutions for channel estimation adapted to beyond diagonal RIS need further study, and few recent works have been proposed to tackle this problem. Moreover, methods that avoid or minimize the use of pilot sequences are of interest. This work formulates a data-driven (semi-blind) joint channel and symbol estimation algorithm for beyond diagonal RIS that avoids a prior pilot-assisted stage while providing decoupled estimates of the involved communication channels. The proposed receiver builds upon a PARATUCK tensor model for the received signal, from which a trilinear alternating estimation scheme is derived. Preliminary numerical results demonstrate the proposed method's performance for selected system setups. The symbol error rate performance is also compared with that of a linear receiver operating with perfect knowledge of the cascaded channel.

This work investigates the application of the Newton's method for the numerical solution of a nonlinear boundary value problem formulated through an ordinary differential equation (ODE). Nonlinear ODEs arise in various mathematical modeling contexts, where an exact solution is often unfeasible due to the intrinsic complexity of these equations. Thus, a numerical approach is employed, using Newton's method to solve the system resulting from the discretization of the original problem. The procedure involves the iterative formulation of the method, which enables the approximation of solutions and the evaluation of convergence with respect to the problem parameters. The results demonstrate that Newton's method provides a robust and efficient solution, highlighting its applicability to complex boundary value problems and reinforcing its relevance for the numerical analysis of nonlinear systems. It is concluded that the methodology discussed is suitable for solving a wide range of boundary value problems, ensuring precision and stability in the results., Comment: 9 pages, 2 figures

The magnetic textures generated by a perpendicularly applied magnetic field at the ferromagnetic layer of $Co/Al_{2}O_{3}/Nb$ thin film heterostructures are investigated using magneto-optical imaging and micromagnetic simulations. It is observed that the stray field caused by flux avalanches in the superconducting layer prints out a non-trivial in-plane texture in the cobalt layer, which remains stable up to temperatures much above the Nb critical temperature. These textures mimic quite closely the dendritic thermomagnetic flux avalanches that penetrate the Nb layer from its edges. For low cobalt thickness, the filamentary magnetic textures occur in pairs with opposite polarity. The previous in-plane magnetization determines the relative location of the magnetic filaments with respect to the avalanche position. Micromagnetic simulations solving the Landau-Lifshitz-Gilbert equation confirm the interpretation given for the experimental findings., Comment: 13 paginas, 5 figuras, supplementary material

A triangular optical cavity can be used to distinguish between two beams with the same intensity profile but different wavefronts. This is what we show in this paper, both theoretically and experimentally, in the case of beams with a doughnut-like intensity profile: one of them having a helical wavefront (vortex beam with orbital angular momentum) and the other with no orbital angular momentum at all (which we call pseudovortex beam). We write the mode decomposition of such beams in the Hermite-Gaussian basis and in the the Laguerre-Gauss basis, respectively, and study how they interact with a triangular cavity in terms of their resonance peaks. The experimental results corroborate the theory, showing that each beam displays its own resonance pattern. Therefore, such cavity may be used to identify beams with orbital angular momentum, distinguishing them from pseudovortices., Comment: 5 pages, 3 figures

In a combined experimental and numerical effort we investigate airfoil tonal noise generation and reduction. The means of noise control are streak generators in form of cylindrical roughness elements. These elements are placed periodically along the span of airfoil at the mid chord streamwise position. Experiments are performed for a wide range of Reynolds number and angle of attack in a companion work (Alva et al. 2024). In the present work we concentrate on numerical investigations for a further investigation of selected cases. We have performed wall-resolved large-eddy simulations for a NACA 0012 airfoil at zero angle of attack and Mach 0.3. Two Reynolds numbers (80,000 and 100,000) have been investigated, showing acoustic results consistent with experiments at the same Reynolds but lower Mach numbers. Roughness elements attenuate tones in the acoustic field and, for the higher Reynolds number, suppress them. Through Fourier decomposition and spectral POD analysis of streamwise velocity data, dominating structures have been identified. Further, the coupling between structures generated by surface roughness and instability modes (Kelvin-Helmholtz) of shear layer has been identified through stability analysis, suggesting stabilisation mechanisms by which the sound generation by the airfoil is reduced by the roughness elements., Comment: 24 pages, 15 figures

In this paper, we investigate the age-of-information (AoI) of a power domain non-orthogonal multiple access (NOMA) network, where multiple internet-of-things (IoT) devices transmit to a common gateway in a grant-free random fashion. More specifically, we consider a framed setup composed of multiple time slots, and resort to the $Q$-learning algorithm to properly define, in a distributed manner, the time slot and the power level each IoT device transmits within a frame. In the proposed AoI-QL-NOMA scheme, the $Q$-learning reward is adapted with the aim of minimizing the average AoI of the network, while only requiring a single feedback bit per time slot, in a frame basis. Our results show that AoI-QL-NOMA significantly improves the AoI performance compared to some recently proposed schemes, without significantly reducing the network throughput.

We evaluate the performance of the LoRaWAN Long-Range Frequency Hopping Spread Spectrum (LR-FHSS) technique using a device-level probabilistic strategy for code rate and header replica allocation. Specifically, we investigate the effects of different header replica and code rate allocations at each end-device, guided by a probability distribution provided by the network server. As a benchmark, we compare the proposed strategy with the standardized LR-FHSS data rates DR8 and DR9. Our numerical results demonstrate that the proposed strategy consistently outperforms the DR8 and DR9 standard data rates across all considered scenarios. Notably, our findings reveal that the optimal distribution rarely includes data rate DR9, while data rate DR8 significantly contributes to the goodput and energy efficiency optimizations.

In this article, we introduce an original hybrid quantum-classical algorithm based on a variational quantum algorithm for solving systems of differential equations. The algorithm relies on a spectral method, which involves encoding the solution functions in the amplitudes of the quantum states generated by different parametrized circuits and transforms the task of solving the differential equations into an optimization problem. We first describe the principle of the algorithm from a theoretical point of view. We provide a detailed pseudo-code of the algorithm, on which we conduct a complexity analysis to highlight its scaling properties. We apply it to a set of examples, showcasing its applicability across diverse sets of differential equations. We discuss the advantages of our method and potential avenues for further exploration and refinement., Comment: The algorithm described in this paper is subject to a patent submission (EP24306601). Use of this method requires a license, which can be obtained from ColibriTD

In this paper, we study the existence of a solution for a class of Dirichlet problems with a singularity and a convection term. Precisely, we consider the existence of a positive solution to the Dirichlet problem $$-\Delta_p u = \frac{\lambda}{u^{\alpha}} + f(x,u,\nabla u)$$ in a bounded, smooth domain $\Omega$. The convection term has exponents with no upper limitations neither in $u$ nor in $\nabla u$. This is somewhat unexpected and rare. So, we address a wide range of problems not yet contained in the literature. The solution of the problem combines the definition of an auxiliary problem, the method of sub- and super-solution and Schauder's fixed point theorem., Comment: 17 pages

Algebraic lattices are those obtained from modules in the ring of integers of algebraic number fields through the canonical or twisted embeddings. In turn, well-rounded lattices are those with maximal cardinality of linearly independent vectors in its set of minimal vectors. Both classes of lattices have been applied for signal transmission in some channels, such as wiretap channels. Recently, some advances have been made in the search for well-rounded lattices that can be realized as algebraic lattices. Moreover, some works have been published studying algebraic lattices obtained from modules in cyclic number fields of odd prime degree $p$. In this work, we generalize some results of a recent work of Tran et al. and we provide new constructions of well-rounded algebraic lattices from a certain family of modules in the ring of integers of each of these fields when $p$ is ramified in its extension over the field of rational numbers., Comment: 11 pages

The accuracy of high precision and fundamental measurements of atomic transition frequencies via laser spectroscopy depends upon fitting the spectral data with a lineshape. With atomic hydrogen and antihydrogen 1S-2S two-photon spectroscopy, computer intensive Monte-Carlo simulations have been used to compute the optical Bloch equations in order to match and interpret the experimental spectra. For the highest resolutions, one tries to minimize saturation effects going to regimes of low excitation probability, where perturbation theory can provide reliable results. Here we describe an analytical approach to the lineshape based on perturbation theory accounting for the AC-Stark shift and ionization. The expressions can be used for beam experiments or integrated over the magnetic field profile for a trapped sample. Theses lineshapes, providing fast results, allow for studies of many systematic effects that influence the accuracy of the determination of the central frequency. This development has relevance to hydrogen beam experiments and to trapped hydrogen and antihydrogen, as developed by the ALPHA collaboration at CERN, for tests of the CPT-symmetry and the highest accuracy measurement on antimatter., Comment: 11 pages, 5 figures

Motivated by the idea of using simple macroscopic examples to illustrate the physics of complex systems, we modify a historic experimental setup in which interacting floating magnets spontaneously self-assemble into ordered clusters. By making the cluster components mechanically stable against their natural tendency to flip and coalesce, we can monitor the torque experienced by individual magnets through the macroscopic tilt angles they acquire when exposed to non-collinear external fields. A mathematical model that reproduces the empirical observations is introduced, enabling us to go beyond the experimental cases considered. The model confirms the existence of alternative orderings as the number of objects increases. Furthermore, a simpler and more mathematically transparent version of our model enables us to establish the conditions under which the cluster structure is stable and when it will collapse. We argue that our simple experimental setup combined with the accompanying models may be useful to describe general features seen in systems composed of mutually repulsive particles in the presence of modulating confining fields., Comment: 14 pages, 8 figures, submitted to PRE

Entangled graph states can be used for quantum sensing and computing applications. Error correction in measurement-based quantum computing schemes will require the construction of cluster states in at least 3 dimensions. Here we generate 1-, 2-, 3-, and 4-dimensional optical frequency-mode cluster states by sending broadband 2-mode vacuum-squeezed light through an electro-optical modulator (EOM) driven with multiple frequencies. We create the squeezed light using 4-wave mixing in Rb atomic vapor and mix the sideband frequencies (qumodes) using an EOM, as proposed by Zhu et al. (1), producing a pattern of entanglement correlations that constitute continuous-variable graph states containing up to several hundred qumodes. We verify the entanglement structure by using homodyne measurements to construct the covariance matrices and evaluate the nullifiers. This technique enables scaling of optical cluster states to multiple dimensions without increasing loss.

General Relativity (GR) is not the only way gravity can be geometrised. Instead of curvature, the Teleparallel Theory attributes gravity to torsion $T$, which is related to the antysimmetric part of connection, and the Symmetric Teleparallel theory no longer preserves metricity, describing gravity through the non-metricity tensor $Q_{\alpha\mu\nu}\equiv \nabla_\alpha g_{\mu\nu}.$ These descriptions give form to what is known as geometrical trinity of gravity. Recently, the extensions of GR have been intensively investigated in order to solve the theoretical impasses which have arisen. In this sense, it is also useful to investigate the extensions of alternative descriptions of gravity, which leads us to the so-called $f(T)$ and $f(Q)$ gravities. In this paper, we consider a family of $f(Q)$ models and obtain their corresponding Tolman-Oppenheimer-{Volkoff} equations applied to {polytropic} stars. Using numerical integration, it is possible to solve a system of differential equations and calculate, among other things, the maximum mass and mass-radius relation allowed. In addition, we explicitly show the non-metricity behavior inside and outside the star.

Let $\delta$ be a derivation in a $K$-algebra $R$ and let ${\rm Aut_{\delta}}(R)$ be the isotropy group with respect to the natural conjugation action of $\rm Aut(R)$ of $K$-automorphisms on the set $\rm Der(R)$ of $K$-derivations: that is, the subgroup of automorphisms that commute with the derivation. We explore the characterization of ${\rm Aut_{\delta}}(R)$ for quantum Weyl algebras and we prove that in the case of the Jordanian plane it is always a finite cyclic group. Furthermore, we obtain an arithmetic characterization that allows us to trivially demonstrate that these algebras are not isomorphic.

One way to personalize and steer generations from large language models (LLM) is to assign a persona: a role that describes how the user expects the LLM to behave (e.g., a helpful assistant, a teacher, a woman). This paper investigates how personas affect diverse aspects of model behavior. We assign to seven LLMs 162 personas from 12 categories spanning variables like gender, sexual orientation, and occupation. We prompt them to answer questions from five datasets covering objective (e.g., questions about math and history) and subjective tasks (e.g., questions about beliefs and values). We also compare persona's generations to two baseline settings: a control persona setting with 30 paraphrases of "a helpful assistant" to control for models' prompt sensitivity, and an empty persona setting where no persona is assigned. We find that for all models and datasets, personas show greater variability than the control setting and that some measures of persona behavior generalize across models., Comment: 20 pages, 12 figures

Artificial spin ices (ASI), containing magnetic monopole quasi-particles emerging at room temperature, have been investigated as a promising system to be applied in alternative low-power information technology devices. However, restrictions associated with the intrinsic energetic connections between opposing magnetic monopoles in conventional ASI need to be overcome to achieve this purpose. Here, photon-magnon scattering in nanomagnets is examined as an approach to locally activate the collective dynamics of interacting magnetic systems at the nanoscale. Low-power white and polarized light were employed as a new tool to manipulate magnetic monopole intensity, leading to tuning on the particles response to external magnetic field and spontaneous magnetization flipping without external field (thermodynamics). Our findings showing evidence of magnetic monopole plasma formation in a regular square ASI system are explained by an analytical model of photon-magnon conversion acting directly on the ASI nanomagnet dipole. Micromagnetic simulations based on the samples parameters and values obtained from the model present a very good qualitative correspondence between theory and observations for the investigated ASI system.

This study presents the results of a search for rotation signature in 250 Gaia DR3 Ultra-Cool Dwarfs (UCDs) with TESS light curves. We identified 71 targets with unambiguous periodicities, of which 61 present rotation signatures and a single source behavior, with periods between 0.133 and 5.81 days. Five UCDs show double-dip features, namely variations with two periods, one approximately double or half the other. The remaining ten UCDs with unambiguous variability present a likely non-single behavior. We also found 20 UCDs showing complex behavior in their light curves, with noticeable fluctuations and irregular structure, with a few exhibiting apparent changes in their temporal structure. The remaining 159 targets show noisy light curves corresponding to low-amplitude signals, whose temporal variation cannot be easily identified. The distribution of the UCDs with rotation signature in the CMD diagram points to a lack of rotating objects within about $11.5

Traditional image clustering techniques only find a single grouping within visual data. In particular, they do not provide a possibility to explicitly define multiple types of clustering. This work explores the potential of large vision-language models to facilitate alternative image clustering. We propose Text-Guided Alternative Image Consensus Clustering (TGAICC), a novel approach that leverages user-specified interests via prompts to guide the discovery of diverse clusterings. To achieve this, it generates a clustering for each prompt, groups them using hierarchical clustering, and then aggregates them using consensus clustering. TGAICC outperforms image- and text-based baselines on four alternative image clustering benchmark datasets. Furthermore, using count-based word statistics, we are able to obtain text-based explanations of the alternative clusterings. In conclusion, our research illustrates how contemporary large vision-language models can transform explanatory data analysis, enabling the generation of insightful, customizable, and diverse image clusterings.

Large language models (LLMs) have taken centre stage in debates on Artificial Intelligence. Yet there remains a gap in how to assess LLMs' conformity to important human values. In this paper, we investigate whether state-of-the-art LLMs, GPT-4 and Claude 2.1 (Gemini Pro and LLAMA 2 did not generate valid results) are moral hypocrites. We employ two research instruments based on the Moral Foundations Theory: (i) the Moral Foundations Questionnaire (MFQ), which investigates which values are considered morally relevant in abstract moral judgements; and (ii) the Moral Foundations Vignettes (MFVs), which evaluate moral cognition in concrete scenarios related to each moral foundation. We characterise conflicts in values between these different abstractions of moral evaluation as hypocrisy. We found that both models displayed reasonable consistency within each instrument compared to humans, but they displayed contradictory and hypocritical behaviour when we compared the abstract values present in the MFQ to the evaluation of concrete moral violations of the MFV., Comment: Final version available at: https://ojs.aaai.org/index.php/AIES/article/view/31704 13 pages, 4 figures, 2 tables

Training data memorization in language models impacts model capability (generalization) and safety (privacy risk). This paper focuses on analyzing prompts' impact on detecting the memorization of 6 masked language model-based named entity recognition models. Specifically, we employ a diverse set of 400 automatically generated prompts, and a pairwise dataset where each pair consists of one person's name from the training set and another name out of the set. A prompt completed with a person's name serves as input for getting the model's confidence in predicting this name. Finally, the prompt performance of detecting model memorization is quantified by the percentage of name pairs for which the model has higher confidence for the name from the training set. We show that the performance of different prompts varies by as much as 16 percentage points on the same model, and prompt engineering further increases the gap. Moreover, our experiments demonstrate that prompt performance is model-dependent but does generalize across different name sets. A comprehensive analysis indicates how prompt performance is influenced by prompt properties, contained tokens, and the model's self-attention weights on the prompt.

There has been a growing interest within the astrophysics community in highly magnetized and fast-spinning white dwarfs (WDs), commonly referred to as HMWDs. WDs with these characteristics are quite uncommon and possess magnetic fields $\geqslant 10^6$ G, along with short rotation periods ranging from seconds to just a few minutes. Based on our previous work, we analyze the emission of Gravitational Waves (GWs) in HMWDs through two mechanisms: matter accretion and magnetic deformation, which arise due to the asymmetry surrounding the star's rotational axis. Here, we perform a thorough self-consistent analysis, accounting for rotation and employing a realistic equation of state to investigate the stability of stars. Our investigation focuses on the emission of gravitational radiation from six rapidly spinning WDs: five of them are situated within binary systems, while one is an AXP, proposed as a magnetic accreting WD. Furthermore, we apply the matter accretion mechanism alongside the magnetic deformation mechanism to assess the influence of one process on the other. Our discoveries indicate that these WDs could potentially act as GW sources for BBO and DECIGO, depending on specific parameters, such as their mass, the angle ($\alpha$) between the magnetic and rotational axes, and the accumulated mass ($\delta m$) at their magnetic poles, which is influenced by the effect of matter accretion. However, detecting this particular class of stars using the LISA and TianQin space detectors seems unlikely due to the challenging combination of parameters such as a large $\delta m$, a large $\alpha$ angle and a small WD mass value., Comment: Accepted for publication in MNRAS

This study explores the use of large language models (LLMs) for translating English into Mambai, a low-resource Austronesian language spoken in Timor-Leste, with approximately 200,000 native speakers. Leveraging a novel corpus derived from a Mambai language manual and additional sentences translated by a native speaker, we examine the efficacy of few-shot LLM prompting for machine translation (MT) in this low-resource context. Our methodology involves the strategic selection of parallel sentences and dictionary entries for prompting, aiming to enhance translation accuracy, using open-source and proprietary LLMs (LlaMa 2 70b, Mixtral 8x7B, GPT-4). We find that including dictionary entries in prompts and a mix of sentences retrieved through TF-IDF and semantic embeddings significantly improves translation quality. However, our findings reveal stark disparities in translation performance across test sets, with BLEU scores reaching as high as 21.2 on materials from the language manual, in contrast to a maximum of 4.4 on a test set provided by a native speaker. These results underscore the importance of diverse and representative corpora in assessing MT for low-resource languages. Our research provides insights into few-shot LLM prompting for low-resource MT, and makes available an initial corpus for the Mambai language.

The Kerr spacetime is a fundamental solution of general relativity (GR), describing the gravitational field around a rotating, uncharged black hole (BH). Kerr spacetime has been crucial in modern astrophysics and it serves as a foundation for the study of gravitational waves (GWs). Possible deviations in Kerr geometry may indicate deviations from GR predictions. In this work, we consider the Johannsen-Psaltis metric, which is a beyond-Kerr metric characterized by a single free parameter, and then we probe this theory framework using several GWs observations from the third Gravitational-wave Transient Catalog (GWTC-3). We find that, for most of the events analyzed, there are no significant deviations from the null hypothesis, i.e. the Kerr metric. Our main findings demonstrate alignment and certain enhancements when compared to previous estimates documented in the literature., Comment: 8 pages, 4 figures

The study proposes a new efficient wireless biosensor based on magnetoelastic waves for the detection of antibodies in human plasma, aiming at the serological diagnosis of COVID-19. The biosensor was functionalized with the N antigen - nucleocapsid phosphoprotein of the SARS-CoV-2 virus. Validation analyses, by sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE), Western blotting, atomic force microscopy (AFM), scanning electron microscopy (SEM), micro-Raman spectroscopy, confirmed the selectivity and effective surface functionalization of the biosensor. The research successfully obtained, expressed and purified the recombinant antigen, while plasma samples from COVID-19 positive and negative patients were used to test the performance of the biosensor. A comparison of performance with the ELISA method revealed equivalent diagnostic power. These results indicate the robustness of the biosensor in reliably differentiating between positive and negative samples, highlighting its potential as an efficient and low-cost tool for the serological diagnosis of COVID-19. In addition to being fast to execute and having the potential for automation in large-scale diagnostic studies, the biosensor fills a significant gap in existing SARS-CoV-2 detection approaches., Comment: 17 pages, 1 table, and 7 figures

Machine learning (ML) and artificial intelligence (AI) approaches are often criticized for their inherent bias and for their lack of control, accountability, and transparency. Consequently, regulatory bodies struggle with containing this technology's potential negative side effects. High-level requirements such as fairness and robustness need to be formalized into concrete specification metrics, imperfect proxies that capture isolated aspects of the underlying requirements. Given possible trade-offs between different metrics and their vulnerability to over-optimization, integrating specification metrics in system development processes is not trivial. This paper defines specification overfitting, a scenario where systems focus excessively on specified metrics to the detriment of high-level requirements and task performance. We present an extensive literature survey to categorize how researchers propose, measure, and optimize specification metrics in several AI fields (e.g., natural language processing, computer vision, reinforcement learning). Using a keyword-based search on papers from major AI conferences and journals between 2018 and mid-2023, we identify and analyze 74 papers that propose or optimize specification metrics. We find that although most papers implicitly address specification overfitting (e.g., by reporting more than one specification metric), they rarely discuss which role specification metrics should play in system development or explicitly define the scope and assumptions behind metric formulations., Comment: 41 pages, 2 figures. Accepted at Artificial Intelligence Review