Search

In this paper we cover a few topics on how to treat inverse problems. There are two different flows of ideas. One approach is based on Morse Lemma. The other is based on analyticity which proves that the number of solutions to the inverse problems is generically isolated for some particular class of dynamical systems.

Researchers and practitioners interested in computational politics rely on automatic content analysis tools to make sense of the large amount of political texts available on the Web. Such tools should provide objective and subjective aspects at different granularity levels to make the analyses useful in practice. Existing methods produce interesting insights for objective aspects, but are limited for subjective ones, are often limited to national contexts, and have limited explainability. We introduce a text analysis framework which integrates both perspectives and provides a fine-grained processing of subjective aspects. Information retrieval techniques and knowledge bases complement powerful natural language processing components to allow a flexible aggregation of results at different granularity levels. Importantly, the proposed bottom-up approach facilitates the explainability of the obtained results. We illustrate its functioning with insights on news outlets, political orientations, topics, individual entities, and demographic segments. The approach is instantiated on a large corpus of French news, but is designed to work seamlessly for other languages and countries.

Data centers have become ubiquitous for today's businesses. From banks to startups, they rely on cloud infrastructure to deploy user applications. In this context, it is vital to provide users with application performance guarantees. Network interference is one of the causes of unpredictable application performance, and many solutions have been proposed over the years. The main objective of this survey is to familiarize the reader with research into network measurement-based resource allocation and control in data centers, focusing on network resources in order to provide cloud performance guarantees. We start with a primer on general network measurement techniques and data center network and applications to give the reader context. We then summarize the characteristics of network traffic and cluster workloads in data centers, which are pivotal for measurement-based allocation and control. We study and compare network monitoring in data centers, giving an overview on their evolution from Software-Defined Networking (SDN) to programmable dataplanes-based. The network monitoring information can serve as input to cluster allocation and scheduling decisions. We next categorize cluster scheduling frameworks, and perform an analysis of those that provide network guarantees in data centers, and we also look at emergent Machine Learning-driven resource allocation and control. We conclude with a discussion about future research directions.

When a particle moves in a Newtonian flow at low Reynolds number, inertia is irrelevant and a linear relationship exists between velocities and forces. For incompressible flows, any force distribution $\mathbf{f}(\mathbf{r})$ acting in the fluid bulk induces flow and motion only through its solenoidal component. For force distributions that are spatially localized (i.e., vanish sufficiently fast at infinity), we derive the representation of the rigid body motion as an explicit linear functional of $\nabla\times\mathbf{f}$, which complements the usual representation in terms of $\mathbf{f}$. We illustrate the utility of this alternative representation, which has the advantage of having the incompressibility constraint built-in, in avoiding certain ambiguities that arise, e.g., when implementing approximations for swimmers., Comment: The manuscript proper and the supplemental material have been merged consecutively into a single PDF file

The influence of space weathering on the observed spectra of C-complex asteroids remains uncertain. This has long hindered our understanding of their composition through telescope observations. Multi-band imaging of Ryugu by ONC-T on Hayabusa2 and that of Bennu by MapCam on OSIRIS-REx found opposite spectral trends of space weathering; Ryugu darkened/reddened while Bennu brightened/blued. How the spectra of Ryugu and Bennu evolved relative to each other would place a constraint for understanding their origins and evolutions. In this study, we compared the space weathering trends on Ryugu and Bennu by applying the results of cross calibration between ONC-T and MapCam. We show that the average Bennu surface is brighter by 18.0 $\pm$ 1.5% at 550 nm and bluer by 0.18 $\pm$ 0.03 $\mu$m$^{-1}$ (480-850 nm slope) than Ryugu. The spectral slopes of surface materials are more uniform on Bennu than on Ryugu at spatial scales $\gtrsim$1 m, but Bennu is more heterogeneous at $\lesssim$1 m. This suggests that lateral mixing due to resurfacing may have been more efficient on Bennu. The reflectance-spectral slope distributions of craters on Ryugu and Bennu appeared to follow two trend lines with an offset before cross calibration, but they converged to a single straight trend without a bend after cross calibration. We show that the spectra of the freshest craters on Ryugu and Bennu are indistinguishable within the uncertainty of cross calibration. These results suggest that Ryugu and Bennu initially had similar spectra before space weathering and that they evolved in completely opposite directions along the same trend line, subsequently evolving into asteroids with different disk-averaged spectra. These findings further suggest that space weathering likely expanded the spectral slope variation of C-complex asteroids, implying that they may have formed from materials with more uniform spectral slopes.

Reasoning with defeasible and conflicting knowledge in an argumentative form is a key research field in computational argumentation. Reasoning under various forms of uncertainty is both a key feature and a challenging barrier for automated argumentative reasoning. It was shown that argumentative reasoning using probabilities faces in general high computational complexity, in particular for the so-called constellation approach. In this paper, we develop an algorithmic approach to overcome this obstacle. We refine existing complexity results and show that two main reasoning tasks, that of computing the probability of a given set being an extension and an argument being acceptable, diverge in their complexity: the former is #P-complete and the latter is #-dot-NP-complete when considering their underlying counting problems. We present an algorithm for the complex task of computing the probability of a set of arguments being a complete extension by using dynamic programming operating on tree-decompositions. An experimental evaluation shows promise of our approach.

Self-chemophoresis is an appealing and quite successful interpretation of the motility exhibited by certain chemically active colloidal particles suspended in a solution of their "fuel": the particle has a phoretic response to self-generated, rather than externally imposed, inhomogeneities in the chemical composition of the solution. The postulated mechanism of chemophoresis is the interaction of the particle (via an adsorption potential) with the chemical inhomogeneities in the surrounding medium. When the range of this interaction is much smaller than any other relevant scale in the system, the (translational and rotational) phoretic velocities can be described in terms of a phoretic coefficient and a slip fluid velocity at the surface of the particle. Using the case of a spherical particle as a simple and physically insightful example, here we exploit an integral representation of the rigid-body motion to critically re-examine this framework. The systematic analysis highlights two steps in the approximation: first the thin-layer approximation (very large particle size), and subsequently a lubrication approximation (slow variation of the adsorption potential along the tangential direction). We also discuss how these approximations could be relaxed and the effect of this on the particle's motion.

With the recent advancements in quantum technologies, the QKD market exploded. World players are scrambling to win the race towards global QKD networks, even before the rules and policies required by such large endeavors were even discussed. Several vendors are on the market, each with specific parameters and advantages (in terms of key rate, link range, KMS software, etc.), hence considerable effort is now made towards standardization. While quantum communications is expected to reach a market size of up to \$36B by 2040, the largest QKD initiative to date is EuroQCI, which, due to its sheer scale, is forcing the market to mature. Although building a QKD network is believed to be trivial today, inter-connecting federated networks on a global scale is a heavy challenge. We propose QKD virtual networks not only as a useful infrastructure abstraction for increased flexibility and granular security, but as an inevitable solution for several problems that future QKD networks will encounter on the way towards widespread adoption.

In recent years, artificial intelligence (AI) has deeply impacted various fields, including Earth system sciences. Here, AI improved weather forecasting, model emulation, parameter estimation, and the prediction of extreme events. However, the latter comes with specific challenges, such as developing accurate predictors from noisy, heterogeneous and limited annotated data. This paper reviews how AI is being used to analyze extreme events (like floods, droughts, wildfires and heatwaves), highlighting the importance of creating accurate, transparent, and reliable AI models. We discuss the hurdles of dealing with limited data, integrating information in real-time, deploying models, and making them understandable, all crucial for gaining the trust of stakeholders and meeting regulatory needs. We provide an overview of how AI can help identify and explain extreme events more effectively, improving disaster response and communication. We emphasize the need for collaboration across different fields to create AI solutions that are practical, understandable, and trustworthy for analyzing and predicting extreme events. Such collaborative efforts aim to enhance disaster readiness and disaster risk reduction.

In recent years, Large Language Models (LLMs) have achieved almost human-like performance on various tasks. While some LLMs have been trained on multilingual data, most of the training data is in English; hence, their performance in English greatly exceeds other languages. To our knowledge, we are the first to collect and translate a large collection of texts, instructions, and benchmarks and train, evaluate, and release open-source LLMs tailored for Romanian. We evaluate our methods on four different categories, including academic benchmarks, MT-Bench (manually translated), and a professionally built historical, cultural, and social benchmark adapted to Romanian. We argue for the usefulness and high performance of RoLLMs by obtaining state-of-the-art results across the board. We publicly release all resources (i.e., data, training and evaluation code, models) to support and encourage research on Romanian LLMs while concurrently creating a generalizable recipe, adequate for other low or less-resourced languages., Comment: arXiv admin note: text overlap with arXiv:2405.07703

Automatic markerless estimation of infant posture and motion from ordinary videos carries great potential for movement studies "in the wild", facilitating understanding of motor development and massively increasing the chances of early diagnosis of disorders. There is rapid development of human pose estimation methods in computer vision thanks to advances in deep learning and machine learning. However, these methods are trained on datasets featuring adults in different contexts. This work tests and compares seven popular methods (AlphaPose, DeepLabCut/DeeperCut, Detectron2, HRNet, MediaPipe/BlazePose, OpenPose, and ViTPose) on videos of infants in supine position. Surprisingly, all methods except DeepLabCut and MediaPipe have competitive performance without additional finetuning, with ViTPose performing best. Next to standard performance metrics (object keypoint similarity, average precision and recall), we introduce errors expressed in the neck-mid-hip ratio and additionally study missed and redundant detections and the reliability of the internal confidence ratings of the different methods, which are relevant for downstream tasks. Among the networks with competitive performance, only AlphaPose could run close to real time (27 fps) on our machine. We provide documented Docker containers or instructions for all the methods we used, our analysis scripts, and processed data at https://hub.docker.com/u/humanoidsctu and https://osf.io/x465b/., Comment: 21 pages, 3 figures, 14 tables

Face recognition and verification are two computer vision tasks whose performance has progressed with the introduction of deep representations. However, ethical, legal, and technical challenges due to the sensitive character of face data and biases in real training datasets hinder their development. Generative AI addresses privacy by creating fictitious identities, but fairness problems persist. We promote fairness by introducing a demographic attributes balancing mechanism in generated training datasets. We experiment with an existing real dataset, three generated training datasets, and the balanced versions of a diffusion-based dataset. We propose a comprehensive evaluation that considers accuracy and fairness equally and includes a rigorous regression-based statistical analysis of attributes. The analysis shows that balancing reduces demographic unfairness. Also, a performance gap persists despite generation becoming more accurate with time. The proposed balancing method and comprehensive verification evaluation promote fairer and transparent face recognition and verification.

In \cite{FGHP}, the first author and his collaborators proved an equivariant Tamagawa number formula for the special value at $s=0$ of a Goss--type $L$--function, equivariant with respect to a Galois group $G$, and associated to a Drinfeld module defined on $\Bbb F_q[t]$ and over a finite, integral extension of $\Bbb F_q[t]$. The formula in question was proved provided that the values at $0$ of the Euler factors of the equivariant $L$--function in question satisfy certain identities involving Fitting ideals of certain $G$--cohomologically trivial, finite $\Bbb F_q[t][G]$--modules associated to the Drinfeld module. In \cite{FGHP}, we prove these identities in the particular case of the Carlitz module. In this paper, we develop general techniques and prove the identities in question for arbitrary Drinfeld modules. Further, we indicate how these techniques can be extended to the more general case of higher dimensional abelian $t$--modules, which is relevant in the context of the proof of the equivariant Tamagawa number formula for abelian $t$--modules given by N. Green and the first author in \cite{Green-Popescu}. This paper is based on a lecture given by the first author at ICMAT Madrid in May 2023 and builds upon results obtained by the second author in his PhD thesis \cite{Ramachandran-thesis}.

Many biological microswimmers are capable of chemotaxis, i.e., they can sense an ambient chemical gradient and adjust their mechanism of motility to move towards or away from the source of the gradient. Synthetic active colloids endowed with chemotactic behavior hold considerable promise for targeted drug delivery and the realization of programmable and reconfigurable materials. Here we study the chemotactic behavior of a Janus particle, which converts ``fuel'' molecules, released at an axisymmetric chemical patch located on a planar wall, into ``product'' molecules at its catalytic cap and moves by self-phoresis induced by the product. The chemotactic behavior is characterized as a function of the interplay between the rates of release (at the patch) and the consumption (at the particle) of fuel, as well as of details of the phoretic response of the particle (i.e., its phoretic mobility). Among others, we find that, under certain conditions, the particle is attracted to a stable ``hovering state'' in which it aligns its axis normal to the wall and rests (positions itself) at an activity-dependent distance above the center of the patch., Comment: 21 pages, 17 figures

To understand the formation and evolution of massive cosmic structures, studying them at high redshift, in the epoch when they formed the majority of their mass is essential. The One-hundred-deg$^2$ DECam Imaging in Narrowbands (ODIN) survey is undertaking the widest-area narrowband program to date, to use Ly$\alpha$-emitting galaxies (LAEs) to trace the large-scale structure (LSS) of the Universe at three cosmic epochs. In this work, we present results at $z$ = 3.1 based on early ODIN data in the COSMOS field. We identify and characterize protoclusters and cosmic filaments using multiple methods and discuss their strengths and weaknesses. We then compare our observations against the IllustrisTNG suite of cosmological hydrodynamical simulations. The two are in excellent agreement, with a similar number and angular size of structures identified above a specified density threshold. We are able to recover the simulated protoclusters with $\log$(M$_{z=0}$/$M_\odot$) $\gtrsim$ 14.4 in $\sim$ 60\% of the cases. With these objects we show that the descendant masses of the protoclusters in our sample can be estimated purely based on our 2D measurements, finding a median $z$ = 0 mass of $\sim10^{14.5}$M$_\odot$. The lack of information on the radial extent of each protocluster introduces a $\sim$0.4~dex uncertainty in its descendant mass. Finally, we show that the recovery of the cosmic web in the vicinity of protoclusters is both efficient and accurate. The similarity of our observations and the simulations imply that our structure selection is likewise robust and efficient, demonstrating that LAEs are reliable tracers of the LSS., Comment: 26 pages, 18 figures; submitted to ApJ

Satellite images have the potential to detect volcanic deformation prior to eruptions, but while a vast number of images are routinely acquired, only a small percentage contain volcanic deformation events. Manual inspection could miss these anomalies, and an automatic system modelled with supervised learning requires suitably labelled datasets. To tackle these issues, this paper explores the use of unsupervised deep learning on satellite data for the purpose of identifying volcanic deformation as anomalies. Our detector is based on Patch Distribution Modeling (PaDiM), and the detection performance is enhanced with a weighted distance, assigning greater importance to features from deeper layers. Additionally, we propose a preprocessing approach to handle noisy and incomplete data points. The final framework was tested with five volcanoes, which have different deformation characteristics and its performance was compared against the supervised learning method for volcanic deformation detection.

In recent years, Large Language Models (LLMs) have achieved almost human-like performance on various tasks. While some LLMs have been trained on multilingual data, most of the training data is in English. Hence, their performance in English greatly exceeds their performance in other languages. This document presents our approach to training and evaluating the first foundational and chat LLM specialized for Romanian.

In the paradigmatic example of quantum measurements, whenever one measures a system which starts in a superposition of two states of a conserved quantity, it jumps to one of the two states, implying different final values for the quantity that should have been conserved. The standard law of conservation for quantum mechanics handles this jump by stating only that the total distribution of the conserved quantity over repeated measurements is unchanged, but states nothing about individual cases. Here however we show that one can go beyond this and have conservation in each individual instance. We made our arguments in the case of angular momentum of a particle on a circle, where many technicalities simplify, and bring arguments to show that this holds in full generality. Hence we argue that the conservation law in quantum mechanics should be rewritten, to go beyond its hitherto statistical formulation, to state that the total of a conserved quantity is unchanged in every individual measurement outcome. As a further crucial element, we show that conservation can be localised at the level of the system of interest and its relevant frame of reference, and is independent on any assumptions on the distribution of the conserved quantity over the entire universe., Comment: 8 pages, 1 table

We re-examine the self-phoresis of a particle that releases(removes) pairs of ions into(from) the electrolyte solution. We show analytically that in the linear regime the mathematical description of this system maps onto that of the correlation-induced (self-)chemophoresis (CICP). This connection provides a unifying perspective of the two phenomena, within which one recovers and extends recent predictions as particular instances of CICP. Conversely, ion-phoretic particles are identified as candidates for experimental investigations into the rich variety of motility patterns predicted by CICP., Comment: The manuscript proper and the supplemental material have been merged consecutively into a single PDF file

In this paper we introduce and study new classes of mappings in metric spaces. The main class of mappings is called generalized orbital triangular contractions and it generalizes some existing results (such as Banach contractions, mappings contracting perimeters of triangles). We prove that these contractions are not necessarily continuous and have a unique fixed point under certain conditions. Moreover, we extend our class to generalized orbital triangular Kannan contractions and generalized orbital triangular Chatterjea contractions., Comment: Submitted to Journal on 24 April 2024

Electrically tunable optical devices present diverse functionalities for manipulating electromagnetic waves by leveraging elements capable of reversibly switching between different optical states. This adaptability in adjusting their responses to electromagnetic waves after fabrication is crucial for developing more efficient and compact optical systems for a broad range of applications including sensing, imaging, telecommunications, and data storage. Chalcogenide-based phase change materials (PCMs) have shown great promise due to their stable, non-volatile phase transition between amorphous and crystalline states. Nonetheless, optimizing the switching parameters of PCM devices and maintaining their stable operation over thousands of cycles with minimal variation can be challenging. In this paper, we report on the critical role of PCM pattern as well as electrical pulse form in achieving reliable and stable switching, extending the operational lifetime of the device beyond 13,000 switching events. To achieve this, we have developed a computer-aided algorithm that monitors optical changes in the device and adjusts the applied voltage in accordance with the phase transformation process, thereby significantly enhancing the lifetime of these reconfigurable devices. Our findings reveal that patterned PCM structures show significantly higher endurance compared to blanket PCM thin films.

This technical report introduces Pegasus-1, a multimodal language model specialized in video content understanding and interaction through natural language. Pegasus-1 is designed to address the unique challenges posed by video data, such as interpreting spatiotemporal information, to offer nuanced video content comprehension across various lengths. This technical report overviews Pegasus-1's architecture, training strategies, and its performance in benchmarks on video conversation, zero-shot video question answering, and video summarization. We also explore qualitative characteristics of Pegasus-1 , demonstrating its capabilities as well as its limitations, in order to provide readers a balanced view of its current state and its future direction.

The goal of this paper is to study the structure of noncommutative weighted shifts, their properties, and to understand their role as models (up to similarity) for $n$-tuples of operators on Hilbert spaces as well as their implications to function theory on noncommutative (resp.commutative) Reinhardt domains. We obtain a Rota type similarity result concerning the joint similarity of $n$-tuples of operators to parts of noncommutative weighted multi-shifts and provide a noncommutative multivariable analogue of Foias-Pearcy model for quasinilpotent operators. The model noncommutative weighted multi-shift which is studied in this paper is the $n$-tuple $W=(W_1,\ldots, W_n)$, where $W_i$ are weighted left creation operators of the full Fock space with $n$ generators associated with a weight sequence $\boldsymbol \mu=\{\mu_\beta\}_{|\beta|\geq 1}$ of nonnegative numbers. We also represent the injective weighted multi-shifts $W_1,\ldots, W_n$ as ordinary multiplications by $Z_1,\ldots, Z_n$ on a Hilbert space of noncommutative formal power series. This leads naturally to analytic function theory in several complex variables. One of the goal for the remainder of the paper is to analyze the extent to which our noncommutative formal power series represent analytic functions in several noncommutative (resp. commutative) variables and to develop a functional calculus for arbitrary $n$-tuples of operators on a Hilbert space., Comment: 49 pages

Let B(H) be the algebra of all bounded linear operators on a Hilbert space H. The main goal of the paper is to find large classes of noncommutative domains in B(H) with prescribed universal operator models, acting on the full Fock space with n generators, and to study these domains and their universal models in connection with the Hardy algebras and the C^*-algebras they generate. While the class of these domains contains the regular noncommutative domains previously studied in the literature, the main focus of the present paper is on the non-regular domains. The multi-variable operator theory of these domains is developed throughout the paper., Comment: 38 pages

In a recent paper, we introduced and studied the class of admissible noncommutative domains $D_{g^{-1}}(H)$ in $B(H)^n$ associated with admissible free holomorphic functions $g$ in noncommutative indeterminates $Z_1,\ldots, Z_n$. Each such a domain admits a universal model ${\bf W}:=(W_1,\ldots, W_n)$ of weighted left creation operators acting on the full Fock space with $n$ generators. In the present paper, we continue the study of these domains and their universal models in connection with the Hardy algebras and the $C^*$-algebras they generate. We obtain a Beurling type characterization of the invariant subspaces of the universal model ${\bf W}:=(W_1,\ldots, W_n)$ and develop a dilation theory for the elements of the noncommutative domain $D_{g^{-1}}(H)$. We also obtain results concerning the commutant lifting and Toeplitz-corrona in our setting as as well as some results on the boundary property for universal models., Comment: 29 pages

We introduce a new type of mappings in metric space which are three-point analogue of the well-known Chatterjea type mappings, and call them generalized Chatterjea type mappings. It is shown that such mappings can be discontinuous as is the case of Chatterjea type mappings and this new class includes the class of Chatterjea type mappings. The fixed point theorem for generalized Chatterjea type mappings is proven.

The aim of the current paper is to introduce a new class of contractive mappings, which are contracting (a feature of) triangles. We prove that maps contracting triangles are continuous and give the fixed point result for such mappings. We emphasize that our main theorem encompasses many functions, with significant applicability, for which the result holds, thereby representing a notable advancement in this research domain.

Class-incremental learning deals with sequential data streams composed of batches of classes. Various algorithms have been proposed to address the challenging case where samples from past classes cannot be stored. However, selecting an appropriate algorithm for a user-defined setting is an open problem, as the relative performance of these algorithms depends on the incremental settings. To solve this problem, we introduce an algorithm recommendation method that simulates the future data stream. Given an initial set of classes, it leverages generative models to simulate future classes from the same visual domain. We evaluate recent algorithms on the simulated stream and recommend the one which performs best in the user-defined incremental setting. We illustrate the effectiveness of our method on three large datasets using six algorithms and six incremental settings. Our method outperforms competitive baselines, and performance is close to that of an oracle choosing the best algorithm in each setting. This work contributes to facilitate the practical deployment of incremental learning.

Does the training of large language models potentially infringe upon code licenses? Furthermore, are there any datasets available that can be safely used for training these models without violating such licenses? In our study, we assess the current trends in the field and the importance of incorporating code into the training of large language models. Additionally, we examine publicly available datasets to see whether these models can be trained on them without the risk of legal issues in the future. To accomplish this, we compiled a list of 53 large language models trained on file-level code. We then extracted their datasets and analyzed how much they overlap with a dataset we created, consisting exclusively of strong copyleft code. Our analysis revealed that every dataset we examined contained license inconsistencies, despite being selected based on their associated repository licenses. We analyzed a total of 514 million code files, discovering 38 million exact duplicates present in our strong copyleft dataset. Additionally, we examined 171 million file-leading comments, identifying 16 million with strong copyleft licenses and another 11 million comments that discouraged copying without explicitly mentioning a license. Based on the findings of our study, which highlights the pervasive issue of license inconsistencies in large language models trained on code, our recommendation for both researchers and the community is to prioritize the development and adoption of best practices for dataset creation and management., Comment: Accepted to FORGE 2024

In this paper we study the role of radiative cooling in a two-fluid model consisting of coupled neutrals and charged particles. We first analyze the linearized two-fluid equations where we include radiative losses in the energy equation for the charged particles. In a 1D geometry for parallel propagation and in the limiting cases of weak and strong coupling, it can be shown analytically that the instability conditions for the thermal mode and the sound waves, the isobaric and isentropic criteria, respectively, remain unchanged with respect to one-fluid radiative plasmas. For the parameters considered in this paper, representative for the solar corona, the radiative cooling produces growth of the thermal mode and damping of the sound waves. When neutrals are included and are sufficiently coupled to the charges, the thermal mode growth rate and the wave damping both reduce by the same factor, which depends on the ionization fraction only. For a heating function which is constant in time, we find that the growth of the thermal mode and the damping of the sound waves are slightly larger. The numerical calculation of the eigenvalues of the general system of equations in a 3D geometry confirm the analytic results. We then run 2D fully nonlinear simulations which give consistent results: a higher ionization fraction or lower coupling will increase the growth rate. The magnetic field contribution is negligible in the linear phase. Ionization-recombination effects might play an important role because the radiative cooling produces a large range of temperatures in the system. In the numerical simulation, after the first condensation phase, when the minimum temperature is reached, the fraction of neutrals increases four orders of magnitude because of the recombination.

Exemplar-free class-incremental learning (EFCIL) poses significant challenges, primarily due to catastrophic forgetting, necessitating a delicate balance between stability and plasticity to accurately recognize both new and previous classes. Traditional EFCIL approaches typically skew towards either model plasticity through successive fine-tuning or stability by employing a fixed feature extractor beyond the initial incremental state. Building upon the foundational FeTrIL framework, our research extends into novel experimental domains to examine the efficacy of various oversampling techniques and dynamic optimization strategies across multiple challenging datasets and incremental settings. We specifically explore how oversampling impacts accuracy relative to feature availability and how different optimization methodologies, including dynamic recalibration and feature pool diversification, influence incremental learning outcomes. The results from these comprehensive experiments, conducted on CIFAR100, Tiny-ImageNet, and an ImageNet-Subset, under-score the superior performance of FeTrIL in balancing accuracy for both new and past classes against ten contemporary methods. Notably, our extensions reveal the nuanced impacts of oversampling and optimization on EFCIL, contributing to a more refined understanding of feature-space manipulation for class incremental learning. FeTrIL and its extended analysis in this paper FeTrIL++ pave the way for more adaptable and efficient EFCIL methodologies, promising significant improvements in handling catastrophic forgetting without the need for exemplars., Comment: arXiv admin note: text overlap with arXiv:2211.13131

Transformer-based language models for automatic code completion have shown great promise so far, yet the evaluation of these models rarely uses real data. This study provides both quantitative and qualitative assessments of three public code language models when completing real-world code. We first developed an open-source IDE extension, Code4Me, for the online evaluation of the models. We collected real auto-completion usage data for over a year from more than 1200 users, resulting in over 600K valid completions. These models were then evaluated using six standard metrics across twelve programming languages. Next, we conducted a qualitative study of 1690 real-world completion requests to identify the reasons behind the poor model performance. A comparative analysis of the models' performance in online and offline settings was also performed, using benchmark synthetic datasets and two masking strategies. Our findings suggest that while developers utilize code completion across various languages, the best results are achieved for mainstream languages such as Python and Java. InCoder outperformed the other models across all programming languages, highlighting the significance of training data and objectives. Our study also revealed that offline evaluations do not accurately reflect real-world scenarios. Upon qualitative analysis of the model's predictions, we found that 66.3% of failures were due to the models' limitations, 24.4% occurred due to inappropriate model usage in a development context, and 9.3% were valid requests that developers overwrote. Given these findings, we propose several strategies to overcome the current limitations. These include refining training objectives, improving resilience to typographical errors, adopting hybrid approaches, and enhancing implementations and usability., Comment: To be published in the proceedings of the 46th IEEE/ACM International Conference on Software Engineering (ICSE 2024)

The paper introduces Hands-Free VR, a voice-based natural-language interface for VR. The user gives a command using their voice, the speech audio data is converted to text using a speech-to-text deep learning model that is fine-tuned for robustness to word phonetic similarity and to spoken English accents, and the text is mapped to an executable VR command using a large language model that is robust to natural language diversity. Hands-Free VR was evaluated in a controlled within-subjects study (N = 22) that asked participants to find specific objects and to place them in various configurations. In the control condition participants used a conventional VR user interface to grab, carry, and position the objects using the handheld controllers. In the experimental condition participants used Hands-Free VR. The results confirm that: (1) Hands-Free VR is robust to spoken English accents, as for 20 of our participants English was not their first language, and to word phonetic similarity, correctly transcribing the voice command 96.71% of the time; (2) Hands-Free VR is robust to natural language diversity, correctly mapping the transcribed command to an executable command in 97.83% of the time; (3) Hands-Free VR had a significant efficiency advantage over the conventional VR interface in terms of task completion time, total viewpoint translation, total view direction rotation, and total left and right hand translations; (4) Hands-Free VR received high user preference ratings in terms of ease of use, intuitiveness, ergonomics, reliability, and desirability.

The discovery of interstellar comet 2I/Borisov offered the unique opportunity to obtain a detailed analysis of an object coming from another planetary system, and leaving behind material in our interplanetary space. We continuously observed 2I/Borisov between October 3 and December 13, 2019 using the 1.52-m Telescopio Carlos S\'{a}nchez equipped with MuSCAT2 instrument, and the 2.54-m Isaac Newton Telescope with Wide Field Camera. We characterize its morphology and spectro-photometric features using the data gathered during this extended campaign. Simultaneous imaging in four bands ($g$, $r$, $i$, and $z_s$) reveals a homogeneous composition and a reddish hue, resembling Solar System comets, and as well a diffuse profile exhibiting familiar cometary traits. We discern a stationary trend fluctuating around a constant activity level throughout October and November 2019. Subsequently, a reduction in activity is observed in December. Dust production and mass loss calculations indicate approximately an average of 4 kg/s before perihelion, while after perihelion the net mass loss is about 0.6 kg/s. Our simulations indicate the most probable size of coma dust particles should be in the range 200-250 nm, and the terminal speed around 300 m/s. The spectrum acquired with the 4.2-m William Herschel Telescope shows the presence of a strong CN line for which we find a gas production rate of $1.2 \times 10^{24}~s^{-1}$. We also detected NH$_2$ and OI bands. The ratio between NH$_2$ and CN productions is $\log (NH_2/CN) =-0.2$. Overall, this observing campaign provides a new understanding of 2I/Borisov's unique characteristics and activity patterns., Comment: accepted to MNRAS on 12th Feb 2024

In a recent paper, PNAS, 118, e1921529118 (2021), it was argued that while the standard definition of conservation laws in quantum mechanics, which is of a statistical character, is perfectly valid, it misses essential features of nature and it can and must be revisited to address the issue of conservation/non-conservation in individual cases. Specifically, in the above paper an experiment was presented in which it can be proven that in some individual cases energy is not conserved, despite being conserved statistically. It was felt however that this is worrisome, and that something must be wrong if there are individual instances in which conservation doesn't hold, even though this is not required by the standard conservation law. Here we revisit that experiment and show that although its results are correct, there is a way to circumvent them and ensure individual case conservation in that situation. The solution is however quite unusual, challenging one of the basic assumptions of quantum mechanics, namely that any quantum state can be prepared, and it involves a time-holistic, double non-conservation effect. Our results bring new light on the role of the preparation stage of the initial state of a particle and on the interplay of conservation laws and frames of reference. We also conjecture that when such a full analysis of any conservation experiment is performed, conservation is obeyed in every individual case.

The high-pressure crystal structure evolution of CH3NH3PbBr3 (MAPbBr3) perovskite has been investigated by single-crystal X-ray diffraction and synchrotron-based powder X-ray diffraction. Single-crystal X-ray diffraction reveals that the crystal structure of MAPbBr3 undergoes two phase transitions following the space-group sequence: Pm-3m-to-Im3-to-Pmn21. The transitions take place at around 0.8 and 1.8 GPa, respectively. This result is contradicting the previously reported phase transition sequence: Pm-3m-to-Im3-to-Pnma. In this work the crystal structures of each of the three phases are determined from single-crystal X-ray diffraction analysis which is later supported by Rietveld refinement of powder X-ray diffraction patterns. The pressure dependence of the crystal lattice parameters and unit-cell volumes are determined from the two aforementioned techniques, as well as the bulk moduli for each phase. The bandgap behaviour of MAPbBr3 has been studied up to around 4 GPa, by the means of single-crystal optical-absorption experiments. The evolution of the bandgap has been well explained using the pressure dependence of the Pb-Br bond distance and Pb-Br-Pb angles as determined from single-crystal X-ray diffraction experiments., Comment: 20 pages, 5 pages

We report a characterization of the high-pressure behavior of zinc-iodate, Zn(IO3)2. By the combination of x-ray diffraction, Raman spectroscopy, and first-principles calculations we have found evidence of two subtle isosymmetric structural phase transitions. We present arguments relating these transitions to a non-linear behavior of phonons and changes induced by pressure on the coordination sphere of the iodine atoms. This fact is explained as a consequence of the formation of metavalent bonding at high-pressure which is favored by the lone-electron pairs of iodine. In addition, the pressure dependence of unit-cell parameters, volume, and bond is reported. An equation of state to describe the pressure dependence of the volume is presented, indicating that Zn(IO3)2 is the most compressible iodate among those studied up to now. Finally, phonon frequencies are reported together with their symmetry assignment and pressure dependence., Comment: 33 pages, 3 tables, 15 figures

We report a study of the high-pressure structural behavior of ZnV2O6, Zn2V2O7, and Zn3V2O8, which has been explored by means of synchrotron powder x-ray diffraction. We found that ZnV2O6 and Zn3V2O8 remain in the ambient-pressure structure up to 15 GPa. In contrast, in the same pressure range, Zn2V2O7 undergoes three phase transitions at 0.7, 3.0, and 10.8 GPa, respectively. Possible crystal structures for the first and second high-pressure phases are proposed. Reasons for the distinctive behavior of Zn2V2O7 are discussed. The compressibility of the different polymorphs has been determined. The response to pressure is found to be anisotropic in all the considered compounds and the room-temperature equations of state have been determined. The bulk moduli of ZnV2O6 (129(2) GPa) and Zn3V2O8 (120(2) GPa) are consistent with a structural framework composed of compressible ZnO6 octahedra and uncompressible VO4 tetrahedra. In contrast, Zn2V2O7 is highly compressible with a bulk modulus of 58(9) GPa, which is almost half of the bulk modulus of the other two vanadates. The large compressibility of Zn2V2O7 and its sequence of structural transitions are related to the fact that this material is less dense than the other zinc vanadates and to the penta-coordination of Zn atoms by oxygen atoms in Zn2V2O7. A comparison to the high-pressure behavior of related compounds is presented., Comment: 35 pages, 15 figures, 2 tables

Neural machine translation systems estimate probabilities of target sentences given source sentences, yet these estimates may not align with human preferences. This work introduces QE-fusion, a method that synthesizes translations using a quality estimation metric (QE), which correlates better with human judgments. QE-fusion leverages a pool of candidates sampled from a model, combining spans from different candidates using a QE metric such as CometKiwi. We compare QE-fusion against beam search and recent reranking techniques, such as Minimum Bayes Risk decoding or QE-reranking. Our method consistently improves translation quality in terms of COMET and BLEURT scores when applied to large language models (LLMs) used for translation (PolyLM, XGLM, Llama2, Mistral, ALMA, and Tower) and to multilingual translation models (NLLB), over five language pairs. Notably, QE-fusion exhibits larger improvements for LLMs due to their ability to generate diverse outputs. We demonstrate that our approach generates novel translations in over half of the cases and consistently outperforms other methods across varying numbers of candidates (5-200). Furthermore, we empirically establish that QE-fusion scales linearly with the number of candidates in the pool., Comment: Accepted at ACL 2024

The Extreme Universe Space Observatory on a Super Pressure Balloon 1 (EUSO-SPB1) was launched in 2017 April from Wanaka, New Zealand. The plan of this mission of opportunity on a NASA super pressure balloon test flight was to circle the southern hemisphere. The primary scientific goal was to make the first observations of ultra-high-energy cosmic-ray extensive air showers (EASs) by looking down on the atmosphere with an ultraviolet (UV) fluorescence telescope from suborbital altitude (33~km). After 12~days and 4~hours aloft, the flight was terminated prematurely in the Pacific Ocean. Before the flight, the instrument was tested extensively in the West Desert of Utah, USA, with UV point sources and lasers. The test results indicated that the instrument had sensitivity to EASs of approximately 3 EeV. Simulations of the telescope system, telescope on time, and realized flight trajectory predicted an observation of about 1 event assuming clear sky conditions. The effects of high clouds were estimated to reduce this value by approximately a factor of 2. A manual search and a machine-learning-based search did not find any EAS signals in these data. Here we review the EUSO-SPB1 instrument and flight and the EAS search., Comment: 18 pages, 19 figures

This study explores non-linear development of the magnetized Rayleigh-Taylor instability (RTI) in a prominence-corona transition region. Using a two-fluid model of a partially ionized plasma, we compare RTI simulations for several different magnetic field configurations. We follow prior descriptions of the numerical prominence model [Popescu Braileanu et al., 2021a,b, 2023] and explore the charged-neutral fluid coupling and plasma heating in a two-dimensional mixing layer for different magnetic field configurations. We also investigate how the shear in magnetic field surrounding a prominence may impact the release of the gravitational potential energy of the prominence material. We show that the flow decoupling is strongest in the plane normal to the direction of the magnetic field, where neutral pressure gradients drive ion-neutral drifts and frictional heating is balanced by adiabatic cooling of the expanding prominence material. We also show that magnetic field within the mixing plane can lead to faster non-linear release of the gravitational energy driving the RTI, while more efficiently heating the plasma via viscous dissipation of associated plasma flows. We relate the computational results to potential observables by highlighting how integrating over under-resolved two-fluid sub-structure may lead to misinterpretation of observational data., Comment: accepted to Philosophical Transactions A

Determining the size distribution of asteroids is key for understanding the collisional history and evolution of the inner Solar System. We aim at improving our knowledge on the size distribution of small asteroids in the Main Belt by determining the parallaxes of newly detected asteroids in the Hubble Space Telescope (HST) Archive and hence their absolute magnitudes and sizes. Asteroids appear as curved trails in HST images due to the parallax induced by the fast orbital motion of the spacecraft. The parallax effect can be computed to obtain the distance to the asteroids by fitting simulated trajectories to the observed trails. Using distance, we can obtain the object's absolute magnitude and size estimation assuming an albedo value, along with some boundaries for its orbital parameters. In this work we analyse a set of 632 serendipitously imaged asteroids found in the ESA HST Archive. An object-detection machine learning algorithm was used to perform this task during previous work. Our raw data consists of 1,031 asteroids trails from unknown objects (not matching any entries in the MPC database). We also found 670 trails from known objects (objects featuring matching entries in the MPC). After an accuracy assessment and filtering process, our analysed HST set consists of 454 unknown objects and 178 known objects. We obtain a sample dominated by potential Main Belt objects featuring absolute magnitudes (H) mostly between 15 and 22 mag. The absolute magnitude cumulative distribution confirms the previously reported slope change for 15 < H < 18, from 0.56 to 0.26, maintained in our case down to absolute magnitudes around H = 20, hence expanding the previous results by approximately two magnitudes. HST archival observations can be used as an asteroid survey since the telescope pointings are statistically randomly oriented in the sky and they cover long periods of time., Comment: Accepted for publication in A&A. 10 pages, 2 tables, 17 figures