Showing 1–50 of 131 results for author: Carvalho, D

An LLM-based Knowledge Synthesis and Scientific Reasoning Framework for Biomedical Discovery

Authors: Oskar Wysocki, Magdalena Wysocka, Danilo Carvalho, Alex Teodor Bogatu, Danilo Miranda Gusicuma, Maxime Delmas, Harriet Unsworth, Andre Freitas

Abstract: We present BioLunar, developed using the Lunar framework, as a tool for supporting biological analyses, with a particular emphasis on molecular-level evidence enrichment for biomarker discovery in oncology. The platform integrates Large Language Models (LLMs) to facilitate complex scientific reasoning across distributed evidence spaces, enhancing the capability for harmonizing and reasoning over h… ▽ More We present BioLunar, developed using the Lunar framework, as a tool for supporting biological analyses, with a particular emphasis on molecular-level evidence enrichment for biomarker discovery in oncology. The platform integrates Large Language Models (LLMs) to facilitate complex scientific reasoning across distributed evidence spaces, enhancing the capability for harmonizing and reasoning over heterogeneous data sources. Demonstrating its utility in cancer research, BioLunar leverages modular design, reusable data access and data analysis components, and a low-code user interface, enabling researchers of all programming levels to construct LLM-enabled scientific workflows. By facilitating automatic scientific discovery and inference from heterogeneous evidence, BioLunar exemplifies the potential of the integration between LLMs, specialised databases and biomedical tools to support expert-level knowledge synthesis and discovery. △ Less

Submitted 26 June, 2024; originally announced June 2024.

Comments: accepted for ACL 2024 System Demonstration Track

Penetration of a spinning sphere impacting a granular medium

Authors: D. D. Carvalho, Y. Bertho, E. M. Franklin, A. Seguin

Abstract: We investigate experimentally the influence of rotation on the penetration depth of a spherical projectile impacting a granular medium. We show that a rotational motion significantly increases the penetration depth achieved. Moreover, we model our experimental results by modifying the frictional term of the equation describing the penetration dynamics of an object in a granular medium. In particul… ▽ More We investigate experimentally the influence of rotation on the penetration depth of a spherical projectile impacting a granular medium. We show that a rotational motion significantly increases the penetration depth achieved. Moreover, we model our experimental results by modifying the frictional term of the equation describing the penetration dynamics of an object in a granular medium. In particular, we find that the frictional drag decreases linearly with the velocity ratio between rotational (spin motion) and translational (falling motion) velocities. The good agreement between our model and our experimental measurements offers perspectives for estimating the depth that spinning projectiles reach after impacting onto a granular ground, such as happens with seeds dropped from aircraft or with landing probes. △ Less

Submitted 8 May, 2024; originally announced May 2024.

Journal ref: Phys. Rev. E 109, 054902 (2024)

Effects of higher-order Casimir-Polder interactions on Rydberg atom spectroscopy

Authors: Biplab Dutta, Joao Carlos de Aquino Carvalho, Guadalupe Garcia-Arellano, Paolo Pedri, Athanasios Laliotis, Chris Boldt, Jivesh Kaushal, Stefan Scheel

Abstract: In the extreme near-field, when the spatial extension of the atomic wavefunction is no longer negligible compared to the atom-surface distance, the dipole approximation is no longer sufficient to describe Casimir-Polder interactions. Here we calculate the higher-order, quadrupole and octupole, contributions to Casimir-Polder energy shifts of Rydberg atoms close to a dielectric surface. We subseque… ▽ More In the extreme near-field, when the spatial extension of the atomic wavefunction is no longer negligible compared to the atom-surface distance, the dipole approximation is no longer sufficient to describe Casimir-Polder interactions. Here we calculate the higher-order, quadrupole and octupole, contributions to Casimir-Polder energy shifts of Rydberg atoms close to a dielectric surface. We subsequently investigate the effects of these higher-order terms in thin-cell and selective reflection spectroscopy. Beyond its fundamental interest, this new regime of extremely small atom surface separations is relevant for quantum technology applications with Rydberg or surface-bound atoms interfacing with photonic platforms. △ Less

Submitted 20 April, 2024; originally announced April 2024.

Comments: to appear in Physical Review Research

Control of the Schrödinger equation in $\mathbb{R}^3$: The critical case

Authors: Pablo Braz e Silva, Roberto de A. Capistrano-Filho, Jackellyny Dassy do Nascimento Carvalho, David dos Santos Ferreira

Abstract: This article deals with the $\dot{H}^{1}$--level exact controllability for the defocusing critical nonlinear Schrödinger equation in $\mathbb{R}^3$. Firstly, we show the problem under consideration to be well-posed using Strichartz estimates. Moreover, through the Hilbert uniqueness method, we prove the linear Schrödinger equation to be controllable. Finally, we use a perturbation argument and sho… ▽ More This article deals with the $\dot{H}^{1}$--level exact controllability for the defocusing critical nonlinear Schrödinger equation in $\mathbb{R}^3$. Firstly, we show the problem under consideration to be well-posed using Strichartz estimates. Moreover, through the Hilbert uniqueness method, we prove the linear Schrödinger equation to be controllable. Finally, we use a perturbation argument and show local exact controllability for the critical nonlinear Schrödinger equation. △ Less

Submitted 11 April, 2024; originally announced April 2024.

Comments: 20 pages. Comments are welcome

Probing molecules in gas cells of subwavelength thickness with high frequency resolution

Authors: Guadalupe Garcia Arellano, Joao Carlos de Aquino Carvalho, Hippolyte Mouhanna, Esther Butery, Thierry Billeton, Frederic Du-Burck, Benoît Darquié, Isabelle Maurin, Athanasios Laliotis

Abstract: Miniaturizing and integrating atomic vapor cells is widely investigated for the purposes of fundamental measurements and technological applications such as quantum sensing. Extending such platforms to the realm of molecular physics is a fascinating prospect that paves the way for compact frequency metrology as well as for exploring light-matter interactions with complex quantum objects. Here, we p… ▽ More Miniaturizing and integrating atomic vapor cells is widely investigated for the purposes of fundamental measurements and technological applications such as quantum sensing. Extending such platforms to the realm of molecular physics is a fascinating prospect that paves the way for compact frequency metrology as well as for exploring light-matter interactions with complex quantum objects. Here, we perform molecular rovibrational spectroscopy in a thin-cell of micrometric thickness, comparable to excitation wavelengths. We operate the cell in two distinct regions of the electromagnetic spectrum, probing $ν_1$+$ν_3$ resonances of acetylene at 1.530$μ$m, within the telecommunications wavelength range, as well as the $ν_3$ and $ν_2$ resonances of $SF_6$ and $NH_3$ respectively, in the mid-infrared fingerprint region around 10.55$μ$m. Thin-cell confinement allows linear sub-Doppler transmission spectroscopy due to the coherent Dicke narrowing effect, here demonstrated for molecular rovibrations. Our experiment can find applications extending to the fields of compact molecular frequency references, atmospheric physics or fundamental precision measurements. △ Less

Submitted 6 March, 2024; originally announced March 2024.

Comments: published in Nature Communications

Journal ref: Nature Communications | (2024) 15:1862

Improving Semantic Control in Discrete Latent Spaces with Transformer Quantized Variational Autoencoders

Authors: Yingji Zhang, Danilo S. Carvalho, Marco Valentino, Ian Pratt-Hartmann, Andre Freitas

Abstract: Achieving precise semantic control over the latent spaces of Variational AutoEncoders (VAEs) holds significant value for downstream tasks in NLP as the underlying generative mechanisms could be better localised, explained and improved upon. Recent research, however, has struggled to achieve consistent results, primarily due to the inevitable loss of semantic information in the variational bottlene… ▽ More Achieving precise semantic control over the latent spaces of Variational AutoEncoders (VAEs) holds significant value for downstream tasks in NLP as the underlying generative mechanisms could be better localised, explained and improved upon. Recent research, however, has struggled to achieve consistent results, primarily due to the inevitable loss of semantic information in the variational bottleneck and limited control over the decoding mechanism. To overcome these challenges, we investigate discrete latent spaces in Vector Quantized Variational AutoEncoders (VQVAEs) to improve semantic control and generation in Transformer-based VAEs. In particular, We propose T5VQVAE, a novel model that leverages the controllability of VQVAEs to guide the self-attention mechanism in T5 at the token-level, exploiting its full generalization capabilities. Experimental results indicate that T5VQVAE outperforms existing state-of-the-art VAE models, including Optimus, in terms of controllability and preservation of semantic information across different tasks such as auto-encoding of sentences and mathematical expressions, text transfer, and inference. Moreover, T5VQVAE exhibits improved inference capabilities, suggesting potential applications for downstream natural language and symbolic reasoning tasks. △ Less

Submitted 1 February, 2024; originally announced February 2024.

Intelligent Data-Driven Architectural Features Orchestration for Network Slicing

Authors: Rodrigo Moreira, Flavio de Oliveira Silva, Tereza Cristina Melo de Brito Carvalho, Joberto S. B. Martins

Abstract: Network slicing is a crucial enabler and a trend for the Next Generation Mobile Network (NGMN) and various other new systems like the Internet of Vehicles (IoV) and Industrial IoT (IIoT). Orchestration and machine learning are key elements with a crucial role in the network-slicing processes since the NS process needs to orchestrate resources and functionalities, and machine learning can potential… ▽ More Network slicing is a crucial enabler and a trend for the Next Generation Mobile Network (NGMN) and various other new systems like the Internet of Vehicles (IoV) and Industrial IoT (IIoT). Orchestration and machine learning are key elements with a crucial role in the network-slicing processes since the NS process needs to orchestrate resources and functionalities, and machine learning can potentially optimize the orchestration process. However, existing network-slicing architectures lack the ability to define intelligent approaches to orchestrate features and resources in the slicing process. This paper discusses machine learning-based orchestration of features and capabilities in network slicing architectures. Initially, the slice resource orchestration and allocation in the slicing planning, configuration, commissioning, and operation phases are analyzed. In sequence, we highlight the need for optimized architectural feature orchestration and recommend using ML-embed agents, federated learning intrinsic mechanisms for knowledge acquisition, and a data-driven approach embedded in the network slicing architecture. We further develop an architectural features orchestration case embedded in the SFI2 network slicing architecture. An attack prevention security mechanism is developed for the SFI2 architecture using distributed embedded and cooperating ML agents. The case presented illustrates the architectural feature's orchestration process and benefits, highlighting its importance for the network slicing process. △ Less

Submitted 12 January, 2024; originally announced January 2024.

Comments: 12 pages, 6 figures, Conference ADVANCE 24 - International Workshop on ADVANCEs in ICT Infrastructures and Services - February 26--29, 2024 - Hanoi, Vietnam

ACM Class: I.2.11; C.2.1; I.2.1

LlaMaVAE: Guiding Large Language Model Generation via Continuous Latent Sentence Spaces

Authors: Yingji Zhang, Danilo S. Carvalho, Ian Pratt-Hartmann, André Freitas

Abstract: Deep generative neural networks, such as Variational AutoEncoders (VAEs), offer an opportunity to better understand and control language models from the perspective of sentence-level latent spaces. To combine the controllability of VAE latent spaces with the state-of-the-art performance of recent large language models (LLMs), we present in this work LlaMaVAE, which combines expressive encoder and… ▽ More Deep generative neural networks, such as Variational AutoEncoders (VAEs), offer an opportunity to better understand and control language models from the perspective of sentence-level latent spaces. To combine the controllability of VAE latent spaces with the state-of-the-art performance of recent large language models (LLMs), we present in this work LlaMaVAE, which combines expressive encoder and decoder models (sentenceT5 and LlaMA) with a VAE architecture, aiming to provide better text generation control to LLMs. In addition, to conditionally guide the VAE generation, we investigate a new approach based on flow-based invertible neural networks (INNs) named Invertible CVAE. Experimental results reveal that LlaMaVAE can outperform the previous state-of-the-art VAE language model, Optimus, across various tasks, including language modelling, semantic textual similarity and definition modelling. Qualitative analysis on interpolation and traversal experiments also indicates an increased degree of semantic clustering and geometric consistency, which enables better generation control. △ Less

Submitted 20 December, 2023; originally announced December 2023.

Impact craters formed by spinning granular projectiles

Authors: Douglas Daniel de Carvalho, Nicolao Cerqueira Lima, Erick de Moraes Franklin

Abstract: Craters formed by the impact of agglomerated materials are commonly observed in nature, such as asteroids colliding with planets and moons. In this paper, we investigate how the projectile spin and cohesion lead to different crater shapes. For that, we carried out DEM (discrete element method) computations of spinning granular projectiles impacting onto cohesionless grains, for different bonding s… ▽ More Craters formed by the impact of agglomerated materials are commonly observed in nature, such as asteroids colliding with planets and moons. In this paper, we investigate how the projectile spin and cohesion lead to different crater shapes. For that, we carried out DEM (discrete element method) computations of spinning granular projectiles impacting onto cohesionless grains, for different bonding stresses, initial spins and initial heights. We found that, as the bonding stresses decrease and the initial spin increases, the projectile's grains spread farther from the collision point, and, in consequence, the crater shape becomes flatter, with peaks around the rim and in the center of craters. Our results shed light on the dispersion of the projectile's material and the different shapes of craters found on Earth and other planetary environments. △ Less

Submitted 25 November, 2023; originally announced November 2023.

Comments: Accepted manuscript for Physical Review E, 108, 054904, (2023)

Journal ref: Physical Review E, 108, 054904, (2023)

Graph-Induced Syntactic-Semantic Spaces in Transformer-Based Variational AutoEncoders

Authors: Yingji Zhang, Marco Valentino, Danilo S. Carvalho, Ian Pratt-Hartmann, André Freitas

Abstract: The injection of syntactic information in Variational AutoEncoders (VAEs) has been shown to result in an overall improvement of performances and generalisation. An effective strategy to achieve such a goal is to separate the encoding of distributional semantic features and syntactic structures into heterogeneous latent spaces via multi-task learning or dual encoder architectures. However, existing… ▽ More The injection of syntactic information in Variational AutoEncoders (VAEs) has been shown to result in an overall improvement of performances and generalisation. An effective strategy to achieve such a goal is to separate the encoding of distributional semantic features and syntactic structures into heterogeneous latent spaces via multi-task learning or dual encoder architectures. However, existing works employing such techniques are limited to LSTM-based VAEs. In this paper, we investigate latent space separation methods for structural syntactic injection in Transformer-based VAE architectures (i.e., Optimus). Specifically, we explore how syntactic structures can be leveraged in the encoding stage through the integration of graph-based and sequential models, and how multiple, specialised latent representations can be injected into the decoder's attention mechanism via low-rank operators. Our empirical evaluation, carried out on natural language sentences and mathematical expressions, reveals that the proposed end-to-end VAE architecture can result in a better overall organisation of the latent space, alleviating the information loss occurring in standard VAE setups, resulting in enhanced performances on language modelling and downstream generation tasks. △ Less

Submitted 14 November, 2023; originally announced November 2023.

Learning the dynamics of a one-dimensional plasma model with graph neural networks

Authors: Diogo D Carvalho, Diogo R Ferreira, Luis O Silva

Abstract: We explore the possibility of fully replacing a plasma physics kinetic simulator with a graph neural network-based simulator. We focus on this class of surrogate models given the similarity between their message-passing update mechanism and the traditional physics solver update, and the possibility of enforcing known physical priors into the graph construction and update. We show that our model le… ▽ More We explore the possibility of fully replacing a plasma physics kinetic simulator with a graph neural network-based simulator. We focus on this class of surrogate models given the similarity between their message-passing update mechanism and the traditional physics solver update, and the possibility of enforcing known physical priors into the graph construction and update. We show that our model learns the kinetic plasma dynamics of the one-dimensional plasma model, a predecessor of contemporary kinetic plasma simulation codes, and recovers a wide range of well-known kinetic plasma processes, including plasma thermalization, electrostatic fluctuations about thermal equilibrium, and the drag on a fast sheet and Landau dam**. We compare the performance against the original plasma model in terms of run-time, conservation laws, and temporal evolution of key physical quantities. The limitations of the model are presented and possible directions for higher-dimensional surrogate models for kinetic plasmas are discussed. △ Less

Submitted 13 May, 2024; v1 submitted 26 October, 2023; originally announced October 2023.

Comments: title changed, accepted for publication in Machine Learning: Science and Technology, code available at https://github.com/diogodcarvalho/gns-sheet-model

Journal ref: Mach. Learn.: Sci. Technol. 5 025048 (2024)

Learning characteristic parameters and dynamics of centrifugal pumps under multi-phase flow using physics-informed neural networks

Authors: Felipe de Castro Teixeira Carvalho, Kamaljyoti Nath, Alberto Luiz Serpa, George Em Karniadakis

Abstract: Electrical submersible pumps (ESP) are the second most used artificial lifting equipment in the oil and gas industry due to their high flow rates and boost pressures. They often have to handle multiphase flows, which usually contain a mixture of hydrocarbons, water, and/or sediments. Given these circumstances, emulsions are commonly formed. It is a liquid-liquid flow composed of two immiscible flu… ▽ More Electrical submersible pumps (ESP) are the second most used artificial lifting equipment in the oil and gas industry due to their high flow rates and boost pressures. They often have to handle multiphase flows, which usually contain a mixture of hydrocarbons, water, and/or sediments. Given these circumstances, emulsions are commonly formed. It is a liquid-liquid flow composed of two immiscible fluids whose effective viscosity and density differ from the single phase separately. In this context, accurate modeling of ESP systems is crucial for optimizing oil production and implementing control strategies. However, real-time and direct measurement of fluid and system characteristics is often impractical due to time constraints and economy. Hence, indirect methods are generally considered to estimate the system parameters. In this paper, we formulate a machine learning model based on Physics-Informed Neural Networks (PINNs) to estimate crucial system parameters. In order to study the efficacy of the proposed PINN model, we conduct computational studies using not only simulated but also experimental data for different water-oil ratios. We evaluate the state variable's dynamics and unknown parameters for various combinations when only intake and discharge pressure measurements are available. We also study structural and practical identifiability analyses based on commonly available pressure measurements. The PINN model could reduce the requirement of expensive field laboratory tests used to estimate fluid properties. △ Less

Submitted 4 October, 2023; originally announced October 2023.

Multi-Bellman operator for convergence of $Q$-learning with linear function approximation

Authors: Diogo S. Carvalho, Pedro A. Santos, Francisco S. Melo

Abstract: We study the convergence of $Q$-learning with linear function approximation. Our key contribution is the introduction of a novel multi-Bellman operator that extends the traditional Bellman operator. By exploring the properties of this operator, we identify conditions under which the projected multi-Bellman operator becomes contractive, providing improved fixed-point guarantees compared to the Bell… ▽ More We study the convergence of $Q$-learning with linear function approximation. Our key contribution is the introduction of a novel multi-Bellman operator that extends the traditional Bellman operator. By exploring the properties of this operator, we identify conditions under which the projected multi-Bellman operator becomes contractive, providing improved fixed-point guarantees compared to the Bellman operator. To leverage these insights, we propose the multi $Q$-learning algorithm with linear function approximation. We demonstrate that this algorithm converges to the fixed-point of the projected multi-Bellman operator, yielding solutions of arbitrary accuracy. Finally, we validate our approach by applying it to well-known environments, showcasing the effectiveness and applicability of our findings. △ Less

Submitted 28 September, 2023; originally announced September 2023.

Spectrally Sharp Near-Field Thermal Emission: Revealing Some Disagreements between a Casimir-Polder Sensor and Predictions from Far-Field Emittance

Authors: J. C. de Aquino Carvalho, I. Maurin, P. Chaves de Souza Segundo, A. Laliotis, D. de Sousa Meneses, D. Bloch

Abstract: Near-field thermal emission largely exceed blackbody radiation, owing to spectrally sharp emission in surface polaritons. We turn Casimir-Polder interaction between Cs(7P1/2) and a sapphire interface, into a sensor sharply filtering, at 24.687 THz, the near-field sapphire emission at ~ 24.5 THz. Temperature evolution of sapphire mode is demonstrated. The Cs sensor, sensitive to both dispersion and… ▽ More Near-field thermal emission largely exceed blackbody radiation, owing to spectrally sharp emission in surface polaritons. We turn Casimir-Polder interaction between Cs(7P1/2) and a sapphire interface, into a sensor sharply filtering, at 24.687 THz, the near-field sapphire emission at ~ 24.5 THz. Temperature evolution of sapphire mode is demonstrated. The Cs sensor, sensitive to both dispersion and dissipation, suggests the polariton to be red-shifted and sharper, as compared, up to 1100 K, to predictions from far-field sapphire emission, affected by birefringence and multiple resonances. △ Less

Submitted 27 September, 2023; originally announced September 2023.

Comments: Accepted (Aug 1, 2023) for publication in Physical Review Letters... The Supplematry Material (ref 62) appears at the end of the single document

Journal ref: Phys. Rev. Lett. 131, 143801 (2023)

Canonical typicality under general quantum channels

Authors: Pedro Silva Correia, Gabriel Dias Carvalho, Thiago R. de Oliveira, Raúl O. Vallejos, Fernando de Melo

Abstract: With the control of ever more complex quantum systems becoming a reality, new scenarios are emerging where generalizations of the most foundational aspects of statistical quantum mechanics are imperative. In such experimental scenarios the often natural correspondence between the particles that compose the system and the relevant degrees-of-freedom might not be observed. In the present work we emp… ▽ More With the control of ever more complex quantum systems becoming a reality, new scenarios are emerging where generalizations of the most foundational aspects of statistical quantum mechanics are imperative. In such experimental scenarios the often natural correspondence between the particles that compose the system and the relevant degrees-of-freedom might not be observed. In the present work we employ quantum channels to define generalized subsystems, which should capture the pertinent degrees-of-freedom, and obtain their associated canonical state. Moreover, we show that generalized subsystems also display the phenomena of canonical typicality, i.e., the generalized subsystem description generated from almost any microscopic pure state of the whole system will behave similarly as the corresponding canonical state. In particular we demonstrate that the property regulating the emergence of the canonical typicality behavior is the entropy of the channel used to define the generalized subsystem. △ Less

Submitted 30 August, 2023; originally announced August 2023.

Comments: 6+7, 3 figures. V1. Comments are welcome

Towards Controllable Natural Language Inference through Lexical Inference Types

Authors: Yingji Zhang, Danilo S. Carvalho, Ian Pratt-Hartmann, Andre Freitas

Abstract: Explainable natural language inference aims to provide a mechanism to produce explanatory (abductive) inference chains which ground claims to their supporting premises. A recent corpus called EntailmentBank strives to advance this task by explaining the answer to a question using an entailment tree \cite{dalvi2021explaining}. They employ the T5 model to directly generate the tree, which can explai… ▽ More Explainable natural language inference aims to provide a mechanism to produce explanatory (abductive) inference chains which ground claims to their supporting premises. A recent corpus called EntailmentBank strives to advance this task by explaining the answer to a question using an entailment tree \cite{dalvi2021explaining}. They employ the T5 model to directly generate the tree, which can explain how the answer is inferred. However, it lacks the ability to explain and control the generation of intermediate steps, which is crucial for the multi-hop inference process. % One recent corpus, EntailmentBank, aims to push this task forward by explaining an answer to a question according to an entailment tree \cite{dalvi2021explaining}. They employ T5 to generate the tree directly, which can explain how the answer is inferred but cannot explain how the intermediate is generated, which is essential to the multi-hop inference process. In this work, we focus on proposing a controlled natural language inference architecture for multi-premise explanatory inference. To improve control and enable explanatory analysis over the generation, we define lexical inference types based on Abstract Meaning Representation (AMR) graph and modify the architecture of T5 to learn a latent sentence representation (T5 bottleneck) conditioned on said type information. We also deliver a dataset of approximately 5000 annotated explanatory inference steps, with well-grounded lexical-symbolic operations. Experimental results indicate that the inference ty** induced at the T5 bottleneck can help T5 to generate a conclusion under explicit control. △ Less

Submitted 7 August, 2023; originally announced August 2023.

News from the Swampland -- Constraining string theory with astrophysics and cosmology

Authors: Nils Schöneberg, Léo Vacher, J. D. F. Dias, Martim M. C. D. Carvalho, C. J. A. P. Martins

Abstract: Our current best guess for a unified theory of gravitation and quantum field theory (string theory) generically predicts a set of requirements for a consistently quantized theory, the Swampland criteria. Refined versions of these criteria have recently been shown to be in mild tension with cosmological observations. We summarize the status of the current impact of and constraints on the Swampland… ▽ More Our current best guess for a unified theory of gravitation and quantum field theory (string theory) generically predicts a set of requirements for a consistently quantized theory, the Swampland criteria. Refined versions of these criteria have recently been shown to be in mild tension with cosmological observations. We summarize the status of the current impact of and constraints on the Swampland conjectures from cosmology, and subject a variety of dark energy quintessence models to recently released cosmological datasets. We find that instead of tightening the tension, the new data allows for slightly more freedom in the Swampland criteria. We further demonstrate that if there is no theoretical argument made to prevent interactions of the moduli fields with the electromagnetic sector, a novel fine-tuning argument arises from the extremely tight current constraints on such interactions. Finally, we conclude with a cautionary tale on model-independent reconstructions of the Swampland criteria from expansion rate data. △ Less

Submitted 8 August, 2023; v1 submitted 27 July, 2023; originally announced July 2023.

Comments: 35 pages, 20 figures, 4 tables. All comments are welcome! [v2: Added citations, corrected eq. 2.6]

RDSim, a fast and comprehensive simulation of radio detection of air showers

Authors: Washington R. de Carvalho Jr., Abha Khakurdikar

Abstract: We present RDSim, a fast and comprehensive framework for the simulation of the radio emission and detection of downgoing air showers. It can handle any downgoing shower that can be simulated with ZHAireS including those induced by CC and NC neutrino interactions and $τ$ decays. RDSim is based on a superposition toymodel that disentangles the Askaryan and geomagnetic components of the shower emissi… ▽ More We present RDSim, a fast and comprehensive framework for the simulation of the radio emission and detection of downgoing air showers. It can handle any downgoing shower that can be simulated with ZHAireS including those induced by CC and NC neutrino interactions and $τ$ decays. RDSim is based on a superposition toymodel that disentangles the Askaryan and geomagnetic components of the shower emission. By using full ZHAireS simulations as input, it is able to estimate the full radio footprint on the ground. A single input simulation at a given energy and arrival direction can be scaled in energy and rotated in azimuth by taking into account all relevant effects. This makes it possible to simulate a huge number of geometries and energies using just a few ZHAireS input simulations. The framework takes into account the main characteristics of the detector, such as trigger setups, thresholds and antenna patterns. To accommodate arrays that use particle detectors for triggering, such as the Auger RD extension, it also features a second toymodel to estimate the muon density at ground level, which is used to perform simple particle trigger simulations. It's speed makes it possible to investigate in detail events with a very low trigger probability, as well as many geometrical effects due to the array layout. In case more detailed studies of the radio detection are needed, RDSim can also be used to sweep the phase-space for the efficient creation of dedicated full simulation sets. This is particularly important in the case of neutrino events, that have extra variables that greatly impact shower characteristics, such as interaction or $τ$ decay depth as well as the type of interaction and it's fluctuations. △ Less

Submitted 14 July, 2023; originally announced July 2023.

Comments: arXiv admin note: text overlap with arXiv:2307.07351

Analysis of the orbital evolution of space debris using a solar sail and natural forces

Authors: Jean Paulo dos S. Carvalho, Rodolpho Vilhena de Moraes, Antonio Fernando Bertachini de A. Prado

Abstract: In this work, the orbital evolution of these objects that are located in the geostationary orbit (GEO) is analyzed. Knowing this, the possibility of using a solar sail is considered to help to clean the space environment. The main natural environmental perturbations that act in the orbit of the debris are considered in the dynamics. Such forces acting in the solar sail can force the growth of the… ▽ More In this work, the orbital evolution of these objects that are located in the geostationary orbit (GEO) is analyzed. Knowing this, the possibility of using a solar sail is considered to help to clean the space environment. The main natural environmental perturbations that act in the orbit of the debris are considered in the dynamics. Such forces acting in the solar sail can force the growth of the eccentricity of these objects in the GEO orbit. Several authors have presented models of the solar radiation pressure considering the single-averaged model. But, doing a literature research, we found that the authors consider the Earth around the Sun in a circular and inclined orbit. Our contribution to the SRP model is in develo** a different approach from other authors, where we consider the Sun in an elliptical and inclined orbit, which is valid for other bodies in the solar system when the eccentricity cannot be neglected. The expression of the SRP is developed up to the second order. We found that the first-order term is much superior to the second-order term, so the quadrupole term can be neglected. Another contribution is the approach to identify the initial conditions of the perigee argument (g) and the longitude of the ascending node (h), where some values of the (g, h) plane contribute to amplify the eccentricity growth. In the numerical simulations we consider real data from space debris removed from the site Stuff in Space. The solar sail helps to clean up the space environment using a propulsion system that uses the Sun itself, a clean and abundant energy source, unlike chemical propellants, to contribute to the sustainability of space exploration. △ Less

Submitted 16 July, 2023; originally announced July 2023.

RDSim: A fast, accurate and flexible framework for the simulation of the radio emission and detection of downgoing air showers

Authors: Washington R. de Carvalho Jr., Abha Khakurdikar

Abstract: RDSim is a fast, accurate and flexible framework for the simulation of the radio emission of downgoing air showers and its detection by an arbitrary array, including showers initiated by neutrino interactions or tau-lepton decays. RDSim was build around speed and is based on simple and fast, yet still accurate, toymodel-like approaches. It models the radio emission using a superposition emission m… ▽ More RDSim is a fast, accurate and flexible framework for the simulation of the radio emission of downgoing air showers and its detection by an arbitrary array, including showers initiated by neutrino interactions or tau-lepton decays. RDSim was build around speed and is based on simple and fast, yet still accurate, toymodel-like approaches. It models the radio emission using a superposition emission model that disentangles the Askaryan and geomagnetic components of the shower radio emission. It uses full ZHAireS simulations as an input to estimate the electric field at any position on the ground. A single input simulation can be scaled in energy and rotated in azimuth, taking into account all relevant effects. This makes it possible to simulate a huge number of geometries and energies using just a few ZHAireS input simulations. RDSim takes into account the main characteristics of the detector, such as trigger setups, thresholds and antenna patterns. To accommodate arrays that use particle detectors for triggering, such as the Auger RD extension, it also features a second toymodel to estimate the muon density at ground level and perform simple particle trigger simulations. Owing to the large statistics made possible by its speed, it can be used to investigate in detail events with a very low trigger probability and geometrical effects due to the array layout, making it specially suited to be used as a fast and accurate aperture calculator. In case more detailed studies of the radio emission and detector response are desired, RDSim can also be used to sweep the phase-space for the efficient creation of dedicated full simulation sets. This is particularly important in the case of neutrino events, that have extra variables that greatly impact shower characteristics, such as interaction or $τ$ decay depth as well as the type of interaction and it's fluctuations. △ Less

Submitted 14 July, 2023; originally announced July 2023.

Equilibration of Isolated Systems: investigating the role of coarse-graining on the initial state magnetization

Authors: Gabriel Dias Carvalho, Luis Fernando dos Prazeres, Pedro Silva Correia, Thiago R de Oliveira

Abstract: Many theoretical and experimental results show that even isolated quantum systems evolving unitarily may equilibrate, since the evolution of some observables may be around an equilibrium value with negligible fluctuations most of the time. There are rigorous theorems giving the conditions for such equilibration to happen. In particular, initial states prepared with a lack of resolution in the ener… ▽ More Many theoretical and experimental results show that even isolated quantum systems evolving unitarily may equilibrate, since the evolution of some observables may be around an equilibrium value with negligible fluctuations most of the time. There are rigorous theorems giving the conditions for such equilibration to happen. In particular, initial states prepared with a lack of resolution in the energy will equilibrate. We investigate how equilibration may be affected by a lack of resolution, or coarse-graining, in the magnetization of the initial state. In particular, for a chaotic spin chain and using exact diagonalization, we show that the level of equilibration of an initial state with a coarse, not well-defined magnetization is different from the level of an initial state with well-defined magnetization. This difference will depend on the degree of coarse-graining and the direction of magnetization. We also analyze the time for the system to reach equilibrium, showing good agreement with theoretical estimates and with some evidence that less resolution leads to faster equilibration. Our study highlights the crucial role of initial state preparation in the equilibration dynamics of quantum systems and provides new insights into the fundamental nature of equilibration in closed systems. △ Less

Submitted 14 December, 2023; v1 submitted 19 May, 2023; originally announced May 2023.

Journal ref: Gabriel D. Carvalho, Luis F. dos Prazeres, Pedro S. Correia, Thiago R. de Oliveira, Physics Letters A, Volume 494, 2024, 129276

Multi-Relational Hyperbolic Word Embeddings from Natural Language Definitions

Authors: Marco Valentino, Danilo S. Carvalho, André Freitas

Abstract: Natural language definitions possess a recursive, self-explanatory semantic structure that can support representation learning methods able to preserve explicit conceptual relations and constraints in the latent space. This paper presents a multi-relational model that explicitly leverages such a structure to derive word embeddings from definitions. By automatically extracting the relations linking… ▽ More Natural language definitions possess a recursive, self-explanatory semantic structure that can support representation learning methods able to preserve explicit conceptual relations and constraints in the latent space. This paper presents a multi-relational model that explicitly leverages such a structure to derive word embeddings from definitions. By automatically extracting the relations linking defined and defining terms from dictionaries, we demonstrate how the problem of learning word embeddings can be formalised via a translational framework in Hyperbolic space and used as a proxy to capture the global semantic structure of definitions. An extensive empirical analysis demonstrates that the framework can help imposing the desired structural constraints while preserving the semantic map** required for controllable and interpretable traversal. Moreover, the experiments reveal the superiority of the Hyperbolic word embeddings over the Euclidean counterparts and demonstrate that the multi-relational approach can obtain competitive results when compared to state-of-the-art neural models, with the advantage of being intrinsically more efficient and interpretable. △ Less

Submitted 16 February, 2024; v1 submitted 12 May, 2023; originally announced May 2023.

Comments: Accepted at the 18th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2024), camera-ready

Learning Disentangled Semantic Spaces of Explanations via Invertible Neural Networks

Authors: Yingji Zhang, Danilo S. Carvalho, André Freitas

Abstract: Disentangled latent spaces usually have better semantic separability and geometrical properties, which leads to better interpretability and more controllable data generation. While this has been well investigated in Computer Vision, in tasks such as image disentanglement, in the NLP domain sentence disentanglement is still comparatively under-investigated. Most previous work have concentrated on d… ▽ More Disentangled latent spaces usually have better semantic separability and geometrical properties, which leads to better interpretability and more controllable data generation. While this has been well investigated in Computer Vision, in tasks such as image disentanglement, in the NLP domain sentence disentanglement is still comparatively under-investigated. Most previous work have concentrated on disentangling task-specific generative factors, such as sentiment, within the context of style transfer. In this work, we focus on a more general form of sentence disentanglement, targeting the localised modification and control of more general sentence semantic features. To achieve this, we contribute to a novel notion of sentence semantic disentanglement and introduce a flow-based invertible neural network (INN) mechanism integrated with a transformer-based language Autoencoder (AE) in order to deliver latent spaces with better separability properties. Experimental results demonstrate that the model can conform the distributed latent space into a better semantically disentangled sentence space, leading to improved language interpretability and controlled generation when compared to the recent state-of-the-art language VAE models. △ Less

Submitted 11 June, 2024; v1 submitted 2 May, 2023; originally announced May 2023.

Comments: ACL 2024

Roles of packing fraction, microscopic friction and projectile spin in cratering by impact

Authors: Douglas Daniel de Carvalho, Nicolao Cerqueira Lima, Erick de Moraes Franklin

Abstract: From small seeds falling from trees to asteroids colliding with planets and moons, the impact of projectiles onto granular targets occurs in nature at different scales. In this paper, we investigate open questions in the mechanics of granular cratering, in particular the forces acting on the projectile, and the roles of granular packing, grain-grain friction and projectile spin. For that, we carri… ▽ More From small seeds falling from trees to asteroids colliding with planets and moons, the impact of projectiles onto granular targets occurs in nature at different scales. In this paper, we investigate open questions in the mechanics of granular cratering, in particular the forces acting on the projectile, and the roles of granular packing, grain-grain friction and projectile spin. For that, we carried out DEM (discrete element method) computations of the impact of solid projectiles on a cohesionless granular medium, where we varied the projectile and grain properties (diameter, density, friction and packing fraction) for different available energies (within relatively small values). We found that a denser region forms below the projectile, pushing it back and causing its rebound by the end of its motion, and that solid friction affects considerably the crater morphology. Besides, we show that the penetration length increases with the initial spin of the projectile, and that differences in initial packing fractions can engender the diversity of scaling laws found in the literature. Finally, we propose an ad hoc scaling that collapsed our data for the penetration length and can perhaps unify existing correlations. Our results provide new insights into the formation of craters in granular matter. △ Less

Submitted 25 November, 2023; v1 submitted 21 April, 2023; originally announced April 2023.

Comments: Accepted manuscript for Physical Review E, 105, 034903, (2023), DOI: 10.1103/PhysRevE.107.044901, https://doi.org/10.1103/PhysRevE.107.044901

Journal ref: Phys. Rev. E 107, 044901, 2023

Analysis of business process automation as linear time-invariant system network

Authors: Mauricio Jacobo-Romero, Danilo S. Carvalho, Andre Freitas

Abstract: In this work, we examined Business Process (BP) production as a signal; this novel approach explores a BP workflow as a linear time-invariant (LTI) system. We analysed BP productivity in the frequency domain; this standpoint examines how labour and capital act as BP input signals and how their fundamental frequencies affect BP production. Our research also proposes a simulation framework of a BP i… ▽ More In this work, we examined Business Process (BP) production as a signal; this novel approach explores a BP workflow as a linear time-invariant (LTI) system. We analysed BP productivity in the frequency domain; this standpoint examines how labour and capital act as BP input signals and how their fundamental frequencies affect BP production. Our research also proposes a simulation framework of a BP in the frequency domain for estimating productivity gains due to the introduction of automation steps. Our ultimate goal was to supply evidence to address Solow's Paradox. △ Less

Submitted 9 February, 2023; originally announced February 2023.

Comments: 16 pages, 24 figures

On the fundamental tone of the $p$-Laplacian on Riemannian manifolds and applications

Authors: Francisco G. de S. Carvalho, Marcos Petrucio Cavalcante

Abstract: We present a general lower bound for the fundamental tone for the $p$-Laplacian on Riemannian manifolds carrying a special kind of function. We then apply our result to the cases of negatively curved simply connected manifolds, a class of warped product manifolds and for a class of Riemannian submersions. We present a general lower bound for the fundamental tone for the $p$-Laplacian on Riemannian manifolds carrying a special kind of function. We then apply our result to the cases of negatively curved simply connected manifolds, a class of warped product manifolds and for a class of Riemannian submersions. △ Less

Submitted 28 January, 2023; originally announced January 2023.

Journal ref: J. Math. Anal. Appl. 506 (2022) 125703

Montague semantics and modifier consistency measurement in neural language models

Authors: Danilo S. Carvalho, Edoardo Manino, Julia Rozanova, Lucas Cordeiro, André Freitas

Abstract: In recent years, distributional language representation models have demonstrated great practical success. At the same time, the need for interpretability has elicited questions on their intrinsic properties and capabilities. Crucially, distributional models are often inconsistent when dealing with compositional phenomena in natural language, which has significant implications for their safety and… ▽ More In recent years, distributional language representation models have demonstrated great practical success. At the same time, the need for interpretability has elicited questions on their intrinsic properties and capabilities. Crucially, distributional models are often inconsistent when dealing with compositional phenomena in natural language, which has significant implications for their safety and fairness. Despite this, most current research on compositionality is directed towards improving their performance on similarity tasks only. This work takes a different approach, and proposes a methodology for measuring compositional behavior in contemporary language models. Specifically, we focus on adjectival modifier phenomena in adjective-noun phrases. We introduce three novel tests of compositional behavior inspired by Montague semantics. Our experimental results indicate that current neural language models behave according to the expected linguistic theories to a limited extent only. This raises the question of whether these language models are not able to capture the semantic properties we evaluated, or whether linguistic theories from Montagovian tradition would not match the expected capabilities of distributional models. △ Less

Submitted 3 April, 2023; v1 submitted 10 October, 2022; originally announced December 2022.

Collaborative behavior of intruders moving amid grains

Authors: Douglas Daniel de Carvalho, Erick de Moraes Franklin

Abstract: We investigate the motion of groups of intruders in a two-dimensional granular system by using discrete numerical simulations. By imposing either a constant velocity or a thrusting force on larger disks (intruders) that move within smaller ones (grains), we obtained instantaneous positions and components of forces for each intruder and grain. We found that (i) intruders cooperate even when at rela… ▽ More We investigate the motion of groups of intruders in a two-dimensional granular system by using discrete numerical simulations. By imposing either a constant velocity or a thrusting force on larger disks (intruders) that move within smaller ones (grains), we obtained instantaneous positions and components of forces for each intruder and grain. We found that (i) intruders cooperate even when at relatively large distances from each other; (ii) the cooperative dynamics is the result of contact chains linking the intruders as well as compaction and expansion of the granular medium in front and behind, respectively, each intruder; (iii) the collaborative behavior depends on the initial arrangement of intruders; and (iv) for some initial arrangements, the same spatial configuration is eventually reached. Finally, we show the existence of an optimal distance for minimum drag for a given set of intruders, which can prove useful for devices stirring the ground or other granular surfaces. △ Less

Submitted 2 December, 2022; originally announced December 2022.

Comments: This article may be downloaded for personal use only. Any other use requires prior permission of the author and AIP Publishing. This article appeared in Phys. Fluids 34, 123306 (2022) and may be found at https://doi.org/10.1063/5.0124556

Journal ref: Physics of Fluids, 34, 123306, 2022

Thermodynamics of static and stationary black holes in Einstein-Gauss-Bonnet gravity with dark matter

Authors: I. D. D. Carvalho, G. Alencar, C. R. Muniz

Abstract: This paper studies Einstein-Gauss-Bonnet (EGB) black holes surrounded by three phenomenological distributions of dark matter halos. The main result is obtaining the analytical solutions for the metric and all thermodynamic quantities, such as Hawking temperature, entropy, constant-volume heat capacity, and Gibbs free energy for static and stationary black hole solutions. Consequently, we determine… ▽ More This paper studies Einstein-Gauss-Bonnet (EGB) black holes surrounded by three phenomenological distributions of dark matter halos. The main result is obtaining the analytical solutions for the metric and all thermodynamic quantities, such as Hawking temperature, entropy, constant-volume heat capacity, and Gibbs free energy for static and stationary black hole solutions. Consequently, we determine a non-null horizon radius at which the black hole halts its evaporation by vanishing the temperature, indicating the emergence of remnants. In the stationary case, we also obtain the ergosphere regions for the found solutions and compare them. Finally, we find local and global phase transitions by studying the behavior of the heat capacity and Gibbs free energy. △ Less

Submitted 14 December, 2022; v1 submitted 21 November, 2022; originally announced November 2022.

Comments: We add the rotation black hole case. The new version has 22 pages and 11 figures

An agent-based approach to procedural city generation incorporating Land Use and Transport Interaction models

Authors: Luiz Fernando Silva Eugênio dos Santos, Claus Aranha, André Ponce de Leon F de Carvalho

Abstract: We apply the knowledge of urban settings established with the study of Land Use and Transport Interaction (LUTI) models to develop reward functions for an agent-based system capable of planning realistic artificial cities. The system aims to replicate in the micro scale the main components of real settlements, such as zoning and accessibility in a road network. Moreover, we propose a novel represe… ▽ More We apply the knowledge of urban settings established with the study of Land Use and Transport Interaction (LUTI) models to develop reward functions for an agent-based system capable of planning realistic artificial cities. The system aims to replicate in the micro scale the main components of real settlements, such as zoning and accessibility in a road network. Moreover, we propose a novel representation for the agent's environment that efficiently combines the road graph with a discrete model for the land. Our system starts from an empty map consisting only of the road network graph, and the agent incrementally expands it by building new sites while distinguishing land uses between residential, commercial, industrial, and recreational. △ Less

Submitted 21 October, 2022; originally announced November 2022.

Comments: 12 pages, 6 figures, XIX Encontro Nacional de Inteligência Artificial e Computacional (ENIAC 2022)

Centralized Training with Hybrid Execution in Multi-Agent Reinforcement Learning

Authors: Pedro P. Santos, Diogo S. Carvalho, Miguel Vasco, Alberto Sardinha, Pedro A. Santos, Ana Paiva, Francisco S. Melo

Abstract: We introduce hybrid execution in multi-agent reinforcement learning (MARL), a new paradigm in which agents aim to successfully complete cooperative tasks with arbitrary communication levels at execution time by taking advantage of information-sharing among the agents. Under hybrid execution, the communication level can range from a setting in which no communication is allowed between agents (fully… ▽ More We introduce hybrid execution in multi-agent reinforcement learning (MARL), a new paradigm in which agents aim to successfully complete cooperative tasks with arbitrary communication levels at execution time by taking advantage of information-sharing among the agents. Under hybrid execution, the communication level can range from a setting in which no communication is allowed between agents (fully decentralized), to a setting featuring full communication (fully centralized), but the agents do not know beforehand which communication level they will encounter at execution time. To formalize our setting, we define a new class of multi-agent partially observable Markov decision processes (POMDPs) that we name hybrid-POMDPs, which explicitly model a communication process between the agents. We contribute MARO, an approach that makes use of an auto-regressive predictive model, trained in a centralized manner, to estimate missing agents' observations at execution time. We evaluate MARO on standard scenarios and extensions of previous benchmarks tailored to emphasize the negative impact of partial observability in MARL. Experimental results show that our method consistently outperforms relevant baselines, allowing agents to act with faulty communication while successfully exploiting shared information. △ Less

Submitted 5 June, 2023; v1 submitted 12 October, 2022; originally announced October 2022.

Formal Semantic Geometry over Transformer-based Variational AutoEncoder

Authors: Yingji Zhang, Danilo S. Carvalho, Ian Pratt-Hartmann, André Freitas

Abstract: Formal/symbolic semantics can provide canonical, rigid controllability and interpretability to sentence representations due to their \textit{localisation} or \textit{composition} property. How can we deliver such property to the current distributional sentence representations to control and interpret the generation of language models (LMs)? In this work, we theoretically frame the sentence semanti… ▽ More Formal/symbolic semantics can provide canonical, rigid controllability and interpretability to sentence representations due to their \textit{localisation} or \textit{composition} property. How can we deliver such property to the current distributional sentence representations to control and interpret the generation of language models (LMs)? In this work, we theoretically frame the sentence semantics as the composition of \textit{semantic role - word content} features and propose the formal semantic geometry. To inject such geometry into Transformer-based LMs (i.e. GPT2), we deploy Transformer-based Variational AutoEncoder with a supervision approach, where the sentence generation can be manipulated and explained over low-dimensional latent Gaussian space. In addition, we propose a new probing algorithm to guide the movement of sentence vectors over such geometry. Experimental results reveal that the formal semantic geometry can potentially deliver better control and interpretation to sentence generation. △ Less

Submitted 11 June, 2024; v1 submitted 12 October, 2022; originally announced October 2022.

Learning Disentangled Representations for Natural Language Definitions

Authors: Danilo S. Carvalho, Giangiacomo Mercatali, Yingji Zhang, Andre Freitas

Abstract: Disentangling the encodings of neural models is a fundamental aspect for improving interpretability, semantic control and downstream task performance in Natural Language Processing. Currently, most disentanglement methods are unsupervised or rely on synthetic datasets with known generative factors. We argue that recurrent syntactic and semantic regularities in textual data can be used to provide t… ▽ More Disentangling the encodings of neural models is a fundamental aspect for improving interpretability, semantic control and downstream task performance in Natural Language Processing. Currently, most disentanglement methods are unsupervised or rely on synthetic datasets with known generative factors. We argue that recurrent syntactic and semantic regularities in textual data can be used to provide the models with both structural biases and generative factors. We leverage the semantic structures present in a representative and semantically dense category of sentence types, definitional sentences, for training a Variational Autoencoder to learn disentangled representations. Our experimental results show that the proposed model outperforms unsupervised baselines on several qualitative and quantitative benchmarks for disentanglement, and it also improves the results in the downstream task of definition modeling. △ Less

Submitted 15 February, 2023; v1 submitted 22 September, 2022; originally announced October 2022.

Comments: Findings of EACL 2023

Estimating productivity gains in digital automation

Authors: Mauricio Jacobo-Romero, Danilo S. Carvalho, André Freitas

Abstract: This paper proposes a novel productivity estimation model to evaluate the effects of adopting Artificial Intelligence (AI) components in a production chain. Our model provides evidence to address the "AI's" Solow's Paradox. We provide (i) theoretical and empirical evidence to explain Solow's dichotomy; (ii) a data-driven model to estimate and asses productivity variations; (iii) a methodology unde… ▽ More This paper proposes a novel productivity estimation model to evaluate the effects of adopting Artificial Intelligence (AI) components in a production chain. Our model provides evidence to address the "AI's" Solow's Paradox. We provide (i) theoretical and empirical evidence to explain Solow's dichotomy; (ii) a data-driven model to estimate and asses productivity variations; (iii) a methodology underpinned on process mining datasets to determine the business process, BP, and productivity; (iv) a set of computer simulation parameters; (v) and empirical analysis on labour-distribution. These provide data on why we consider AI Solow's paradox a consequence of metric mismeasurement. △ Less

Submitted 8 October, 2022; v1 submitted 3 October, 2022; originally announced October 2022.

Comments: 11 pages and 9 figures

Horizon Fractalization in Black Strings Ungravity

Authors: I. D. D. Carvalho, J. Furtado, R. R. Landim, G. Alencar

Abstract: In this paper we study the scalar(tensor) and vector unparticle corrections for cosmic and black strings. Initially we have considered an static cosmic string ansatz from which we obtain the solution in terms of first and second kind Bessel functions. We have also obtained the solution for black string in the unparticle scenario. We could identify two regimes, namely, a gravity dominated regime an… ▽ More In this paper we study the scalar(tensor) and vector unparticle corrections for cosmic and black strings. Initially we have considered an static cosmic string ansatz from which we obtain the solution in terms of first and second kind Bessel functions. We have also obtained the solution for black string in the unparticle scenario. We could identify two regimes, namely, a gravity dominated regime and an ungravity dominated regime. In the gravity dominated regime the black string solution recovers the usual solution for black strings. The Hawking temperature was also studied in both regimes and in the ungravity dominated regime. As in the static and rotating black hole, we found a fractalization of the event horizon. This points to the fact that fractalization is a natural consequence of unparticles. Finally, we study the thermodynamic of the black string in the ungravity scenario by computing the entropy, heat capacity and free energy. For both cases we find that, depending on the region of the parameter $d_U$, we can have phase transitions. △ Less

Submitted 29 September, 2022; originally announced September 2022.

Comments: 18 pages, 6 figures

No Pattern, No Recognition: a Survey about Reproducibility and Distortion Issues of Text Clustering and Topic Modeling

Authors: Marília Costa Rosendo Silva, Felipe Alves Siqueira, João Pedro Mantovani Tarrega, João Vitor Pataca Beinotti, Augusto Sousa Nunes, Miguel de Mattos Gardini, Vinícius Adolfo Pereira da Silva, Nádia Félix Felipe da Silva, André Carlos Ponce de Leon Ferreira de Carvalho

Abstract: Extracting knowledge from unlabeled texts using machine learning algorithms can be complex. Document categorization and information retrieval are two applications that may benefit from unsupervised learning (e.g., text clustering and topic modeling), including exploratory data analysis. However, the unsupervised learning paradigm poses reproducibility issues. The initialization can lead to variabi… ▽ More Extracting knowledge from unlabeled texts using machine learning algorithms can be complex. Document categorization and information retrieval are two applications that may benefit from unsupervised learning (e.g., text clustering and topic modeling), including exploratory data analysis. However, the unsupervised learning paradigm poses reproducibility issues. The initialization can lead to variability depending on the machine learning algorithm. Furthermore, the distortions can be misleading when regarding cluster geometry. Amongst the causes, the presence of outliers and anomalies can be a determining factor. Despite the relevance of initialization and outlier issues for text clustering and topic modeling, the authors did not find an in-depth analysis of them. This survey provides a systematic literature review (2011-2022) of these subareas and proposes a common terminology since similar procedures have different terms. The authors describe research opportunities, trends, and open issues. The appendices summarize the theoretical background of the text vectorization, the factorization, and the clustering algorithms that are directly or indirectly related to the reviewed works. △ Less

Submitted 2 August, 2022; originally announced August 2022.

ACM Class: I.2; I.2.7; I.5.3

Design and Initial Performance of the Prototype for the BEACON Instrument for Detection of Ultrahigh Energy Particles

Authors: D. Southall, C. Deaconu, V. Decoene, E. Oberla, A. Zeolla, J. Alvarez-Muñiz, A. Cummings, Z. Curtis-Ginsberg, A. Hendrick, K. Hughes, R. Krebs, A. Ludwig, K. Mulrey, S. Prohira, W. Rodrigues de Carvalho, Jr., A. Rodriguez, A. Romero-Wolf, H. Schoorlemmer, A. G. Vieregg, S. A. Wissel, E. Zas

Abstract: The Beamforming Elevated Array for COsmic Neutrinos (BEACON) is a planned neutrino telescope designed to detect radio emission from upgoing air showers generated by ultrahigh energy tau neutrino interactions in the Earth. This detection mechanism provides a measurement of the tau flux of cosmic neutrinos. We have installed an 8-channel prototype instrument at high elevation at Barcroft Field Stati… ▽ More The Beamforming Elevated Array for COsmic Neutrinos (BEACON) is a planned neutrino telescope designed to detect radio emission from upgoing air showers generated by ultrahigh energy tau neutrino interactions in the Earth. This detection mechanism provides a measurement of the tau flux of cosmic neutrinos. We have installed an 8-channel prototype instrument at high elevation at Barcroft Field Station, which has been running since 2018, and consists of 4 dual-polarized antennas sensitive between 30-80 MHz, whose signals are filtered, amplified, digitized, and saved to disk using a custom data acquisition system (DAQ). The BEACON prototype is at high elevation to maximize effective volume and uses a directional beamforming trigger to improve rejection of anthropogenic background noise at the trigger level. Here we discuss the design, construction, and calibration of the BEACON prototype instrument. We also discuss the radio frequency environment observed by the instrument, and categorize the types of events seen by the instrument, including a likely cosmic ray candidate event. △ Less

Submitted 29 March, 2023; v1 submitted 20 June, 2022; originally announced June 2022.

Comments: 21 pages, 20 figures

Journal ref: NIM-A Volume 1048, March 2023, 167889

Systematicity, Compositionality and Transitivity of Deep NLP Models: a Metamorphic Testing Perspective

Authors: Edoardo Manino, Julia Rozanova, Danilo Carvalho, Andre Freitas, Lucas Cordeiro

Abstract: Metamorphic testing has recently been used to check the safety of neural NLP models. Its main advantage is that it does not rely on a ground truth to generate test cases. However, existing studies are mostly concerned with robustness-like metamorphic relations, limiting the scope of linguistic properties they can test. We propose three new classes of metamorphic relations, which address the proper… ▽ More Metamorphic testing has recently been used to check the safety of neural NLP models. Its main advantage is that it does not rely on a ground truth to generate test cases. However, existing studies are mostly concerned with robustness-like metamorphic relations, limiting the scope of linguistic properties they can test. We propose three new classes of metamorphic relations, which address the properties of systematicity, compositionality and transitivity. Unlike robustness, our relations are defined over multiple source inputs, thus increasing the number of test cases that we can produce by a polynomial factor. With them, we test the internal consistency of state-of-the-art NLP models, and show that they do not always behave according to their expected linguistic properties. Lastly, we introduce a novel graphical notation that efficiently summarises the inner structure of metamorphic relations. △ Less

Submitted 26 April, 2022; originally announced April 2022.

Comments: Findings of the Association for Computational Linguistics 2022

Contacts, motion and chain-breaking in a two-dimensional granular system displaced by an intruder

Authors: Douglas Daniel de Carvalho, Nicolao Cerqueira Lima, Erick de Moraes Franklin

Abstract: We investigate numerically how the motion of an intruder within a two-dimensional granular system affects its structure and produces drag on the intruder. We made use of discrete numerical simulations in which a larger disk (intruder) is driven at constant speed amid smaller disks confined in a rectangular cell. By varying the intruder's velocity and the basal friction, we obtained the resultant f… ▽ More We investigate numerically how the motion of an intruder within a two-dimensional granular system affects its structure and produces drag on the intruder. We made use of discrete numerical simulations in which a larger disk (intruder) is driven at constant speed amid smaller disks confined in a rectangular cell. By varying the intruder's velocity and the basal friction, we obtained the resultant force on the intruder and the instantaneous network of contact forces, which we analyze at both the cell and grain scales. We found that there is a bearing network that percolates forces from the intruder toward the walls, being responsible for jammed regions and high values of the drag force, and a dissipative network that percolates small forces within the grains, in agreement with previous experiments on compressed granular systems. In addition, we found the anisotropy levels of the contact network for different force magnitudes and regions, that the force network can reach regions far downstream of the intruder by the end of the intruder's motion, that the extent of the force network decreases with decreasing the basal friction, and that the void region (cavity) that appears downstream the intruder tends to disappear for lower values of the basal friction. Interestingly, our results show that grains within the bearing chains creep while the chains break, revealing the mechanism by which bearing chains collapse. △ Less

Submitted 18 March, 2022; originally announced March 2022.

Comments: Accepted manuscript for Physical Review E, 105, 034903, (2022), DOI: 10.1103/PhysRevE.105.034903, https://doi.org/10.1103/PhysRevE.105.034903

Journal ref: Phys. Rev. E 105, 034903, 2022

Hierarchically Structured Scheduling and Execution of Tasks in a Multi-Agent Environment

Authors: Diogo S. Carvalho, Biswa Sengupta

Abstract: In a warehouse environment, tasks appear dynamically. Consequently, a task management system that matches them with the workforce too early (e.g., weeks in advance) is necessarily sub-optimal. Also, the rapidly increasing size of the action space of such a system consists of a significant problem for traditional schedulers. Reinforcement learning, however, is suited to deal with issues requiring m… ▽ More In a warehouse environment, tasks appear dynamically. Consequently, a task management system that matches them with the workforce too early (e.g., weeks in advance) is necessarily sub-optimal. Also, the rapidly increasing size of the action space of such a system consists of a significant problem for traditional schedulers. Reinforcement learning, however, is suited to deal with issues requiring making sequential decisions towards a long-term, often remote, goal. In this work, we set ourselves on a problem that presents itself with a hierarchical structure: the task-scheduling, by a centralised agent, in a dynamic warehouse multi-agent environment and the execution of one such schedule, by decentralised agents with only partial observability thereof. We propose to use deep reinforcement learning to solve both the high-level scheduling problem and the low-level multi-agent problem of schedule execution. Finally, we also conceive the case where centralisation is impossible at test time and workers must learn how to cooperate in executing the tasks in an environment with no schedule and only partial observability. △ Less

Submitted 6 March, 2022; originally announced March 2022.

Unsupervised machine learning approaches to the $q$-state Potts model

Authors: Andrea Tirelli, Danyella O. Carvalho, Lucas A. Oliveira, J. P. Lima, Natanael C. Costa, Raimundo R. dos Santos

Abstract: In this paper with study phase transitions of the $q$-state Potts model, through a number of unsupervised machine learning techniques, namely Principal Component Analysis (PCA), $k$-means clustering, Uniform Manifold Approximation and Projection (UMAP), and Topological Data Analysis (TDA). Even though in all cases we are able to retrieve the correct critical temperatures $T_c(q)$, for $q = 3, 4$ a… ▽ More In this paper with study phase transitions of the $q$-state Potts model, through a number of unsupervised machine learning techniques, namely Principal Component Analysis (PCA), $k$-means clustering, Uniform Manifold Approximation and Projection (UMAP), and Topological Data Analysis (TDA). Even though in all cases we are able to retrieve the correct critical temperatures $T_c(q)$, for $q = 3, 4$ and $5$, results show that non-linear methods as UMAP and TDA are less dependent on finite size effects, while still being able to distinguish between first and second order phase transitions. This study may be considered as a benchmark for the use of different unsupervised machine learning algorithms in the investigation of phase transitions. △ Less

Submitted 18 March, 2022; v1 submitted 13 December, 2021; originally announced December 2021.

Comments: Added computation of critical exponents; exposition improved

The Impact of Data Distribution on Q-learning with Function Approximation

Authors: Pedro P. Santos, Diogo S. Carvalho, Alberto Sardinha, Francisco S. Melo

Abstract: We study the interplay between the data distribution and Q-learning-based algorithms with function approximation. We provide a unified theoretical and empirical analysis as to how different properties of the data distribution influence the performance of Q-learning-based algorithms. We connect different lines of research, as well as validate and extend previous results. We start by reviewing theor… ▽ More We study the interplay between the data distribution and Q-learning-based algorithms with function approximation. We provide a unified theoretical and empirical analysis as to how different properties of the data distribution influence the performance of Q-learning-based algorithms. We connect different lines of research, as well as validate and extend previous results. We start by reviewing theoretical bounds on the performance of approximate dynamic programming algorithms. We then introduce a novel four-state MDP specifically tailored to highlight the impact of the data distribution in the performance of Q-learning-based algorithms with function approximation, both online and offline. Finally, we experimentally assess the impact of the data distribution properties on the performance of two offline Q-learning-based algorithms under different environments. According to our results: (i) high entropy data distributions are well-suited for learning in an offline manner; and (ii) a certain degree of data diversity (data coverage) and data quality (closeness to optimal policy) are jointly desirable for offline learning. △ Less

Submitted 10 February, 2023; v1 submitted 23 November, 2021; originally announced November 2021.

Gravitational bending angle with finite distances by Casimir wormholes

Authors: I. D. D. Carvalho, G. Alencar, C. R. Muniz

Abstract: In this paper, we investigate the gravitational bending angle due to the Casimir wormholes, which consider the Casimir energy as the source. Furthermore, some of these Casimir wormholes regard Generalized Uncertainty Principle (GUP) corrections of Casimir energy. We use the Ishihara method for the Jacobi metric, which allows us to study the bending angle of light and massive test particles for fin… ▽ More In this paper, we investigate the gravitational bending angle due to the Casimir wormholes, which consider the Casimir energy as the source. Furthermore, some of these Casimir wormholes regard Generalized Uncertainty Principle (GUP) corrections of Casimir energy. We use the Ishihara method for the Jacobi metric, which allows us to study the bending angle of light and massive test particles for finite distances. Beyond the uncorrected Casimir source, we consider many GUP corrections, namely: the Kempf, Mangano and Mann (KMM) model, the Detournay, Gabriel and Spindel (DGS) model, and the so-called type II model for the GUP principle. We also find the deflection angle of light and massive particles in the case of the receiver and the source are far away from the lens. In this case, we also compute the optical scalars: convergence and shear for these Casimir wormholes as a gravitational weak lens. △ Less

Submitted 17 August, 2021; v1 submitted 22 June, 2021; originally announced June 2021.

Comments: 25 pages, 1 figure, Minor corrections, typos fixed, references added

Linear probing of molecules at micrometric distances from a surface with sub-Doppler frequency resolution

Authors: J. Lukusa Mudiayi, I. Maurin, T. Mashimo, J. C. de Aquino Carvalho, D. Bloch, S. K. Tokunaga, B. Darquié, A. Laliotis

Abstract: We report on precision spectroscopy of sub-wavelength confined molecular gases. This was obtained by rovibrational selective reflection of $\mathrm{NH_3}$ and $\mathrm{SF_6}$ gases using a quantum cascade laser at $λ\approx 10.6 μm$. Our technique probes molecules at micrometric distances ($\approx λ/2π$) from the window of a macroscopic cell with sub-MHz resolution, allowing molecule-surface inte… ▽ More We report on precision spectroscopy of sub-wavelength confined molecular gases. This was obtained by rovibrational selective reflection of $\mathrm{NH_3}$ and $\mathrm{SF_6}$ gases using a quantum cascade laser at $λ\approx 10.6 μm$. Our technique probes molecules at micrometric distances ($\approx λ/2π$) from the window of a macroscopic cell with sub-MHz resolution, allowing molecule-surface interaction spectroscopy. We exploit the linearity and high-resolution of our technique to gain novel spectroscopic information on the $\mathrm{SF_6}$ greenhouse gas, useful for enriching molecular databases. The natural extension of our work to thin-cells will allow compact frequency references and improved measurements of the Casimir-Polder interaction with molecules. △ Less

Submitted 13 June, 2021; originally announced June 2021.

Comments: accepted for publication in Physical Review Letters

Evaluating Meta-Feature Selection for the Algorithm Recommendation Problem

Authors: Geand Trindade Pereira, Moises Rocha dos Santos, Andre Carlos Ponce de Leon Ferreira de Carvalho

Abstract: With the popularity of Machine Learning (ML) solutions, algorithms and data have been released faster than the capacity of processing them. In this context, the problem of Algorithm Recommendation (AR) is receiving a significant deal of attention recently. This problem has been addressed in the literature as a learning task, often as a Meta-Learning problem where the aim is to recommend the best a… ▽ More With the popularity of Machine Learning (ML) solutions, algorithms and data have been released faster than the capacity of processing them. In this context, the problem of Algorithm Recommendation (AR) is receiving a significant deal of attention recently. This problem has been addressed in the literature as a learning task, often as a Meta-Learning problem where the aim is to recommend the best alternative for a specific dataset. For such, datasets encoded by meta-features are explored by ML algorithms that try to learn the map** between meta-representations and the best technique to be used. One of the challenges for the successful use of ML is to define which features are the most valuable for a specific dataset since several meta-features can be used, which increases the meta-feature dimension. This paper presents an empirical analysis of Feature Selection and Feature Extraction in the meta-level for the AR problem. The present study was focused on three criteria: predictive performance, dimensionality reduction, and pipeline runtime. As we verified, applying Dimensionality Reduction (DR) methods did not improve predictive performances in general. However, DR solutions reduced about 80% of the meta-features, obtaining pretty much the same performance as the original setup but with lower runtimes. The only exception was PCA, which presented about the same runtime as the original meta-features. Experimental results also showed that various datasets have many non-informative meta-features and that it is possible to obtain high predictive performance using around 20% of the original meta-features. Therefore, due to their natural trend for high dimensionality, DR methods should be used for Meta-Feature Selection and Meta-Feature Extraction. △ Less

Submitted 11 June, 2021; v1 submitted 7 June, 2021; originally announced June 2021.

The gravitational bending angle by static and spherically symmetric black holes in bumblebee gravity

Authors: I. D. D. Carvalho, G. Alencar, W. M. Mendes, R. R. Landim

Abstract: This work investigates the influence of the Lorentz symmetry breaking in the bending angle of massive particles and light for bumblebee black hole solutions. The solutions analyzed break the Lorentz symmetry due to a non-zero vacuum expectation value of the bumblebee field. We use the Ishihara method, which allows us to study the bending angle of light for finite distances, and it is applicable to… ▽ More This work investigates the influence of the Lorentz symmetry breaking in the bending angle of massive particles and light for bumblebee black hole solutions. The solutions analyzed break the Lorentz symmetry due to a non-zero vacuum expectation value of the bumblebee field. We use the Ishihara method, which allows us to study the bending angle of light for finite distances, and it is applicable to non-asymptotically flat spacetimes when considering the receiver viewpoint. In order to analyze the deflection of massive particles, we systematize the Ishihara method for its application in the Jacobi metric. This systematization allows the study of the deflection angle of massive particles using the Gauss-Bonnet theorem. We consider two backgrounds: the first was found by Bertolami et al. and is asymptotically flat. The second was found recently by Maluf et al. and is not asymptotically flat due to an effective cosmological constant. △ Less

Submitted 17 May, 2021; v1 submitted 5 March, 2021; originally announced March 2021.

Comments: 7 pages, 1 figure, Minor corrections, typos fixed, references added

Fuzzy clustering algorithms with distance metric learning and entropy regularization

Authors: Sara Ines Rizo Rodriguez, Francisco de Assis Tenorio de Carvalho

Abstract: The clustering methods have been used in a variety of fields such as image processing, data mining, pattern recognition, and statistical analysis. Generally, the clustering algorithms consider all variables equally relevant or not correlated for the clustering task. Nevertheless, in real situations, some variables can be correlated or may be more or less relevant or even irrelevant for this task.… ▽ More The clustering methods have been used in a variety of fields such as image processing, data mining, pattern recognition, and statistical analysis. Generally, the clustering algorithms consider all variables equally relevant or not correlated for the clustering task. Nevertheless, in real situations, some variables can be correlated or may be more or less relevant or even irrelevant for this task. This paper proposes partitioning fuzzy clustering algorithms based on Euclidean, City-block and Mahalanobis distances and entropy regularization. These methods are an iterative three steps algorithms which provide a fuzzy partition, a representative for each fuzzy cluster, and the relevance weight of the variables or their correlation by minimizing a suitable objective function. Several experiments on synthetic and real datasets, including its application to noisy image texture segmentation, demonstrate the usefulness of these adaptive clustering methods. △ Less

Submitted 18 February, 2021; originally announced February 2021.

CHARET: Character-centered Approach to Emotion Tracking in Stories

Authors: Diogo S. Carvalho, Joana Campos, Manuel Guimarães, Ana Antunes, João Dias, Pedro A. Santos

Abstract: Autonomous agents that can engage in social interactions witha human is the ultimate goal of a myriad of applications. A keychallenge in the design of these applications is to define the socialbehavior of the agent, which requires extensive content creation.In this research, we explore how we can leverage current state-of-the-art tools to make inferences about the emotional state ofa character in… ▽ More Autonomous agents that can engage in social interactions witha human is the ultimate goal of a myriad of applications. A keychallenge in the design of these applications is to define the socialbehavior of the agent, which requires extensive content creation.In this research, we explore how we can leverage current state-of-the-art tools to make inferences about the emotional state ofa character in a story as events unfold, in a coherent way. Wepropose a character role-labelling approach to emotion tracking thataccounts for the semantics of emotions. We show that by identifyingactors and objects of events and considering the emotional stateof the characters, we can achieve better performance in this task,when compared to end-to-end approaches. △ Less

Submitted 19 July, 2021; v1 submitted 15 February, 2021; originally announced February 2021.

Comments: Preprint

The Surface response around a sharply resonant surface polariton mode is simply a Lorentzian

Authors: J. C. de Aquino Carvalho, D. Bloch

Abstract: At the planar interface between a material and vacuum, the complex surface response S(omega)=[eps(omega) 1]/[eps(omega)+1], with eps(omega) the relative complex dielectric permittivity of the material, exhibit resonances, typical of the surface polariton modes, when eps(omega) ~ 1. We show that for a moderately sharp resonance, S(omega) is satisfactorily described with a mere (complex) Lorentzian,… ▽ More At the planar interface between a material and vacuum, the complex surface response S(omega)=[eps(omega) 1]/[eps(omega)+1], with eps(omega) the relative complex dielectric permittivity of the material, exhibit resonances, typical of the surface polariton modes, when eps(omega) ~ 1. We show that for a moderately sharp resonance, S(omega) is satisfactorily described with a mere (complex) Lorentzian, independently of the details affecting the various bulk resonances describing ) eps(omega). Remarkably, this implies a quantitative correlation between the resonant behaviors of Re[S(omega)] and Im[S(omega)], respectively associated to dispersive and dissipative effects in the surface near-field. We show that this "strong resonance" approximation easily applies, and discuss its limits, based upon published data for sapphire, CaF2 and BaF2. Extension to interfaces between two media or to a non planar interface is briefly considered. △ Less

Submitted 26 May, 2021; v1 submitted 13 February, 2021; originally announced February 2021.

Comments: accepted for publication in Optics Letters

Velocity preserving transfer between highly excited atomic states: Black Body Radiation and Collisions

Authors: J. C. de Aquino Carvalho, I. Maurin, H. Failache, D. Bloch, A. Laliotis

Abstract: We study the excitation redistribution from cesium $7\mathrm{P}_{1/2}$ or $7\mathrm{P}_{3/2}$ to neighboring energy levels by Black Body Radiation (BBR) and inter atomic collisions using pump-probe spectroscopy inside a vapor cell. At low vapor densities we measure redistribution of the initial, velocity-selected, atomic excitation by BBR. This preserves the selected atomic velocities allowing us… ▽ More We study the excitation redistribution from cesium $7\mathrm{P}_{1/2}$ or $7\mathrm{P}_{3/2}$ to neighboring energy levels by Black Body Radiation (BBR) and inter atomic collisions using pump-probe spectroscopy inside a vapor cell. At low vapor densities we measure redistribution of the initial, velocity-selected, atomic excitation by BBR. This preserves the selected atomic velocities allowing us to perform high resolution spectroscopy of the $\mathrm{6D\rightarrow 7F}$ transitions. This transfer mechanism could also be used to perform sub-Doppler spectroscopy of the cesium highly-excited $\mathrm{nG}$ levels. At high densities we observe interatomic collisions redistributing the excitation within the cesium $\mathrm{7P}$ fine and hyperfine structure. We show that $\mathrm{7P}$ redistribution involves state-changing collisions that preserve the initial selection of atomic velocities. These redistribution mechanisms can be of importance for experiments probing high lying excited states in dense alkali vapor. △ Less

Submitted 19 January, 2021; originally announced January 2021.

Comments: accepted for publication in Journal of Physics B: Atomic, Molecular and Optical Physics