Pasar al contenido principal

CIENCIAS DE LA COMPLEJIDAD PARA LA INTELIGENCIA ARTIFICIAL: DE LA DINAMICA REDES COMPLEJAS A LOS FENOMENOS COLECTIVOS DE REDES NEURONALES EN INTERACCION

PID2024-157526NB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para la Investigación y el Desarrollo Experimental
Subprograma Subprograma Estatal de Generación de Conocimiento Científico-Técnico y Desarrollo Experimental
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2024
Plan Estatal 2024-2027 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario AGENCIA ESTATAL CONSEJO SUPERIOR DE INVESTIGACIONES CIENTIFICAS (CSIC)

Resultados relacionados

Resultados totales (Incluyendo duplicados): 11
Encontrada(s) 1 página(s)

DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Transfer learning-enhanced deep reinforcement learning for aerodynamic airfoil optimization subject to structural constraints

  • Ramos, David
  • Lacasa, Lucas
  • Valero, Eusebio
  • Rubio, Gonzalo
© 2025 Author(s). Published under an exclusive license by AIP Publishing., The main objective of this paper is to introduce a transfer learning-enhanced deep reinforcement learning (DRL) methodology that is able to optimize the geometry of any airfoil based on concomitant aerodynamic and structural integrity criteria. To showcase the method, we aim to maximize the lift-to-drag ratio C L / C D while preserving the structural integrity of the airfoil—as modeled by its maximum thickness—and train the DRL agent using a list of different transfer learning (TL) strategies. The performance of the DRL agent is compared with Particle Swarm Optimization (PSO), a traditional gradient-free optimization method. Results indicate that DRL agents are able to perform purely aerodynamic and hybrid aerodynamic/structural shape optimization that the DRL approach outperforms PSO in terms of computational efficiency and aerodynamic improvement and that the TL-enhanced DRL agent achieves performance comparable to the DRL one, while further saving substantial computational resources., The authors thank useful discussions within the NUMATH group. D.R., E.V., and G.R. acknowledge funding from Project TIFON (No. PLEC2023-010251) funded by MCIN/AEI/10.13039/501100011033, Spain, funding from the European Union (Project HERFUSE) under GA No. 101140567. Views and opinions expressed are those of the authors only and do not necessarily reflect those of the European Union or Clean Aviation Joint Undertaking. Neither the European Union nor Clean Aviation JU can be held responsible for them. L.L. acknowledges partial financial support from via Project MISLAND (No. PID2020-114324GB-C22), CSxAI (PID2024-157526NB-I00) and the Maria de Maeztu Program for units of Excellence (Grant No. CEX2021-001164-M), the three of them funded by MICIU/AEI/10.13039/501100011033, and funding from the European Commission Chips Joint Undertaking Project No. 101194363 (NEHIL). G.R. acknowledges partial financial support received by the Grant DeepCFD (Project No. PID2022-137899OB-I00) funded by MICIU/AEI/10.13039/501100011033 and by ERDF, EU. Finally, all authors gratefully acknowledge the Universidad Politécnica de Madrid (www.upm.es) for providing computing resources on Magerit Supercomputer., With funding from the Spanish government through the "María de Maeztu Unit of Excellence" accreditation (CEX2021-001164-M), Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Inferring Leader-Follower Behavior from Presence Data in the Marine Environment: A Case Study on Reef Manta Rays

  • Fernández-Gracia, Juan
  • Rodríguez-García, Jorge Pablo
  • Peel, Lauren R.
  • Klemm, Konstantin
  • Meekan, Mark G.
  • Eguíluz, Víctor M.
Social interactions are fundamental in animal groups, including humans, and can take various forms, such as competition, cooperation, or kinship. Understanding these interactions in marine environments has been historically challenging due to data collection difficulties. However, advancements in acoustic telemetry now enable the remote analysis of such behaviors. This study proposes a method to derive leader-follower networks from presence data collected by a single acoustic receiver at a specific location. Using the Kolmogorov-Smirnov distance, the method analyzes lag times between consecutive presences of individuals to infer directed relationships. Tested on simulated data, it was then applied to detection data from acoustically tagged reef manta rays (Mobula alfredi) frequenting a known site. Results revealed temporal patterns, including circadian rhythms and burst-like behavior with power-law distributed time gaps between presences. The inferred leader-follower network highlighted key behavioral patterns when compared to an appropriate random null model: females followed males more often than expected, males showed stronger but fewer associations with specific females, and smaller individuals were less consistent in following others than larger ones. These findings align with ecological insights, revealing structured social interactions and providing a novel framework for studying marine animal behavior through network theory., VME and JFG acknowledge partial support from projects MISLAND (PID2020-114324GB-C22) and CSxAI (PID2024-157526NB-I00) funded by MICIU/AEI/10.13039/501100011033/FEDER, UE. JPR was supported by Govern de les Illes Balears through the Vicenç Mut program. Fieldwork and data collection was funded by the Save Our Seas Foundation (SOSF) and supported by The Manta Trust and SOSF – D’Arros Research Centre (SOSF-DRC). Collection of acoustic telemetry data in Seychelles was approved by, and conducted with the knowledge of, the Seychelles Bureau of Standards and the Seychelles Ministry of Environment, Energy and Climate Change., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

A Certifiable Machine Learning-Based Pipeline to Predict Fatigue Life of Aircraft Structures

  • Ladrón, Ángel
  • Sánchez-Domínguez, Miguel
  • Rozalén, Javier
  • Sánchez, Fernando R.
  • Vicente, Javier de
  • Lacasa, Lucas
  • Valero, Eusebio
  • Rubio, Gonzalo
Fatigue life prediction is essential in both the design and operational phases of any aircraft, and in this sense safety in the aerospace industry requires early detection of fatigue cracks to prevent in-flight failures. Robust and precise fatigue life predictors are thus essential to ensure safety. Traditional engineering methods, while reliable, are time consuming and involve complex workflows, including steps such as conducting several Finite Element Method (FEM) simulations, deriving the expected loading spectrum, and applying cycle counting techniques like peak-valley or rainflow counting. These steps often require collaboration between multiple teams and tools, added to the computational time and effort required to achieve fatigue life predictions. Machine learning (ML) offers a promising complement to traditional fatigue life estimation methods, enabling faster iterations and generalization, providing quick estimates that guide decisions alongside conventional simulations. In this paper, we present a ML-based pipeline that aims to estimate the fatigue life of different aircraft wing locations given the flight parameters of the different missions that the aircraft will be operating throughout its operational life. We validate the pipeline in a realistic use case of fatigue life estimation, yielding accurate predictions alongside a thorough statistical validation and uncertainty quantification. Our pipeline constitutes a complement to traditional methodologies by reducing the amount of costly simulations and, thereby, lowering the required computational and human resources., The authors acknowledge funding from project TIFON (PLEC2023-010251) funded by MCIN/AEI/10.13039/501100011033, Spain. LL acknowledges partial support from projects MISLAND (PID2020-114324GB-C22), project CSxAI (PID2024-157526NB-I00) funded by MICIU/AEI/10.13039/501100011033/FEDER, UE, project MdM Seal of Excellence (CEX2021-001164-M) funded by MICIU/AEI/10.13039/501100011033 and from the European Commission Chips Joint Undertaking project No. 101194363 (NEHIL)., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001164-M)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Otros (Other)

A Certifiable Machine Learning-Based Pipeline to Predict Fatigue Life of Aircraft Structures

  • Ladrón, Ángel
  • Sánchez-Domínguez, Miguel
  • Rozalén, Javier
  • Sánchez, Fernando R.
  • Vicente, Javier de
  • Lacasa, Lucas
  • Valero, Eusebio
  • Rubio, Gonzalo
34 pages, 17 figures., Fatigue life prediction is essential in both the design and operational phases of any aircraft, and in this sense safety in the aerospace industry requires early detection of fatigue cracks to prevent in-flight failures. Robust and precise fatigue life predictors are thus essential to ensure safety. Traditional engineering methods, while reliable, are time consuming and involve complex workflows, including steps such as conducting several Finite Element Method (FEM) simulations, deriving the expected loading spectrum, and applying cycle counting techniques like peak-valley or rainflow counting. These steps often require collaboration between multiple teams and tools, added to the computational time and effort required to achieve fatigue life predictions. Machine learning (ML) offers a promising complement to traditional fatigue life estimation methods, enabling faster iterations and generalization, providing quick estimates that guide decisions alongside conventional simulations. In this paper, we present a ML-based pipeline that aims to estimate the fatigue life of different aircraft wing locations given the flight parameters of the different missions that the aircraft will be operating throughout its operational life. We validate the pipeline in a realistic use case of fatigue life estimation, yielding accurate predictions alongside a thorough statistical validation and uncertainty quantification. Our pipeline constitutes a complement to traditional methodologies by reducing the amount of costly simulations and, thereby, lowering the required computational and human resources., The authors acknowledge funding from project TIFON (PLEC2023-010251) funded by MCIN/AEI/10.13039/501100011033, Spain. LL acknowledges partial support from projects MISLAND (PID2020-114324GB-C22), project CSxAI (PID2024-157526NB-I00) funded by MICIU/AEI/10.13039/501100011033/FEDER, UE, project MdM Seal of Excellence (CEX2021-001164-M) funded by MICIU/AEI/10.13039/501100011033 and from the European Commission Chips Joint Undertaking project No. 101194363 (NEHIL)., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001164-M)., No
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Improving prediction region accuracy in marine animal movement with temporal fusion transformer

  • Medina Hernández, Jorge
  • Rodríguez-García, Jorge Pablo
  • McMahon, Clive R.
  • Sequeira, Ana M. M.
  • Eguíluz, Víctor M.
Predicting marine animal movements from satellite tracking data remains challenging, limiting conservation and ecosystem management efforts. To address this, we trained the Temporal Fusion Transformer (TFT) neural network on tracking data from 434 southern elephant seals to forecast locations and fill data gaps (imputation) within 7-day windows. Compared to state-space models, TFT reduced location errors by 15% and produced more efficient prediction regions, identifying where seals were likely to be found while using less area: a fivefold reduction for forecasting and 30-40% reduction for imputation. The model performed best near the continental shelf and at low-to-moderate movement speeds, with bathymetry, water temperature and current direction being the most influential environmental factors affecting the model output. When applied to new geographic regions not represented in the training dataset, model performance declined by approximately 30% across most evaluation metrics, indicating challenges in transferring learned patterns to unfamiliar environments. Our findings show that deep learning is a promising tool for analyzing large, sparse tracking datasets. The enhanced predictive capabilities have potential for dynamic conservation measures, such as forecasting the spatial evolution of animals to minimize conflicts with human activities and environmental disturbances., This research was funded by the Sustainable Ocean Alliance for the project Artificial Intelligence & Animal Movement (AIAM, ref. D017) project under the SOA Grants program. J.P.R. acknowledges funding from the Vicenç Mut program of Govern de les Illes Balears. This research was supported by Ministerio de Ciencia e Innovacion, Agencia Estatal de Investigación through the Maria de Maeztu Units of Excellences Program (project CEX2021-001164-M) funded by MICIU/AEI/10.13039/501100011033, and by project CSxAI (PID2024-157526NB-I00), funded by MICIU/AEI/10.13039/501100011033/FEDER, UE, and by an ARC DP210103091 (awarded to A.M.M.S. and V.M.E.). J.M.H. acknowledges financial support from MDM-2017-0711-19-1, funded by MICIU/AEI/10.13039/501100011033 and by FSE invierte en tu futuro. Data collection was supported by the French Polar Institute (program 109: PI. H. Weimerskirch and 1201: PI. C. Gilbert and C. Guinet), the SNO-MEMO and CNES-TOSCA, in collaboration with The Integrated Marine Observing System (IMOS) with data sourced from Australia’s Integrated Marine Observing System (IMOS) – IMOS is enabled by the National Collaborative Research Infrastructure Strategy (NCRIS)., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001164-M)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Predictability of temporal network dynamics in normal ageing and brain pathology

  • Caligiuri, Annalisa
  • Papo, David
  • Yener, Görsev
  • Güntekin, Bahar
  • Galla, Tobias
  • Lacasa, Lucas
  • Zanin, Massimiliano
Spontaneous brain activity generically displays transient spatiotemporal coherent structures, which can selectively be affected in various neurological and psychiatric pathologies. Here we model the full brain’s electroencephalographic activity as a high-dimensional functional network performing a trajectory in a latent graph phase space. This approach allows us to investigate the orbital stability of brain’s activity and in particular its short-term predictability. We do this by constructing a non-parametric statistic quantifying the expansion of initially close functional network trajectories. We apply the method to cohorts of healthy ageing individuals, and patients previously diagnosed with Parkinson’s or Alzheimer’s disease. Results not only characterize brain dynamics from a new angle, but further show that functional network predictability varies in a marked scale-dependent way across healthy controls and patient groups. The path towards both pathologies is markedly different. Furthermore, healthy ageing’s predictability appears to strongly differ from that of Parkinson’s disease, but much less from that of patients with Alzheimer’s disease., This work was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme [851255]. The authors further acknowledge partial support from projects MISLAND [PID2020-114324GB-C22], CSxAI [PID2024-157526NB-I00] MICIU/AEI/10.13039/501100011033/FEDER, UE and the María de Maeztu project CEX2021-001164-M, funded by MICIU/AEI/10.13039/501100011033. AC acknowledges funding by the Maria de Maeztu Programme [MDM-2017-0711] and the AEI [MICIU/AEI/10.13039/501100011033] under the FPI programme., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001164-M)., No
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Otros (Other)

Predictability of temporal network dynamics in normal ageing and brain pathology

  • Caligiuri, Annalisa
  • Papo, David
  • Yener, Görsev
  • Güntekin, Bahar
  • Galla, Tobias
  • Lacasa, Lucas
  • Zanin, Massimiliano
24 pages., Spontaneous brain activity generically displays transient spatiotemporal coherent structures, which can selectively be affected in various neurological and psychiatric pathologies. Here we model the full brain's electroencephalographic activity as a high-dimensional functional network performing a trajectory in a latent graph phase space. This approach allows us to investigate the orbital stability of brain's activity and in particular its short-term predictability. We do this by constructing a non-parametric statistic quantifying the expansion of initially close functional network trajectories. We apply the method to cohorts of healthy ageing individuals, and patients previously diagnosed with Parkinson's or Alzheimer's disease. Results not only characterise brain dynamics from a new angle, but further show that functional network predictability varies in a marked scale-dependent way across healthy controls and patient groups. The path towards both pathologies is markedly different. Furthermore, healthy ageing's predictability appears to strongly differ from that of Parkinson's disease, but much less from that of patients with Alzheimer's disease., This work was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme [851255]. The authors further acknowledge partial support from projects MISLAND [PID2020-114324GB-C22], CSxAI [PID2024-157526NB-I00] MICIU/AEI/10.13039/501100011033/FEDER, UE and the María de Maeztu project CEX2021-001164-M, funded by MICIU/AEI/10.13039/501100011033. AC acknowledges funding by the Maria de Maeztu Programme [MDM-2017-0711] and the AEI [MICIU/AEI/10.13039/501100011033] under the FPI programme., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001164-M)., No
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Leveraging chaotic transients in the training of artificial neural networks

  • Jiménez-González, Pedro
  • Soriano, Miguel C.
  • Lacasa, Lucas
Traditional algorithms to optimize artificial neural networks when confronted with a supervised learning task are usually exploitation-type relaxational dynamics such as gradient descent (GD). Here, we explore the dynamics of the neural network trajectory along training for unconventionally large learning rates. We show that for a region of values of the learning rate, the GD optimization shifts away from purely exploitationlike algorithm into a regime of exploration-exploitation balance, as the neural network is still capable of learning but the trajectory shows sensitive dependence on initial conditions—as characterized by positive network maximum Lyapunov exponent. Interestingly, the characteristic training time required to reach an acceptable accuracy in the test set reaches a minimum precisely in such learning rate region, further suggesting that one can accelerate the training of artificial neural networks by locating at the onset of chaos. Our results—initially illustrated for the MNIST classification task—qualitatively hold for a range of supervised learning tasks, learning architectures (including both shallow and deep multilayer perceptrons and convolutional neural networks) and other hyperparameters (different activation functions and weight regularization), and showcase the emergent, constructive role of transient chaotic dynamics in the training of artificial neural networks., P.J. acknowledges funding from Maria de Maeztu (MdM) Seal of Excellence (CEX2021-001164-M) via the FPI programme (Grant No. PRE2022-104148), funded by the MICIU/AEI/10.13039/501100011033. L.L. acknowledges partial support from project CSxAI (PID2024-157526NB-I00) funded by MICIU/AEI/10.13039/501100011033/FEDER, UE. M.C.S. and L.L. acknowledge partial support from a Maria de Maeztu grant (CEX2021-001164-M) funded by the MICIU/AEI/10.13039/501100011033 and from the European Commission Chips Joint Undertaking Project No. 101194363 (NEHIL)., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001164-M)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Otros (Other)

On the Role of Consistency Between Physics and Data in Physics-Informed Neural Networks (v1)

  • Becerra-Zuniga, Nicolás
  • Lacasa, Lucas
  • Valero, Eusebio
  • Rubio, Gonzalo
24 pages, 7 Figures, 3 Tables., Physics-informed neural networks (PINNs) have gained significant attention as a surrogate modeling strategy for partial differential equations (PDEs), particularly in regimes where labeled data are scarce and physical constraints can be leveraged to regularize the learning process. In practice, however, PINNs are frequently trained using experimental or numerical data that are not fully consistent with the governing equations due to measurement noise, discretization errors, or modeling assumptions. The implications of such data-to-PDE inconsistencies on the accuracy and convergence of PINNs remain insufficiently understood. In this work, we systematically analyze how data inconsistency fundamentally limits the attainable accuracy of PINNs. We introduce the concept of a consistency barrier, defined as an intrinsic lower bound on the error that arises from mismatches between the fidelity of the data and the exact enforcement of the PDE residual. To isolate and quantify this effect, we consider the 1D viscous Burgers equation with a manufactured analytical solution, which enables full control over data fidelity and residual errors. PINNs are trained using datasets of progressively increasing numerical accuracy, as well as perfectly consistent analytical data. Results show that while the inclusion of the PDE residual allows PINNs to partially mitigate low-fidelity data and recover the dominant physical structure, the training process ultimately saturates at an error level dictated by the data inconsistency. When high-fidelity numerical data are employed, PINN solutions become indistinguishable from those trained on analytical data, indicating that the consistency barrier is effectively removed. These findings clarify the interplay between data quality and physics enforcement in PINNs providing practical guidance for the construction and interpretation of physics-informed surrogate models., The authors acknowledge funding from project TIFON (PLEC2023-010251) funded by MCIN/AEI/10.13039/501100011033, Spain. This research has received funding from the European Union (ROSAS, project number 101138319). This research has received funding from the European Research Council Executive Agency under TRANSDIF-FUSE project, Grant Agreement No. 101167322. Views and opinions expressed are, however, those of the authors only and do not necessarily reflect those of the European Union. Neither the European Union nor the granting authority can be held responsible for them. LL acknowledges partial support from project CSxAI (PID2024-157526NB-I00) funded by MICIU/AEI/10.13039/501100011033/FEDER, UE, from a Maria de Maeztu Seal of Excellence grant (CEX2021-001164-M) funded by the MICIU/AEI/10.13039/501100011033, and from the European Commission Chips Joint Undertaking project No. 101194363 (NEHIL)., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001164-M)., No
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Fluid dynamics reduction methods for temporal networks

  • Lacasa, Lucas
Temporal networks, defined as sequences of time-aggregated adjacency matrices, sample latent graph dynamics and trace trajectories in graph space. By interpreting each adjacency matrix as a different time snapshot of a scalar field, we show how fluid-mechanics methods can be applied to construct two distinct eigendecompositions of temporal networks. The first builds on the proper orthogonal decomposition (POD) of flowfields and decomposes the evolution of a network in terms of a basis of orthogonal network eigenmodes which are ordered in terms of their relative importance, hence enabling compression of temporal networks as well as their projection in low-dimensional embeddings. The second proposes a numerical approximation of the Koopman operator, a linear operator acting on a suitable observable of the graph space which provides the best linear approximation of the latent graph dynamics. Its eigendecomposition provides a data-driven spectral description of the temporal network dynamical stability, in terms of dynamic modes which grow, decay or oscillate over time. We illustrate and validate the application of both eigendecompositions in a suite of synthetic generative models of temporal networks with varying complexity., Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. The author acknowledges partial support from project CSxAI (PID2024-157526NB-I00) funded by MICIU/AEI/10.13039/501100011033/FEDER, UE, project Maria de Maeztu CEX2021-001164-M funded by the MICIU/AEI/10.13039/501100011033, and from the European Commission Chips Joint Undertaking project No. 101194363 (NEHIL)., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001164-M)., Peer reviewed
DIGITAL.CSIC. Repositorio Institucional del CSIC
Publicación Artículo científico (JournalArticle)

Transitions between discrete developmental stages during language acquisition through syntax network morphospace

  • Barceló-Coblijn, Lluís
  • Eguíluz, Víctor M.
  • Seoane Iglesias, Luis
Syntax is an aspect of human language that sorts linguistic structures hierarchically. It is summarized by dependency trees with words as nodes and edges reflecting syntactic subordination. By merging trees from several sentences, we obtain syntax graphs or networks, which display distinct shapes depending on whether the language capability is well formed, still developing, or pathological. Such graphs make syntactic capacity quantifiable at a systemic level. What is the structure of syntax networks during ontogeny in typically developing children? Do cognitively challenged children develop language through alternative routes? We quantify and portray typical development of syntax networks in Dutch, and find that children affected by Down syndrome, hearing impairment, and specific language impairment track the typical developmental path but halt along the way. Through a broad dataset and a thorough mathematical quantification we (1) confirm and refine a proposed sharp transition in language development, (2) correlate network traits with syntax maturation, and (3) quantify aspects that fall short in atypical development—suggesting potential diagnostic tools. We find grounds to hypothesize a gap in syntax maturation that typically developing children bridge over seamlessly, but which divides atypically developing children regardless of their condition. Our quantitative analysis enables a rigorous visualization of linguistic development trajectories, an old (yet mostly qualitative) theme in linguistics. Similar works should allow to test specific linguistic concepts (such as development trajectories) on solid quantitative grounds—as done here. Future efforts should generalize to other languages and clinical conditions, seeking patterns and universalities in language development., L.B.-C. was supported by Grant No. PID2021-128404NA-I00 funded by MICIU/AEI/10.13039/50110001103 and by ERDF/EU. V.M.E. received partial support from project CSxAI (PID2024-157526NB-I00) funded by MICIU/AEI/10.13039/501100011033/FEDER, UE. L.F.S. received funding from the Occident Foundation (Grant No. FJSCNB-2022-12-B) and from the Spanish State Research Agency, AEI, through Project No. PID2023-153225NA-I00 funded by MICIU/AEI/10.13039/501100011033 and ERDF, EU. This work was started at IFISC, where LSF was benefiting from a fellowship at the María de Maeztu Program for Units of Excellence in R&D (MDM-2017-0711)., Peer reviewed