Search

The Lattice Boltzmann Method (LBM) is a computational technique of Computational Fluid Dynamics (CFD) that has gained popularity due to its high parallelism and ability to handle complex geometries with minimal effort. Although LBM frameworks are increasingly important in various industries and research fields, their complexity makes them difficult to modify and can lead to suboptimal performance. This paper presents miniLB, the first, to the best of our knowledge, SYCL-based LBM mini-app.miniLB addresses the need for a performance-portable LBM proxy app capable of abstracting complex fluid dynamics simulations across heterogeneous computing systems. We analyze SYCL semantics for performance portability and evaluate miniLB on multiple GPU architectures using various SYCL implementations. Our results, compared against a manually-tuned FORTRAN version, demonstrate effectiveness of miniLB in assessing LBM performance across diverse hardware, offering valuable insights for optimizing large-scale LBM frameworks in modern computing environments., Comment: 12 pages, conference

We investigate the synchronization behavior and the emergence of chimera states in a system of two interacting populations of maps possessing chaotic neural-like dynamics. We characterize four collective states on the space of coupling parameters of the system: complete synchronization, generalized synchronization, chimera states, and incoherence. We quantify the information exchange between the two neuron populations in chimera states. We have found a well-defined direction of the flow of information in chimera states, from the desynchronized population to the synchronized one. The incoherent population functions as a driver of the coherent neuron population in a chimera state. This feature is independent of the population sizes or population partitions. Our results yield insight into the communication mechanisms arising in brain processes such as unihemispheric sleep and epileptic seizures that have been associated to chimera states., Comment: 8 pages, 8 figures

Large Language Models (LLMs) have revolutionized artificial intelligence, demonstrating remarkable computational power and linguistic capabilities. However, these models are inherently prone to various biases stemming from their training data. These include selection, linguistic, and confirmation biases, along with common stereotypes related to gender, ethnicity, sexual orientation, religion, socioeconomic status, disability, and age. This study explores the presence of these biases within the responses given by the most recent LLMs, analyzing the impact on their fairness and reliability. We also investigate how known prompt engineering techniques can be exploited to effectively reveal hidden biases of LLMs, testing their adversarial robustness against jailbreak prompts specially crafted for bias elicitation. Extensive experiments are conducted using the most widespread LLMs at different scales, confirming that LLMs can still be manipulated to produce biased or inappropriate responses, despite their advanced capabilities and sophisticated alignment processes. Our findings underscore the importance of enhancing mitigation techniques to address these safety issues, toward a more sustainable and inclusive artificial intelligence.

We investigate the collective behavior of a system of social agents subject to the competition between two mass media influences considered as external fields. We study under what conditions either of two mass media with different intensities can impose its message to the majority. In addition to a collective state dominated by the stronger mass media and a disordered phase, we characterize two nontrivial effects as the parameters of the system are varied: (i) the appearance of a majority sharing the state of the weaker mass media, and (ii) the emergence of an alternative ordering in a state different from those of either media. We explore the dependence of both phenomena on the topology of the network of interactions. We show that the presence of long-range interactions rather than random connections is essential for the occurrence of both effects. The model can be extended to include multiple mass media and we illustrate it by considering three mass media fields acting on the system. Nontrivial collective behaviors persist for some ranges of parameters: the weakest mass media can convince the majority, and the system can spontaneously order against all applied fields., Comment: 8 pages, 15 figures, sent to Physical Review E

Independent retrospective analyses of the effectiveness of reducing deforestation and forest degradation (REDD) projects are vital to ensure climate change benefits are being delivered. A recent study in Science by West et al. (1) appeared therefore to be a timely alert that the majority of projects operating in the 2010s failed to reduce deforestation rates. Unfortunately, their analysis suffered from major flaws in the choice of underlying data, resulting in poorly matched and unstable counterfactual scenarios. These were compounded by calculation errors, biasing the study against finding that projects significantly reduced deforestation. This flawed analysis of 24 projects unfairly condemned all 100+ REDD projects, and risks cutting off finance for protecting vulnerable tropical forests from destruction at a time when funding needs to grow rapidly.

Graphs can be leveraged to model polyphonic multitrack symbolic music, where notes, chords and entire sections may be linked at different levels of the musical hierarchy by tonal and rhythmic relationships. Nonetheless, there is a lack of works that consider graph representations in the context of deep learning systems for music generation. This paper bridges this gap by introducing a novel graph representation for music and a deep Variational Autoencoder that generates the structure and the content of musical graphs separately, one after the other, with a hierarchical architecture that matches the structural priors of music. By separating the structure and content of musical graphs, it is possible to condition generation by specifying which instruments are played at certain times. This opens the door to a new form of human-computer interaction in the context of music co-creation. After training the model on existing MIDI datasets, the experiments show that the model is able to generate appealing short and long musical sequences and to realistically interpolate between them, producing music that is tonally and rhythmically consistent. Finally, the visualization of the embeddings shows that the model is able to organize its latent space in accordance with known musical concepts.

Today digital revolution is having a dramatic impact on the pharmaceutical industry and the entire healthcare system. The implementation of machine learning, extreme-scale computer simulations, and big data analytics in the drug design and development process offers an excellent opportunity to lower the risk of investment and reduce the time to the patient. Within the LIGATE project, we aim to integrate, extend, and co-design best-in-class European components to design Computer-Aided Drug Design (CADD) solutions exploiting today's high-end supercomputers and tomorrow's Exascale resources, fostering European competitiveness in the field. The proposed LIGATE solution is a fully integrated workflow that enables to deliver the result of a virtual screening campaign for drug discovery with the highest speed along with the highest accuracy. The full automation of the solution and the possibility to run it on multiple supercomputing centers at once permit to run an extreme scale in silico drug discovery campaign in few days to respond promptly for example to a worldwide pandemic crisis., Comment: Paper Accepted to the 20th ACM International Conference on Computing Frontiers (CF'23)

We investigate an agent-based model for the emergence of corruption in public contracts. There are two types of agents: business people and public servants. Both business people and public servants can adopt two strategies: corrupt or honest behavior. Interactions between business people and public servants take place through defined payoff rules. Either type of agent can switch between corrupt or honest strategies by comparing their payoffs after interacting. We measure the level of corruption in the system by the fractions of corrupt and honest agents for asymptotic times. We study the effects of the group size of the interacting agents, the dispersion with respect to the average salary of the public servants, and a parameter representing the institutional control of corruption. We characterize the fractions of honest and corrupt agents as functions of these variables. We construct phase diagrams for the level of corruption in the system in terms of these variables, where three collective states can be distinguished: i) a phase where corruption dominates; ii) a phase where corruption remains in less than $50\%$ of the agents; and iii) a phase where corruption disappear. Our results indicate that a combination of large group sizes of interacting servants and business people and small dispersion of the salaries of public servants, contributes to the decrease of systemic corruption in public contracts., Comment: Submitted to Heliyon Journal

Virtual screening is an early stage of the drug discovery process that selects the most promising candidates. In the urgent computing scenario it is critical to find a solution in a short time frame. In this paper, we focus on a real-world virtual screening application to evaluate out-of-kernel optimizations, that consider input and architecture features to improve the computation efficiency on GPU. Experiment results on a modern supercomputer node show that we can almost double the performance. Moreover, we implemented the optimization using SYCL and it provides a consistent benefit with the CUDA optimization. A virtual screening campaign can use this gain in performance to increase the number of evaluated candidates, improving the probability of finding a drug.

We investigate the existence of solutions to viscous ergodic Mean Field Games systems in bounded domains with Neumann boundary conditions and local, possibly aggregative couplings. In particular we exploit the associated variational structure and search for constrained minimizers of a suitable functional. Depending on the growth of the coupling, we detect the existence of global minimizers in the mass subcritical and critical case, and of local minimizers in the mass supercritical case, notably up to the Sobolev critical case., Comment: 23 pages