Inria-Brazil: Inria and the LNCC strengthen their partnership

Inria’s researchers have historically had close ties with Brazil through the LNCC (Laboratório Nacional de Computação Científica - the National Laboratory for Scientific Computing), Brazil’s leading body for research into computer science. These ties were strengthened in 2020 through the signing of a memorandum of understanding aimed at promoting scientific cooperation between the two institutes. "The aim of the Inria-Brazil partnership is to promote research excellence on topics of shared interest, particularly in the fields of high performance computing (HPC), de la data science, de l’artificial intelligence et du computational science", explains Patrick Valduriez, scientific director of Inria-Brazil, emeritus researcher at Inria and former head of the Zenith project team at the Côte d'Azur University Inria centre. He has worked regularly with Brazilian researchers since the 1990s.  “This partnership is also geared towards innovation with a view towards impacting society and industry”.

The Santos Dumont super computer (named after the Brazilian aviation pioneer) at the LNCC, which was built by Atos, is the most powerful computer in academia in Latin America. Located in Petrópolis, its total processing capacity is somewhere in the region of 5.1 quadrillion operations a second.

A new strategic agreement was signed at the Inria-Brazil Workshop (IBW), the first in-person conference staged by Inria-Brazil which ran from 10 to 14 April 2023 in São Paulo in partnership with the University of São Paulo (USP), and in Petrópolis in Rio de Janeiro state in partnership with the LNCC and Rio de Janeiro Federal University (UFRJ).

And he continues: "This includes Embrapa, a major stakeholder in agricultural research in Brazil, and the INRAE (France’s national institute for agronomic research), the IRD (France’s development research institute) and the CIRAD (the Centre for International Cooperation in Agronomic Research for Development) in France. Digital agriculture is a very important subject for both our countries. Brazil has set a target of being able to feed 40% of the world’s population within the next ten years."

Inria-Brazil, which has around twenty or so collaborative projects involving around twenty Inria project teams and the same number of teams from Brazil (around 200 researchers and PhD students in France and Brazil), is now exploring the possibility of further collaborations on subjects specific to Brazil alongside French groups with significant research and development activity in Brazil and who have signed framework agreements with Inria. Examples include Atos, Total Energies and Stellantis. Researchers from these companies also attended the Inria-Brazil workshop in April, alongside around a hundred or so other participants.

© Inria Brasil

Inria-Brazil Workshop (IBW)

Among the talks and discussions was a presentation of the artificial intelligence research laboratory C4AI, which is jointly funded by IBM and FAPESP (the São Paolo Research Foundation), in addition to a range of sessions on mathematical research applied to epidemiology.

Health, and in particular research into dengue, a disease that is highly prevalent in Brazil, is one of the main shared priorities, and an area of focus for both the INSERM laboratory in France and the Fiocruz laboratory in Brazil. Mostafa Adimy, head of Dracula a project team based in Grenoble (Inria Lyon Centre), presented a report on research that is currently being carried out aimed at modelling the impact of PrEP (a preventive drug for people with high exposure to HIV) on the spread of AIDS, while his Brazilian colleague Suani T. R. Pinho (Bahia Federal University Physics Institute) unveiled the latest discoveries made at his labs linked to the real-time use of epidemiological data for risk modelling.

“One of our success stories, which I presented as part of a talk I gave on 12 April, is Provlake, a data management system with the capacity to effectively query data in distributed workflows and services”, explains Patrick Valduriez. “This system came about as a result of a collaboration with UFRJ and IBM Brazil through a PhD I co-supervised. This led to transfer to IBM Brazil, which then incorporated the software into its service offer in Brazil for its clients in the energy sector.”

Clément Maria's research provides another example of an ongoing collaboration, leading to discussions at the workshop. Clément Maria is a guest researcher at the Getulio Vargas Foundation School of Mathematics (FGV-Emap in Rio de Janeiro) as well as being a member of DataShape, a project team that is a joint undertaking involving the Inria Saclay Centre and the Côte d'Azur University Inria Centre. “Myself and my colleagues from the FGV/EMAp are currently seeking to refine existing methods for extracting more semantic information on the ‘form’ of data, employing methods which are topologically more precise.  We are also in discussions with statisticians from the EMAp on random methods for improving combinatorial algorithms in topology.” 

Each team brings its own strengths to projects, whether it's in terms of researchers or software and equipment. HPDaSc an Inria associate team, jointly managed by Patrick Valduriez and Fabio Porto, at the LNCC, is comprised of a permanent group of 25 researchers on both sides of the Atlantic. “From a research point of view, the challenge lies in combining machine learning with analysis and simulation; analysing and viewing data in real-time; and finding a way of effectively processing complex workflows combining analysis, modelling and simulation”, explains the emeritus researcher. The new methods that are developed are validated using software developed in both countries: OpenAlea and Pl@Ntnet, on the Inria side, and DfAnalyzer and Savime, in Brazil.