Inria Awards 2012
Paul-Louis George : Inria - Dassault Systèmes 2012 Innovation Award winner
© Inria / Photo Christian Tourniaire
Team leader of the Gamma3 project team (Inria, Troyes University of Technology), Paul Louis George is one of the inventors of the GHS3D volume mesh, a software used throughout the world by players in the industry, researchers and academics. Integrated in several software for 3D calculations by finished elements, GHS3D helps obtain simulations which are particularly reliable and high performing. A success based on teamwork undertaken over the long term and which is now being rewarded with the innovation award.
"When I took up the subject during the second half of the 1980s, meshing was not fashionable, and very few people were interested in it, especially in France," Paul Louis George recollects. But the conditions for making it a promising subject were all there: industry players and software publishers were in great need of solutions that would enable 3D simulation, for example, the behaviour of structures under constraints, and the evolution of the abilities of machines making it possible to tackle problems in 3 dimensions. So Paul Louis George became interested in meshing so as not to abandon this subject which turned out to be very rich in scientific problems which are constantly renewed and fed by the needs of players in the industry and by revolutions in information technology. "This field is located at the intersection between the continuous model (mathematical equation) and the discrete model which is its translation into a form which can be understood by the computer." Hence, it is an essential step in simulation. Everything is part of the art of correctly dividing space into elements (into tetrahedrons here), and reasoning over these pieces in order to be able to make calculations.
Paul Louis George has unquestionably become a master in the field. The GHS3D software which he developed with Frédéric Hecht and Eric Saltel, represents a great advance in the industrial exploitation of numeric 3D simulation and has become a reference in the field. Distributed by Distène, which originated from an Inria start-up, it is now used by about a hundred publishers of simulation software throughout the world (Dassault Systèmes, Siemens, ANSYS, Autodesk, etc.) and is used by large industrial companies such as EDF, Safran, Alcan, etc., who praise its quality, reliability and speed.
This exemplary transfer is the fruit of twenty years' teamwork under the conditions that Paul Louis George considers ideal: “We were able to devote ourselves to research completely by exploiting our complementary skills. …And there was very little competition! “And with good reason! The subject is complex and arduous, for mathematical and technical reasons as well as those related to computing, and very few researchers tackled it! Paul Louis George and his colleagues have overcome these obstacles, armed with discipline and pragmatism and guided by a strong ambition. “Our approach is supported by a certain number of obsessions which characterise us and our strength, explains Paul Louis George. In particular, we do not handle the various angles of the problem separately; we handle them simultaneously so that the result will be coherent. This requires a turn of mind which makes it possible to go beyond theory when necessary.” Another obsession of the researcher is to always reduce the calculation time more, which gives the team a good head start in relation to others. A third obsession is tackling complex subjects, like those found in fluid mechanics concerning aeroplanes, thanks to data from the industry. "We have always had contacts with players in the industry, especially with Dassault Aviation which had a high demand and with which the team has built a real collaboration, says Paul Louis George. But I have very few industrial contracts, after all is said and done. My aim is to develop generic solutions. The subject should constitute a real challenge. This is what is satisfying: solving problems which were thought incapable of solution only a few years ago.”
A fourth obsession which contributes to the success of the software from the team is that the numeric experiments required for their validation are very advanced: upwards, by using several billion elements, and downwards, by increasing or reducing the size of elements, for example. “It's very instructive because in this way you come across quite astonishing anomalies in the behaviour of algorithms which enable you to increase the knowledge required to best solve problems.”
Due to these ambitious viewpoints and accumulated know-how, the team is now in the lead worldwide with respect to subjects like anisotropic meshing for which he developed original theories allowing for industrial exploitation. The researcher also contributes to the dissemination of his work. Although he sometimes gives lectures, most of all Paul Louis George likes to write. So "Mesh generation" is THE benchmark in the field: "The credit isn't mine. I write easily and nobody else did it before me,” he apologises.
Furthermore, Paul Louis George deplores the fact that it has always been difficult to consider meshing worthy of interest and that it figures so little in university degree courses, although that is changing. But his main preoccupation is to ensure the transfer of the knowledge and skills required to be able to offer long-lasting software. Since the departure of his acolytes from the GHS trio, he reconstituted a small circle around GHS3D with Houman Borouchaki, a lecturer at the Troyes University of Technology who has been contributing to the software for a long time, and Adrien Loseille and Frédéric Alauzet, two young researchers who enrich the skills of the team in fluid mechanics and can take over from them. And what about the Inria prize? "I wouldn't like it to overshadow the contribution of the other members of the team."
Bruno Stoufflet, director of forecasting and scientific strategy at Dassault Aviation
"I first met Paul-Louis George when I was a Ph. D student at Inria in 1982-83. He's a little disconcerting when you first meet him because of his caustic and bohemian side, but working together is very nice: relations are simple and marked by very great professionalism. The collaboration we started several years ago between the team of P.-L. George and Dassault Aviation is based on trust and time, which is very nice. A researcher in his soul, he is really keen on the development of the best software of its category at a global level and he was able to infuse a real team spirit in his collaborators, something which is not common. The meshing tools developed in this way are now integrated in the industrial design production line for the shapes of our aircraft".
Inria and Dassault Systèmes created an award for innovation dedicated to computer science and information technology. Dominique Florack, Senior Executive Vice-President, Products-Research & Development of Dassault Systèmes, explains how it came about and the selection of the 2012 winner.
Why was Paul-Louis George chosen as the 2012 winner?
Firstly for the field, that of simulation, which has become crucial for companies. Indeed, it makes it possible to test the applications of a product virtually and thus to benefit from a return on experience several years before the product exists! Nowadays, this approach is indispensable for ensuring the relevance of a
product and remaining competitive.
Next, for the direct contribution of Paul-Louis George to the step enabling the adaptation of mathematical modelling to numeric means, with a meshing technology which is the best in the world in my opinion. It’s very good quality allows for realistic calculations and its performance level is unequalled. This researcher has also effected an exceptional transfer, because I estimate that approximately half of the players in the field use his technology.
In your opinion, what are the conditions for the success of a public - private research collaboration?
We have a lot of experience in research collaborations with several public bodies. From our point of view, the conditions for success are present if a project is completely formulated from both a scientific and an industrial viewpoint. By entrusting its distribution to Distène, Paul-Louis George was able to focus on the core of this particularly long and difficult research. As he said himself: "do not invent questions which are not asked, conclude the research and do not be content with one publication"
- 1980 : Paul Louis George writes a thesis in applied mathematics at Pierre and Marie Curie University on the subject of the resolution of parabolic problems.
- 1981 : He joins Inria Paris-Rocquencourt in the Modulef project team where he contributes to the development of a library of programs of finished elements before turning to problems of making of geometric objects discrete, to which he has dedicated his career.
- 1988 : He brings out one of the first automatic 3D meshes with his colleagues.
- 1996 : He leads the Gamma team (later Gamma3) which has acquired international renown in the field.
2012 Prize Winners
© Inria / Photo Christian TourniaireInria Awards 2012 Paul-Louis George : Inria - Dassault Systèmes 2012 Innovation Award winner
Francis Bach - © Inria / Photo J.M. RamèsInria Awards 2012 Young researcher Inria award: Francis Bach
© Inria / Photo Candice BacheletInria Awards 2012 David Margery: Support for research and innovation Award
© Collège de France / Photo Patrick ImbertInria Awards 2012 Pierre-Louis Lions : Grand Prize Inria