Marcelo Viana (IMPA, Brazil) interviews Marcus Sarkis
	Marcus Sarkis at IMPA, Brazil

Brain drain has always plagued the developing world, depriving it of badly needed scientific manpower, as many of their brightest minds head toward wealthier and more developed areas of the globe in search for better salaries and better working environments. While it is unlikely that the situation can be reversed in the near future, some positive signs are appearing as a result of incentives by several governments for the return of their national scientists, together with the emergence of institutes offering highly competitive working conditions, most notably in countries like China, India, and Brazil. In order to begin compiling a profile of those scientists who are coming back, I interviewed Marcus Sarkis, an applied mathematician, who traded a most promising career in the United States for an equally promising and even more challenging career in his native Brazil. Since 2001, Prof. Sarkis is an Associate Professor at the Pure and Applied Mathematics Institute (Instituto de Matemática Pura e Aplicada - IMPA) in Rio de Janeiro. Let his own words explain how and why he got here, and how he feels about it:

MV: Marcus, where are you from and where did you study?

MS: I was born in Rio, but I grew up in several places around Brazil. Eventually, I went to ITA (Brazil's Aeronautics Institute of Technology) in São José dos Campos (in the state of São Paulo), from where I graduated.

MV: At that time, did you feel academically oriented or were you planning to become an engineer?

MS: From the start, I intended to get a PhD degree and get involved in research. The atmosphere at ITA and other institutes in the S. José dos Campos area, like INPE (National Institute for Space Research), really encouraged that.

MV: Did you envisage research in academia, or rather in industry?

MS: I was thinking of working at a place like INPE, possibly in satellite construction, but then came the opportunity to go to Petrobrás (Brazil's national oil company), where I did a course in oil engineering, and actually worked for a few months offshore as an operation leader and production engineer. However, my attraction to mathematics and my will to do research were stronger, and so I decided to enroll as a graduate student at the Catholic University of Rio de Janeiro where I got my Master's degree.

MV: Then it was time to move on to the PhD. Did you plan to go to the USA from the start?

MS: Well, I knew I wanted to go abroad, but France was actually my first choice, because I had made some contacts there, in fact I had visited some institutions there while I was studying for my Master's degree. But I also applied to the USA and somehow the paperwork went faster at the Courant Institute. I was accepted there, got my scholarship from CNPq (Brazil's National Research Agency), and so I decided to go there.

MV: How was your experience at the Courant Institute?

MS: The first year was tough. My English was not so good, and therefore, I felt out of place. In the beginning you tend to interact more with people in the same situation, like Courant colleagues and other NYU students. Also, there was a lot of pressure because of the written and oral examinations. But Courant Institute became the place where I strengthened a great part of my mathematical skills, and also I learned to appreciate the diversity of cultures present in New York City.

MV: Who was your thesis advisor?

MS: Initially I worked with Prof. Robert Kohn and I actually took the oral examination with him. At that point I was focussing on PDEs, weak solutions, homogenization theory, optimal design, areas like that. But I began to be more and more interested in Numerical Analysis, and thus, I changed my advisor to Prof. Olof Widlund, with whom I did my thesis on finite elements theory and who has become a good friend.

MV: What was your thesis about?

MS: I focussed on elliptic problems with highly variable, possibly even discontinuous coefficients. I treated both linear and non-linear problems, but I guess the fundamental part of my thesis concerned the linear case. The problem is that after discretization you get a huge system of equations: you want to have iterative algorithms for solving this system that can handle those discontinuities, such that the iteration time be relatively independent of the coefficients.

MV: In which kind of situations do you find such strongly variable coefficients that motivate this study?

MS: They appear, for instance, in multi-phase flow on porous media with highly heterogeneous permeabilities such as those that appear in petroleum reservoirs. In material science or multi-phase flow associated to different properties (compressibility, densities, viscosities and others) or transmission of waves on stratified and heterogeneous media. For such problems, or in others involving highly different scales, special discretization schemes and fast and robust solvers are required. I believe that due to the recent large interest in MEMS and nanotechnology, and the huge growth of computational power, such techniques will be required more often.

MV: So, you defended your thesis in 1994. Where did you go next?

MS: I got a NSF Post-Doctoral Fellowship, which gave me great freedom to continue doing research. Typically, these fellowships cover half of your salary, the other half coming from the host institution. So I went to the University of Colorado at Boulder where they have a very strong group in Numerical Analysis, both in Mathematics and in Computer Science with applications especially related to aerospace sciences. I got very involved in the research going on there: fluid dynamics, parallel computing, domain decomposition and so on. I could have stayed there for the rest of my life!

Marcus Sarkis, as a postdoc at the University of Colorado at Boulder

MV: Yet, in 2001 you decided to come back to Brazil. Before we get into that, I would like you to tell me about the grant you were awarded in 1999.

MS: After Boulder, I moved to Massachusetts for an appointment as an Assistant Professor at the Worcester Polytechnic Institute. There I got an NSF Career award. These awards are the most prestigious awards in the USA for new faculty members. The award recognizes outstanding new professors who intend to develop academic careers involving both research and education. It supports projects that stimulate the discovery process in which the excitement of research and inspired teaching are integrated within the context of the mission of an institution.

MV: So when you decided to come back, why did you do it? Was it for family reasons; did your wife want to come?

MS: Well, family reasons did have some weight because my mother was not feeling well at the time. But mostly, I wanted to come back and do my share in Brazil. I always felt a moral obligation to contribute. Besides, although I was very well integrated into the research going on in the USA, I also felt that a good part of my effort was, in some sense, being diluted. The problem was that I was training students but they were moving to industry and the commercial job market. These kind of skills, like parallel computation and so on, are in much need in the market and are very well paid. So, I felt I was not creating a scientific family, a school in my area, and I was convinced I could be much more influential here in Brazil. Especially at IMPA, an institution that I know always attracts a large number of excellent students, and then disseminates them across Latin America and other regions.

MV: What about your wife, how did she feel about this move?

MS: When I met my wife Michele she was a PhD student in Hispanic literature and I was a Post-Doctorate student. We talked of going to Puerto Rico, where she is from, or coming to Brazil, to be near our respective relatives. But we ended up lingering in the USA for a while. In the beginning she was not so enthusiastic about coming here, and her professional situation is not completely settled yet. But we are both very happy about being here now.

MV: You and your group have a parallel computer downstairs. What do you use it for?

MS: A lot of what I do involves domain decomposition techniques, which are particularly suited for parallel computing. Domain decomposition algorithms attempt to solve a problem on an entire domain D by using solutions on subdomains of D. The subdomain can be chosen small enough to be allocated to individual processors of a parallel computer. Due to the good localization of the associated data, well-known sequential and fast algorithms can be used locally for each subdomain. The proper treatment of interfacing solutions is the key to the method with a rich mathematical structure and elegant corresponding analysis. The applications considered are: porous media flow, visco-elastoplastic materials, theoretical Schwarz methods, numerical homogenization and computational fluid dynamics in general.

MV: So, you have students and you have computer facilities at your disposal. What would you say are the main differences between being in the USA and here in Brazil?

MS: Well I guess the main negative difference is that there are fewer resources for research in Brazil; funding of science is still fairly unstable, and you end up spending a lot of time writing research proposals to various funding agencies. If I had more money I would be organizing more conferences here, with my students I would go more frequently to international conferences, we would have more visitors and larger computers. On the positive side I immediately see the quality and motivation of the students. Here, if you want to teach a really advanced course, you can, while you often cannot do this even in the best places in the USA. I mean, if I offer an advanced course here, the students will take it and participate actively. That is a great opportunity for attracting young people to your area!