Oral-History:Alain Bensoussan

About Alain Bensoussan

Alain Bensoussan was born in Tunis, Tunisia on May 20th, 1940. He completed a Master's degree in Mathematics and Sciences at Ecole Polytechnique in 1962 and a Master's degree in Economics and Statistics at the Ecole Nationale de la Statistique et de l' Administration Economique in 1965. He also completed his Ph.D. in Mathematics at the University of Paris in 1969 under J. L. Lions. After completing his degree he started at the University Paris-Dauphine as Associate Professor working in "Mathematics for Decision Sciences", later serving as Chairman of the Mathematics department from 1975 to 1977 and achieving Emeritus status in 2004. Bensoussan also served as a part-time Professor at the Ecole Polytechnique from 1970 to 1986 and at the Ecole Normale Supérieure from 1980 to 1985. From 1984 to 1996 he was President of the INRIA, from 1996 to 2003 he was President of CNES, and from 1999 to 2002 he was Chairman of the European Space Agency (ESA). Joining UT Dallas in 2004, he currently serves as a Professor of Risk and Decisions Analysis, as well as Ashbel Smith Professor.

Bensoussan's research interests include control theory, applied mathematics, analysis and stochastic processes, and automatic and applied mathematics. For his contributions to the field, he has received several awards and honors, including most recently the 2014 W. T. & Idalia Reid Prize from SIAM.

In this interview, Alain Bensoussan discusses his career in control. He outlines his work under J. L. Lions, and at Paris-Dauphine, INRIA, CNES, and UT Dallas. He reflects on the differences between the French and US research and education systems, and highlights the challenges and successes of his career. Additionally, he provides advice for young people interested in a career in the field of control.

About the Interview

ALAIN BENSOUSSAN: An Interview Conducted by Michael P. Polis, IEEE History Center, 8 July 2015.

Interview #758 for the IEEE History Center, The Institute of Electrical and Electronics Engineers Inc.

Copyright Statement

This manuscript is being made available for research purposes only. All literary rights in the manuscript, including the right to publish, are reserved to the IEEE History Center. No part of the manuscript may be quoted for publication without the written permission of the Director of IEEE History Center.

Request for permission to quote for publication should be addressed to Oral History Program, IEEE History Center at Stevens Institute of Technology, Samuel C. Williams Library, 3rd Floor, Castle Point on Hudson, Hoboken NJ 07030 USA or ieee-history@ieee.org. It should include identification of the specific passages to be quoted, anticipated use of the passages, and identification of the user.

It is recommended that this oral history be cited as follows:

Alain Bensoussan, an oral history conducted in 2015 by Michael P. Polis, Paris, France at SIAM CT 15.

Interview

INTERVIEWEE: Alain Bensoussan
INTERVIEWER: Michael P. Polis
DATE: 8 July 2015
PLACE: Paris, France at SIAM CT 15

Video

Early Life and Education

Polis:

Good afternoon. My name is Michael Polis and I am here in Paris to conduct an oral history with Alain Bensoussan. This oral history is a cooperative effort between the SIAM CT 15 conference and the IEEE Control Systems Society. Alain, maybe you can tell us a little bit about your, where you grew up? What made you decide to go into mathematics and control, and a little bit about yourself?

Bensoussan:

Okay. Thank you. First of all let me tell you that I am very delighted that you are my interviewer. It is a real pleasure. We have known each other for more than 40 years. Okay.

So in fact what I want to say is I always wanted to become a professor. Not because I like especially to teach or even do research but because professorship was associated with freedom in my idea. So to some extent I was reluctant to work for industry or work for government because I did not like bosses. I thought a professor was the best thing because you are supposed to be free. It is not as true as I believed but to some extent it is. And I am lucky I became a professor and of course, what field? It was not necessarily mathematics because in my days mathematics was pure mathematics, and applied mathematics did not exist. And, especially in the French context, mathematics was very elitist. If you were not a Bourbaki guy, or things like that then you could not think of a career here in mathematics. So I did not expect to become a mathematician. I was more interested in economics, and we can talk about that later. But when I discovered applied mathematics by attending lectures of Jacques-Louis Lions, I thought this is my field. That is what I want to do. And then that was the beginning of the story.

Polis:

I see. How did you make your choice of a university to go to?

Bensoussan:

First of all, I graduated from Ecole Polytechnique which in France is one of the best with Ecoles Normales Superieures for someone who wanted to do scientific studies and. as common in France you go to these schools, engineering schools or business schools and universities. Now it is changing a lot, in those days universities you go to for PhD research and so on. So I did my research with Jacques-Louis Lions and then I was very fortunate, if your question refers to a job, I was very fortunate, really very fortunate that I got my PhD in March ’69, after what we call in France, the ’68 revolution. An interesting outcome of this revolution was that it changed the university system in France, money was not a problem, and it created a lot of new universities. So I finished my PhD in March ’69, I defended, and I got a job immediately in Paris. In Paris-Dauphine, which was one of the new universities, and this is an incredible luck, but today would be absolutely incredible. It was an opportunity which is hard to believe. So I was at Paris-Dauphine, really at the beginning, when we started, with my French colleagues, Jean-Pierre Aubin, and Ivar Ekeland, a new domain called “mathematics for decision sciences”. It was extremely successful and it remains one of the strong points of the university.

Polis:

You talked a little bit about how Jacques-Louis Lions was very influential on you. How did you really choose a dissertation area?

Bensoussan:

Okay. That is an interesting story that I want to tell because it shows how visionary Lions was, as well as what I consider my input in this business. So, Lions was very well known for his work on Partial Differentiation Equations, PDEs. He was renowned as a student of Laurent Schwartz, and Schwartz said that Lions was his best student, so you can imagine. Lions did not know much about the control field, but in looking for new areas, he was introduced to control theory and naturally, he started to do control for PDEs because of his expertise in PDEs and because not much existed in this area at that time. The only work on control of PDEs that existed at that time was from some people in Russia, the former Soviet Union. An important point is that he knew very little about stochastic things. In those days the analyst and the probabilists were very far apart. To show how visionary he was, when we met he gave me a book by David Sworder that was written in ’65 or ’66 about adaptive control. So Lions did not know anything about adaptive control. He did not even know anything about stochastic control but he said maybe there is something interesting in this, look at it. So I found it extremely visionary because he was able to foresee things which appeared very promising, even though at the same time he had no specific knowledge, and no special expertise in that area, so that is fantastic. And now to my contribution. Lions was doing deterministic things, he suggested I look at adaptive, and in between was stochastic. So I thought it would be better to stay in between, and I adopted the strategy of focusing on stochastic control, which then was probably more fruitful than adaptive control. And still very open in those days. And I consider this to be my contribution because I decided myself; it was not Lions, to work on stochastic control. And of course Lions said okay, no problem with that. I think it was a very wise strategy because stochastic control was just starting in those days. Not many things were available. There were a lot of deterministic results, but not so much in stochastic and even now it is extremely promising in terms of possibilities of research and new problems and so on. And that was my decision.

Polis:

Were there other students that were essentially your peers that you worked with or that you talked to about these things other than simply Lions?

Bensoussan:

Well, you know in those days it was just starting, for instance most of my dissertation was on filtering, Kalman filtering for distributed parameter systems. At the same time Ruth Curtain was doing very similar things with Peter Falb. We didn’t know about each other’s work, although subsequently we met and became good friends. This field was moving very fast. I was one of the contributors, but there were many others. Balakrishnan, who passed away recently, was very interested, so he invited me to UCLA in ’69. I was very happy. For the first time I was going to the U.S. And I was very happy to see the landing on the moon in the U.S. in Los Angeles in July ’69. So Balakrishnan was collaborating with Lions. There were many people and not just me, there was Wendell Fleming, also Harold Kushner and Sanjoy Mitter at MIT and Michel Delfour who were working in this area, and you. I met Kalman and of course, you mentioned peers, there was Pierre Faure, who was really a top guy, a top person, deceased too early unfortunately. Kalman’s best student, according to Kalman. He was my school mate at Ecole Polytechnique and although we cooperated, we never published together, even though we worked more or less in the same field. And I was closely following what he was doing. And then he moved to industry, and it was different, but this is typically one of the guys who I really was in contact with and he developed a lot of control and automatic control in Europe and in France in particular. That is what I can say on this topic,

INRIA and CNES

Polis:

You have held some administrative positions in really large organizations particularly in INRIA as well as CNES. Could you tell us a little bit about your experience with those and how you got into them? How you made the decision to go ahead and accept that responsibility?

Bensoussan:

Well, okay. Sure. Of course. By the way, I did not seek these positions. I accepted them but I was not especially looking for them. After all I am not the only one. Lions, my advisor who was really a big mentor also. He held the same positions. He was President of INRIA and also President of the French Space Agency. So obviously you can think he pushed me, recommended me, and he advanced my name. For INRIA, to some extent I am not saying it was natural but I was there from the very beginning and I followed Lions so it was a challenge of course, especially following Lions. It was a big challenge but it was not unnatural.

The story of how I got to the Space Agency is more interesting. It was ’96 and the idea in those days was that it was very important to put space and telecom and information technology together. It is a time of the project of Bill Gates (Teledesic) to put hundreds of satellites into space and as Bill Gates was saying there would be no difference between a child in Africa and a child in New York. It was a big project which was very much ahead of its time but probably it’s only going to be realized now or in the near future. In those days ’96 it was really a hot idea. So the French Space Agency wanted to have somebody who was not from the space domain, someone who maybe had a new vision and so on and at the same time someone coming from having experience in information technology. That was one big element explaining why I was chosen. I had also an administrative background because I had such a position already at INRIA, and this was the idea. This idea of connection between the two things, space and information technology was going to evolve more slowly. It was premature in those days, too early. Particularly, foreshadowed by the failure of a big project launched by Motorola, Irridium. The idea is materializing now but differently. Then when you are in space you love space. It is just fascinating, really fascinating. I want to say also something interesting. I come from control theory as you know and control theory started with space, modern control theory. Right? Modern control theory. Control theory existed before, but it was the space race that really launched modern control theory. Remember that Pontryagin’s book and Bellman’s book were out in ‘57, which was the year of Sputnik. And in those days the two were really going in parallel. And when I went to the French Space Agency we had the rocket called Ariane and I knew that control was used to guide these rockets so I wanted to see if it was true or not. And I must tell you it is exactly like in the books. Which was a sort of surprise to me. When I got all the files about the way the Ariane rocket is guided, I could check that all the algorithms were the ones that I knew from the books like Linear Quadratic Gaussian, LQG, were exactly applied. That was very interesting to see that control theory was really implemented very concretely. The trouble with that is that since space is very conservative once the technology works they keep it for the decades. They just changed the LQG control into robust control, but now it is the same technology so there is little research for us and that is why control moved to other areas and space was no longer the driver of control research. But still, what we learn in the books and in papers, the early papers, Kalman and so on is exactly what they do. It is interesting.

Polis:

Let’s go back a second. How big was INRIA, and how big was CNES when you were running it?

Bensoussan:

Well, when I left it, let’s see it was a big thing. When I left INRIA probably all together because some were really full-time, and others were connected so I would say around 2000 people counting everyone, with a little bit less than 1000 full-time. Now, I want to say the momentum I left in 1996, almost 19 years ago, has never declined. Never, it is really fantastic. I think now they have maybe 4000 all together, it is a fantastic place. I also want to say something about this since it just comes to mind. I remember that Wendell Fleming was asked, in the ‘70s maybe, to write a report to an agency, probably it was NSF, I don’t remember all the details, but what I remember about this report is that it said there should be an organization in the U.S. similar to INRIA. We were very happy of course and felt it was a nice comment.

French v. US Research Systems

Polis:

Well, I guess the French system is a little bit different than the U.S. In the U.S. the universities tend to be quite autonomous and people do their research in their universities. Whereas in France it seems like there are sort of government agencies where even though you are a university professor you actually do a lot of your work in the government-sponsored agency. Could you tell us a little bit about that, and also given your experience recently in the U.S., could you maybe compare the two systems?

Bensoussan:

Well, I would say that recently the two systems are getting closer, the differences have been reduced. You see for instance that now INRIA is very much connected to the university. The big research organization in France is called the CNRS. CNRS contains all the research areas. The origin of this difference is probably because to some extent in France the university was considered more of a teaching organization especially for what we call undergraduate in the U.S. In undergrad is not much research. It was felt that the university would not provide enough research emphasis, especially for emerging fields, so we needed to have something dedicated. It was in particular the idea of General de Gaulle who created INRIA and The French Space Agency, and other such organizations. The idea was that we needed to do something in certain specific areas and that research was one part. By the way, government research labs exist also in the U.S. You have many National Labs such as Argonne, or Los Alamos, run by DOE, the Department of Energy, or the many NASA Labs.

Polis:

Perhaps the difference is in the U.S. most of these labs started during the second world war and the major theme was the atomic bomb. Whereas I think in France it was a different objective.

Bensoussan:

No. It was the same. It was the need to have to do something in specific fields. One field, atomic energy led to the creation of the French Atomic Energy Agency and the need for information technology led to INRIA. The idea was that France was not doing enough in information technology. Moreover research had to be boosted. When NASA was created it was the same. The difference is that the universities in the U.S. are very strong and very autonomous in their funding coming, most of it from students, but not so much from the Federal government, except for research grants. And in France it was totally different. But now, as I said, because things have changed, because of globalization the systems worldwide are becoming very similar so there are fewer differences.

Extended Stays and Collaborations

Polis:

Are there any special places that you travelled for example on sabbatical or extended visits that you might want to talk about?

Bensoussan:

My time in France you mean?

Polis:

Yes. Particularly in France but-

Bensoussan:

Yes. First of all I was going a lot to the U.S. of course and I have been very happy to visit many countries. I never took a sabbatical. Never. Even now I am in the U.S. for already eleven years but haven’t taken a sabbatical. I was allowed to spend half a year in Hong Kong, so I said maybe asking for a sabbatical is too much so I don’t have really this type of experience. Besides while I was at the University of Paris-Dauphine I spent four years in Brussels. In Brussels in a new organization sponsored by some people at Dauphine. It was interesting because it gave me, it was early in the ‘70s, and Europe was far from being organized like it is now and especially in terms of research. There was nothing to connect the Europeans. In fact, it was more likely that the Europeans would meet for the first time in California than elsewhere. This was the early days and it was really interesting because I met many Europeans in this place in Brussels, that I probably would not have met otherwise. I know it was different to leave France, and I learned a lot about the differences when I moved to the U.S.

Polis:

I said sabbaticals. I probably should have said extended stays.

Bensoussan:

Yes. Kokotovich invited me to Illinois during the summer of 1980. I gave a class in Illinois in his lab and then I wrote a book about my lecture notes. It was very nice experience. I was 40 years old then and it was the first time I gave lectures in the U.S. That was my only extended stay outside France, because after that I started as President of INRIA 1984. So it was then impossible to leave for an extended period.

Polis:

So your stay was related to the control field. I know that you are doing economic systems, can you elaborate?

Bensoussan:

I always was interested in economics. As I said in my University, Dauphine, I created, together with my colleagues something dedicated to mathematics for decision sciences. It was always a field of interest. I work in inventory control and then with Lions’ in the theory of quasi-rational inequalities. It is the application of course but I want to be clear I never really implemented that in industry but in terms of motivating interesting mathematics. To some extent you can think of mathematical finance as the same type of thing. Not all the mathematicians who work in mathematical finance work for banks, but they develop a lot of very interesting mathematics especially in stochastic control. I want to say that I always found very interesting problems in applications. I was never interested in pure math. And probably also not capable to do anything but problems coming from application in particular, not just economics of management science, but mostly those. By the way, at UT-Dallas, I’m not in the math department, I‘m in the School of Management, and in Hong Kong I am in the Engineering College and the Department of System Engineering and Engineering Management. So to some extent, except in Dauphine where we created mathematics, but not a mathematics department in terms of purely mathematics, we clearly wanted mathematics for decision sciences. So really, surprisingly, I found mathematics more interesting to be imbedded in engineering or management colleges or schools other than mathematics. I have a very good relation with math department at UT-Dallas. Even students there but I am not from the math department.

Polis:

I think the first time that we met was in Canada. Could you tell us a little bit about your collaboration with Michel Delfour and how that happened?

Bensoussan:

Oh, that’s easy. In fact Michel Delfour was a student of Sanjoy Mitter so I first met Mitter. Sanjoy invited me to MIT, and we worked together and I was visiting him frequently and met Michel, and we’ve kept in contact since then. I think it was before Michel moved back to Canada. We worked in control of delay equations a lot and then Michel, Sanjoy, DaPrato, and I, had this book on infinite dimensional systems. It was a very nice thing to do. Very interesting and I kept him as a very good friend. Of course when you have friends like that you keep them. That’s the case with you.

Literary Contributions

Polis:

You have written a number of books. I think the French tend to write books more than just publish papers but a book is really an extensive work. Could you tell us about how you got involved in writing so many books?

Bensoussan:

Well, as you say it is, and I would not write the books now as I wrote them before. You know the French style, even though I was not at all in pure mathematics, but still we were told that when we write a theorem we have to write all assumptions. If you do so, then it’s kind of accepted as mathematics. Now, I’m much more relaxed. So I write formal things, I don’t care really about all the accurate assumptions, and so on. I think now I want to stress more with pedagogical aspects rather than mathematical rigor. So in other words how to convey, because I believe that finally mathematics should be very simple. There is a simple idea and you should convey it. Of course there is a lot of work to put rigor in it, but mathematics is very intuitive. It is not at all as people think to have the sort of sequence of arguments and logic, it is first intuition. You have to show the power of mathematics to students and that is what I am writing now. Less books of course but what I want to do. I am writing a book by the way, taking from the old and new things in control theory. This is a challenge, but I think that in control theory there is so much that the core is forgotten and the new scholars who do control, do it in one area or another area and they forget about what is in common. So I have this project to write a book in explaining the common things and then the specific things. It is one of the big achievements of control theory and one of the weaknesses. It has succeeded so much, so well, that it is imbedded now and we forget control as a core discipline, however people doing control in different areas have many things I believe in common. This is not just in control theory. It comes up in other disciplines but it is one of the things I see in control. Maybe one challenge for our discipline is how to convey this idea that there is really a discipline which covers all these aspects and many many applications. This is a good thing but we should not spread the discipline so much that it appears there is nothing in common.

Comments on Education

Polis:

It seems the tradition in France is much more mathematical than the one in the U.S. You know in the U.S. we constantly complain that our undergraduate students don’t have the math background that we would like them to have particularly for advanced control courses. Has that changed in France or is that still the case?

Bensoussan:

We have a strong mathematics school and the evidence of that is the Fields Medal where France has the same, I think, the same number of winners as those in the U.S. Which, if you compare the two countries is not bad. That is, mathematics has always been one of the strengths of French education. Now I would say, probably again with globalization, the world is more integrated and the students travel a lot. Now Mathematics is using a lot of computing, something which did not exist in the past. Which is a fantastic opportunity because you can really implement mathematics. Okay. That is very good but the drawback is of course that computers may hide the strength of mathematics, the beauty of mathematics and this intuition. So I notice this in the U.S. when I went to the U.S. in 2004. Before that I left teaching when I was at the Space Agency. So from ’96 to 2004 things have changed a lot, but the thing, which I noticed is that students are happy when they can apply a mathematical formula. They don’t care where it comes from. So that’s understandable but of course it has a lot of drawbacks because you cannot control that the formula is correct because you just apply, but if you have something wrong in the formula you have no clue. And then you put that in the computer and it gives you figures and you have no way to check whether this is good or not. Okay. So it is very important that you teach your students that you have to challenge, you have to challenge Matlab. Matlab makes mistakes. A lot of mistakes I can show you that Matlab does. It does not mean that Matlab is wrong but you have to challenge it, and you have to understand what you are doing and that is why as I said before you have to mention intuition. So mathematics is not something boring. It is something very intuitive. It is very helpful. But you need to control completely what you are doing. These aspects are spreading worldwide. All the mathematical exercises that were done in the past are disappearing. It is true also in France not just in the U.S. and there is a new way to do mathematics, I agree. But what remains really important is that mathematics is a fantastic way to describe things. It is a language. It is a fantastic language. The best language in the world and also a language which is very adequate to describe reasoning, but beyond that it is a way to formalize intuition and that is what we have to focus on teaching these days. And I guess it is the same throughout the world because recently I see from the Chinese, who I believe used to be extremely good, like the French, in mathematics and so on. But they are not an exception. They are like the Americans, like the French and others. The way to combine mathematics and computers, is the important thing to teach. Student will appreciate.

Polis:

How do you go about teaching intuition?

Bensoussan:

Okay. First of all, that is why I don’t care now about making precise the assumptions and so on. I want to explain one thing. Why is this result occurring? And you can explain it in very simple cases. That is one way. I think a very good way and once you have understood it, it can be generalized to more complex situations. It is a challenge to follow this approach but in many cases it is possible. Even Laurent Schwartz was saying mathematics is very easy. It is very, it should be simple. It looks very difficult but it is in fact simple provided that you explain what the key thing is. What is the key idea? There is no recipe for that but once you have that in mind you can say okay, what I am telling is too difficult. Okay. And then I should simplify it, simplify it, simplify it. And you do your best.

Move from France to US

Polis:

We talked about your moved from France to the U.S. Could you tell us a little bit about what your thinking was that made you make that move?

Bensoussan:

Oh, first of all I am very grateful to the U.S. and to my university, to Texas and Dallas, which I did not know by the way. It is very fortunate that the U.S. allows people to stay forever. In France and Europe in many places you have to retire and I don’t think it is a good idea. The U.S. as always is ahead. And I was very fortunate so in fact you are asking a very good question because, of course it would be natural for somebody like me to go to another place in Europe but it does not work because other European countries have the same problem as France. In other words you have to retire at a fixed age. Maybe U.K. now, United Kingdom has introduced but it is only contracts. Even in Hong Kong I don’t have tenure, only contracts. So, because I am too old, they give me a contract and then they may stop it or not. And in the U.K. it is the same. Canada is now like the U.S. I think.

Polis:

Very close. They, I am not sure but in the past it was retirement at sixty five. In the U.S. also. In the U.S. it is only the last twenty years or so.

Bensoussan:

That is a very good thing. First of all, it is very good for people like me because I am happy to be with young people, and I cannot work by myself in my home. I need to be in a professional place. It has challenges because you have to write proposals, you have to grade students so it is something which is not so nice but it is a price to pay to be a professional and that is very good for staying in good health and good spirits.

UT Dallas

Polis:

Who recruited you to UT Dallas?

Bensoussan:

Oh, that is a good point. First of all, I was writing to some friends. And one of them was Suresh Sethi who is a person who has many friends in the world, and I was one of them and we worked on some papers together and he works in control theory in management science. He is one of the stars in this domain. I did not know he was in Dallas. He was in Canada at first. He was in Toronto. He invited me in Toronto a long time ago and I knew he moved to the U.S. I thought he was in Arizona or somewhere. I must say also that in the U.S. and Canada you have the cold places. I could not go to Minneapolis, even New York. Boston, of course if MIT gives me a job I would go but it is so cold. So for me the United States is California, Florida, Texas is part of it. So I was very fortunate. I thought Arizona, he was in Arizona. So I wrote to him. He was in Dallas. And then he was very nice because he introduced me to the Dean who is a very open.

Polis:

Who was the Dean at that time?

Bensoussan:

The Dean is Hasan Pirkul, the Dean of the School of Management. A very quantitative school. Very quantitative, so they like people with a quantitative background. And the Dean is from Information Sciences, Management Information Systems and so on. So he gave me a contract for three years and then let’s see after three years I got tenure and then I got a chair, so I am very grateful to them. By the way I want to say that Texas is a wonderful state. Let me make some, if you allow me, some publicity. It is the most dynamic place in the U.S. Dallas-Fort Worth is the fourth place in the U.S. after New York, Los Angeles and Chicago. Number four, and the fastest growing, and UTD, the University of Texas in Dallas, when I went there in 2004 it was 14,000 students and now it is 24,000 students. How many places in the U.S. are growing so fast? So it is fantastic growth. Very, very nice. Very interesting. It is not Stanford, I agree. It is not MIT, but it is a very dynamic place, I am amazed. Engineering has grown quickly, and so it is a very nice place to be.

Polis:

At one point UT Dallas was mainly graduate work. Has that changed?

Bensoussan:

Oh, yes. That has changed. It was graduate because UT Dallas was a spinoff of Texas Instruments. So it is a new university, in the sense that it is less than 40 years old. Texas Instruments had labs and they had something like Bell Labs and they at some point decided to stop that, and they gave it to the state of Texas to do something, which was graduate work, and so they started graduate studies. And as a full-fledged university it is very recent. We are part of the University of Texas system and the major one is Austin, which is a university of 50,000 students. But it was surprising that a place like Dallas Fort Worth had no big university. There was the University of Texas at Arlington; Arlington, Texas, Arlington is between Dallas and Fort Worth. It was not Dallas. Dallas is a really good place and it had no university of that stature, not like MIT, but among the first tier universities, and that is the objective of UTD, although we are still not there. So, first tier in Texas you have only UT-Austin, Rice and Texas A&M. That is all. UTD would be the next one. So it is a very interesting challenge and although we are behind, we are now taken very seriously in the Texas university system. Thus, UT-Dallas is certainly an interesting place because it is in a big city with lots of industry and opportunities.

Polis:

Do you collaborate with some of the groups in engineering?

Bensoussan:

Oh, yes, a lot. Engineering, as I told you was very limited in the beginning of the university so engineering was only Electrical Engineering and Computer Science. Now they have started mechanical engineering, bioengineering. They have started Systems Engineering so they are growing a lot and I work a lot with them. I work in particular, surprisingly, with computer science people on issues related to cyber security, which is very popular in the U.S. And because of my interest in risk management, something which I learned from space. Space is a very risky business. Risk management is a very interesting new discipline, I think. There are a lot of interesting connections with stochastic control, and other things. And it turns out that in Dallas we have a strong group in cyber security. And they are very open to mathematics, very open to mathematical methods and so I work a lot with them. I wrote many papers with them and they get a lot of NSF grants. And they put me as a Co-PI and I am very happy with these people. This is one thing, and second thing is Mechanical Engineering and Systems Engineering. In fact, the Dean of the School of Engineering is Mark Spong, a control guy. So they are very much in favor of control. So we have excellent relations, and in mechanical engineering there is also somebody who is very very close to control, and in Systems Engineering as well. So there are a lot of interesting collaborations, and I get students from these places. I also get students from math department. So it works. We have very good relations, and so that is very fruitful.

Career Highlights

Polis:

Could you reflect a little bit on what you might consider the highlights of your career?

Bensoussan:

In terms of research? In terms of what?

Polis:

Whatever.

Bensoussan:

First of all, I was very lucky because I started control when it was associated with the space effort. It was the beginning, especially for stochastic control, and distributed parameter systems, and so on. So I was very lucky to be there in the early days, they very [were] interesting. Also with INRIA it was fantastic to see the boom with internet, and INRIA was involved right away and we created a lot of things around the internet. Also Europe. Europe was not organized at all in terms of research. Again people did not know each other at all. And we organized everything in Europe, so I was lucky to be there at times of pioneering. Another thing for instance, INRIA and the start-up and all these things which existed only in Silicon Valley, and now exists in many places, but they did not exist at all in France. In France the mathematical community was far from industry, and the concept of high tech research connected to industry was not there. I was very fortunate to be able to push at INRIA that and now it is a tremendous success. So to some extent I was lucky to be there at a good time. Good times for some things. In the Space Agency it was the beginning of the new Ariane. Unfortunately, the first launch was a failure, but I learned a lot. Of course it is very unfortunate but again I learned a lot about what to do and we could talk a lot about this and in particular the importance of having an idea of risk and taking decision with risk, and not so many people even in industry are doing that. It is progressing. So I was very lucky but I had also the opportunity to face big problems and big challenges. The first failure of Ariane was four months after I became President of the French Space Agency. The minister told me “You know, this time it is not your fault but next time it will be yours”. So when you are in something like that you have to do something. So there were interesting challenges.

Career Challenges

Polis:

Were there other challenges in your career that you might want to talk about?

Bensoussan:

Well, now maybe I can talk about the future challenges. Now for control as I said control has a challenge. It is extremely useful but people are not so much aware of that, so it is a little bit hidden and it is very important that people understand. Thus, its challenge is to be a discipline which is really at the forefront of progress, also innovation but also visible. Visible. This is a challenge. I think statistics is today in a fantastic boom. But to me they may have the same problem, for instance statistics people speak about big data. Everybody is in big data now. The number one discipline should be statistics, right? I am very interested in statistics. In fact I started as a statistician. I moved to control but my training in school was in statistics. After Polytechnique I went to a statistics school. Concerning opportunities, big data is certainly one of them. Who is going to benefit from that? I believe control theory has a lot to benefit from it and I want to explain why? For instance IBM is claiming that they are doing a lot with big data. I am not even speaking of universities, so a company like IBM and the guy in charge says ”with big data, we are going to do stochastic optimization. When I hear stochastic optimization it rings a bell. Right? Between stochastic optimization and stochastic control there is not much difference. What they have in mind is that it will be possible to incorporate all uncertainties in our decisions. That is the idea. This is not the case now or in the past. And that is what they think is the objective of big data. So our challenge now, my challenge, your challenge maybe and for all the control guys is really to be there. The statisticians have the same challenge because before deciding you need to deal with data. You need to forecast. You need to do statistical work. All right. Remember statistics and controls were very close to each other in the past. But also, the computer scientists come in because they do data mining and things like that. This is computer science. Okay. So I think one of the challenges we have now and it is an interesting one is to be part of this new interesting evolution and to have our full share of it because we deserve it. What we do is useful to that and we need to connect at the same time with all the disciplines also. I think it is one of the challenges I would like to highlight for control these days in the future.

Polis:

Are there any anecdotes or stories maybe on the lighter side that you want to share with us?

Bensoussan:

There could be many. I mentioned the story about adaptive control. I don’t know if it is a story or an anecdote but I find it interesting. It is very important to see what will be successful and what even advanced may fail. I mentioned this story about what Fleming wrote about INRIA which was very nice and maybe I will refer to the joke of Bill Levine, which I like about control and protrol. Maybe something which comes to my mind, when Kalman--by the way, I met Kalman a month ago; Very interesting to see him again. When he was meeting with Russian scientists and they were looking at their progress compared to the progress of the space race and making some kind of parallel...

Polis:

When was this?

Bensoussan:

Oh, it was in the late ‘60’s, ‘70s, you know, in those days you had the competition with the Americans and the Russians and all theorems, there was one Russian and one American so there was Weiner-Kolmogorov. There was Pontryagin-Bellman. It was nice to see this competition. Of course things have changed a lot so. Well, that is a few of them.

Advice to Young People

Polis:

What advice would you give to a young person that is thinking about a career in the control area?

Bensoussan:

Okay. So maybe I will refer to a few things that I alluded to before. Mathematics is a friend. You have to learn mathematics. It is very important. You have to learn in the right way so not to forget about the intuition but you have to learn mathematics. Without mathematics you can do little. Of course now something I mentioned, someone in control should look at big data applications. So if you can be in this area it is a very good thing to do. I will mention things, which I am aware of in management science. We have seen mathematical finance, the boom of mathematical finance. I remember the times when mathematics of finance was so limited. There was nothing except the discount cash flow and that was all the mathematics in finance. Now when you see what has happened you are amazed. Now the new thing is Marketing; you could believe marketing is just something with no mathematics. Just the art of advertisement but not at all. It is extremely quantitative these days. And because of something called revenue management, which is a fast expanding domain where control theory is extremely, could be extremely important. It is not so much known so if someone is interested in using control theory in the context of management science I would really encourage him or her to look at revenue management. In engineering I see two things. One I am not really an expert but I think it is important. It is control for Nano systems. Because we have Nano systems everywhere now. Right. And we need to control them and how to do with them? It is not like controlling a system. To some extent it is going back to the early days of control, which means external representation, stability, and this seems to be important for Nano systems. Nano technology is extremely fashionable, extremely popular and control has something to say. So control of Nano system would be a very good domain. And another one is totally the opposite. It is control of the very large systems like the grid. When you look at this problem with all the aspects it is really a very large coordinated system, which goes to every local consumer and local and large producers, so there is a major coordination problem which is management of the grid. So this is a stochastic control problem for a very large, very large system. So you have Nano system and you have very large systems. And that I think are very interesting challenges for control theorists. So there are good times. You need to, I would say to the student, be careful in your choices. You have to challenge your professor. You have to challenge your books. You have to find by yourself what a good strategy is. What you like and not just accept what somebody tells you. You have to find yourself; the advisor can give you some information but your decision is yours. That is very important. You have to help yourself and decide by yourself. That is what I would say to students.

Polis:

Well, Alain, thank you very much. On behalf of both the Control Systems Society as well as SIAM CT 15 it has been a pleasure talking to you and again. Thank you very much.

Bensoussan:

Thank you, Mike. It is a pleasure.