Jaime San Martín: “If all goes well, Chile will have a new supercomputer next year”

In this interview with the Universidad de Chile, the academic explains the importance of the service provided by the Guacolda-Leftraru supercomputing unit, Chile’s upcoming challenges in this area and in artificial intelligence, and the need to strengthen collaboration in the country in order to compete at a global level.

“This is a computer that will allow us to grow up to five times in capacity”, says Jaime San Martín, the most recent winner of the National Prize for Exact Sciences, researcher at the Center for Mathematical Modeling (CMM) of the University of Chile and scientific director of the National Laboratory for High Performance Computing (NLHPC).

“This is not an isolated award or an award for a single person. I think it is rather an award to a community”, says Jaime San Martín about his winning the 2023 National Prize for Exact Sciences. When he received the news of his distinction from the Minister of Education, he says that he remembered with emotion his family, especially his grandparents and parents, who managed to reunite in Chile after living through the Spanish Civil War, Franco’s dictatorship and World War II, as well as his childhood in the house they built and where he lives to this day, in the San Joaquín neighborhood of La Legua. “Chile is very important to my family. There is a lot of gratitude from those who have already left to the people of Chile, for having welcomed this family and given them the opportunity to grow”, he says.

Jaime San Martín thinks that this important distinction is due in part to the supercomputing dream he embarked on together with academics from the University of Chile and other institutions in the late 2000s, an initiative that allowed Leftraru, the country’s first supercomputer, with a capacity of 70 teraFlops of theoretical performance, to be launched in 2014. “Then, there was a feeling that if we did not have supercomputing capacity there were areas of science that we were not going to be able to cultivate in Chile”, he assures. Then, in 2019, Guacolda was added, a unit with which the National Laboratory for High Performance Computing (NLHPC), where he serves as scientific director, reached 266 teraFlops of nominal performance.

“It is one of the nicest stories that I have had to live, because I have met many people and many places where Chile is producing and growing”, says Professor San Martín about this project, who emphasizes that there is currently a community of more than 500 users of the supercomputing unit that is now housed in the Beauchef campus of the University of Chile. “The type of service we provide is of international level, but we have to think about the next generation of computers. When you buy a supercomputer, you already have to have in mind what you are going to need in the next four to five years,” he says.

– Why is it so important today for the country to have supercomputing units such as Guacolda-Leftraru?

The design of almost anything today requires large simulations. In drug design, simulations allow some molecules to be selected so that they can be tested in the laboratory. But there are hundreds of other problems that have to do with everyday life. For example, if you want to improve Santiago’s traffic, monitor our infrastructure or fresh water reserves. In art, for 3D simulation. Then there are all the issues that have to do with climate change and estimating what Santiago’s climate is going to be like in the next 30 years. You have very complex and computationally demanding models that allow you to make those estimates. If I am an astronomer and I want to simulate the evolution of galaxies, that is pure supercomputing. If I’m a nuclear physicist and I want to simulate nuclear fusion, that’s supercomputing. You don’t do experiments anymore by dropping bombs to measure what’s going on. If I’m a mathematician and I want to make some conjecture about prime numbers. Well, I’m going to use a lot of computation. In everyday and scientific life we use it, and today it is an indispensable edge.

– What is the present situation regarding the use of supercomputing capacity in Chile?

Very few companies use supercomputing in Chile today. In the more productive world, this tool should be introduced little by little and, in fact, some large companies use it, but it should be much more common. It is like in the 50’s, when the first computers appeared. Very few companies in the world used computers and today the vast majority use them. The same thing is going to happen here. All over the world, large companies do use supercomputers. In Chile, this is going to enter little by little, so I would expect that in the next ten years this will be more and more common.

On the other hand, one of our most important users today, after the university system, are government work groups. There are several groups from the MOP and other ministries that use supercomputing, for example, to determine how vulnerable our coastal infrastructure is to tidal waves, tsunamis, etcetera. That is done via simulation. I would expect the use of supercomputing to be more intense in the State. If you want to improve urban transportation, if you want to use satellite imagery to see how we are taking care of our forests.

– What are the country’s needs in this area?

We have become convinced that the best thing to do is to share this infrastructure. It is the only efficient way to have a world-class infrastructure, perhaps not in size, but in the service provided. From the point of view of the supercomputer world, Leftraru and Guacolda are very small, but they still cover the needs of the country quite well. But of course you have to be thinking about the next generation of computers. When you buy a supercomputer, you have to have in mind what you are going to need in the next four to five years.

The only way to compete globally in things like this is with large groups that break down the barriers between institutions and disciplines, to be able to talk, understand each other and come up with an answer, a solution. I believe that competition is good and is what made the national scientific system grow so much, but we have reached a point where competition alone is no longer useful. Both are important, but what we have neglected in Chile is to finance and encourage collaboration, which is fundamental for the great challenges.

– How do you see the future development of supercomputing in Chile? Do you have new projects in mind?

Last year we signed an agreement between 44 institutions. We had an event with the Minister of Science, Flavio Salazar, and we said that Chile deserves to have a world-class supercomputing center. To this end, we asked the State to make a direct investment in a center that we are going to build, so that any money that the State puts into this center will reach everyone equally.

In the short term, which for me is three to four years, Chile should have resolved how this supercomputing center is going to work, where everyone has equal access. Solving that means a couple of things that are not easy. One is governance, the other is how many resources we are going to put into this as a society. Investments have to be very well studied, thinking about a computer that lasts about five years and that this has to be a continuous cycle. We should invest in the next four years something like $5 million dollars in a machine. It is money, but it is not so much money in relation to what it means for the country.

New supercomputer for Chile and technological challenges

Professor San Martín comments that a bidding process for $1,050 million pesos is currently open for the construction of a new supercomputer for Chile, an infrastructure that is expected to cover the country’s supercomputing needs for the next three years. With this new machine, which would be added to the installed capacity of Guacolda-Leftraru, Chile could reach the petaFLOP performance by 2024.

– What can you tell us about the supercomputing unit that is currently in the bidding process?

If all goes well, we will have a new supercomputer next year. If Guacolda is about four times more powerful than Leftraru, this computer should be up to five times more powerful than Guacolda. It’s not a single machine, it’s several styles of machines. We are going to make a more important investment in what is called GPU, which is more used in artificial intelligence and other applications. We made the definitions based on an estimation of what we should buy with the budget we have. I believe that this computer we are buying today will satisfy for at least two or three years the needs we are seeing arising in research, development and innovation. This is a bidding process that is being carried out by a project made up of 26 universities and more than 70 researchers. We hope that at the beginning of next year it will be up and running.

– Will this new infrastructure seek to cover requirements associated with artificial intelligence applications?

Yes. There are many groups in Chile that are making artificial intelligence applications, using our infrastructure or others in the country, and we certainly have the capabilities to do so, both in terms of human resources and infrastructure. Of course, more is always needed. With the computer we are buying now, there is a good budget to buy specific infrastructure for artificial intelligence. This is not science fiction, it is already a reality in Chile. Now, we should not overdo movies either. Chile has a small research capacity compared to the world and other countries. But I do believe that we can give the “stick to the cat” in some things, in some application.

For example, ALeRCE [Automatic Learning for the Rapid Classification of Events], which is a Chilean development led by Francisco Foster, an astronomer who arrived at the CMM of the U. de Chile about eight or ten years ago. This is a support system for astronomy that is very focused on this new large telescope that is about to be inaugurated, the Vera C. Rubin, which is going to start operating very soon, where the amount of information per night that it will generate is crazy. ALeRCE has been developed to be a broker for this telescope, which is an intermediate stage between the observatory and the users. It’s like someone summarizing relevant information and telling you what to look out for based on your interest. There are tools for everything, artificial intelligence, machine learning. This is a spectacular collaboration between many people, not only astronomers, many people who come from computer science, mathematics, many engineers.

Graduate, teacher, student…

Jaime San Martín’s life is intimately linked to the University of Chile. After his school years at the Miguel León Prado Institute on Gran Avenida, and despite his initial taste for chemistry, in 1978 he entered the Faculty of Physical Sciences and Mathematics at our campus, following the advice of a professor at his school, who recommended that he explore his tastes more freely in the common plan of the Engineering career. “It was the best advice I ever received in my life. I entered the University of Chile and began to see the different areas that interested me. Besides, it was a world where diversity was the great virtue, as well as depth and rigor. Here you meet the whole of Chilean society, much of which I did not know”, he says.

“Those were very rich years, in the midst of the dictatorship, of course, but the Engineering School of the University of Chile was a very special place”, he recalls of that time. Then, in 1986, he joined the Department of Mathematical Engineering, a unit in which he still works as a Full Professor.

– What is your vision of the great challenges facing the University today?

The great challenge is to remain a relevant university for the country. There are many ways of defining this. It has to do with continuing to be a place where opportunities are given; with always being concerned about the quality of all the processes that occur in the University, from training, research and extension of the projects in which we work; maintaining good levels of internal democracy, and that it is possible to have different positions and that this takes place in an atmosphere of respect. Many of these aspects are satisfied today, but I believe that we have to work permanently on them. They are constant goals to which we must put energy.

– What message would you like to convey to students and future generations of professionals at the University?

It is a great privilege to be able to study at a university like the Universidad de Chile. It is not the only one in Chile where one should feel privileged. But entering the Universidad de Chile is a privilege, in general, because of the level of education you are going to receive, and that is something I transmit to my students. Every year I teach first year classes and the first thing I tell them, at the beginning of the course, is that they and I are very privileged. I am privileged to have so many good students in the classroom and they are privileged to enter a place that is going to give them a very good education. And this has a responsibility in the working world. When they leave, they have a profound responsibility to society.

By Cristian Fuentes, journalist at the Universidad de Chile.

Posted on Oct 13, 2023 in News