A Lifelong Researcher - Robert van de Geijn

"I remember when I was five," said Robert van de Geijn, newly retired Professor Emeritus of Computer Science at the University of Texas at Austin, and lead of the Science of High-Performance Computing group at the Oden Institute for Computational Engineering and Sciences. 

"Back in 1967 or ‘68 probably. My father was a medical physicist - which is a computational scientist back before it was fashionable to be a computational scientist. This was before computers were generally available, but an army buddy of his was the manager of an IBM Business Center in the Netherlands. So, in the middle of the night, he would give my father access to their business machines to do his scientific calculations, and my dad would drag me along in the dark in these rooms with these gigantic computers that don't do more than the most basic calculator does now." 

He paused in a moment of reflection. 

"And, you know, I think that that's probably where it started."

The work that van de Geijn has done in the time between living this story and telling it has impacted computer science (CS) on a fundamental level. 

In the early 1970s, theoretical computer scientists like Edsger W. Dijkstra postulated that algorithms could be developed systematically to be correct. This was very exciting with the sole drawback that no one could figure out how to take it from theory to practice - and they kept on not figuring it out until Robert van de Geijn and Maggie Myers, partners in research and life, decided to figure it out. 

It was they who discovered that applying those techniques was very practical in their professional domain of dense linear algebra and that by rigorously applying those techniques, many new algorithms could be discovered. But of course, it wasn't as simple as popping a new technique into a vending machine and shaking it until a half dozen new career-defining algorithms fell out. 

“What really happened,” explained van de Geijn, “was that we were both teaching courses that introduced these techniques to undergraduates while we were discovering new algorithms. At some point in time, we asked ourselves what the process was by which we were managing to come up with these new algorithms while other people couldn't, and what we recognized was that we were subconsciously applying these techniques that had been pushed early in the 1970s. Once we recognized that, we could make it into something formal."

“For a computer scientist - a true computer scientist - it is all about exposing the system,” he insisted. “Notice that science in general is about looking at a phenomenon and capturing what the system behind the phenomena is, which then gives you the theory, which might be theory in physics might be in chemistry, it might be mathematics, it could be computer science. So, in some sense, computer science is a pure science, because it's all about discovering systems.”

Given that van de Geijn and Myers wave the banner of CS as a pure science, it’s perhaps unexpected that their legacy is being written by an offshoot of their research: software libraries.

"We developed these theories of how software should be developed,” said van de Geijn. “That's the scientific contribution that we've made. To demonstrate to the world that these techniques that we created are valid, we did an implementation - a scientific experiment to verify our theories. And then, lo and behold, creating these implementations as libraries was so successful, that the world came along and said, ‘this is fantastic, we're going to fund this, we're going to embrace these libraries, and everybody's going to use them.’"

These extraordinarily valuable software libraries have been embraced and funded by companies like Intel, Facebook, and Oracle.

"The problem is that our stuff is robust enough that we actually get very little feedback. If your software has a lot of bugs, you get a lot of complaints, and then you know that there's a lot of people out there because they are yelling at you," van de Geijn explained.

Given their expertise and experience, van de Geijn and Myers are no doubt in store for a peaceful retirement - or as close an approximation to retirement as two lifelong researchers can stomach.

As Myers described it, “We don’t have to work quite as hard.” 

“I still work as hard because she’s my boss,” quipped van de Geijn.