Photo

Joseph F. Traub.Credit Eileen Barroso/Columbia Engineering

Joseph F. Traub, who founded the computer science department at Columbia University and who helped develop algorithms used in scientific computing in physics and mathematics as well as on Wall Street, died on Monday in Santa Fe, N.M. He was 83.

The cause could not be immediately determined, his wife, Pamela McCorduck, said.

Professor Traub took a detour from physics and became immersed in computing in the 1950s, well before there was a discipline known as computer science.

As an educator, he was a skillful advocate, both for more resources and more respectability for the young field at a time when the people grappling with the big machines in computer centers were regarded in much of academia as glorified mechanics.

He became the dean of the computer science department at Carnegie Mellon University in Pittsburgh in 1971 and had built the faculty to 50 people from a handful by 1979, when he left for Columbia, where he had been recruited to become the founding dean of the new computer science department.

At Columbia, Professor Traub recalled in an interview in 2001, his challenge was to “convince one of the great arts and sciences universities in the United States that computer science was really central.”

Not long after he arrived, he delivered a university lecture, “What Will Be the Intellectual Impact of Computers?” His prescient answer, in essence, was that the impact would be broad, deep and in ways that had not yet been imagined.

“Joe Traub was one of the first to recognize the potential of computer science education at universities,” said H.T. Kung, a professor at Harvard and a former student of Professor Traub.

Professor Traub encouraged his peers to address the social and economic implications of their work. He was the founding chairman of the Computer Science and Telecommunications Board of the National Academies of Sciences, Engineering and Medicine, the nation’s leading advisory group on science and technology, and he served two terms as the head of the board, from 1986 to 1992 and again from 2005 to 2009.

Joseph Frederick Traub was born on June 24, 1932, in Karlsruhe, Germany, the only child of Leo Traub and the former Mimi Nussbaum. Leo Traub was a banker in Karlsruhe, but after the Nazis seized the bank in 1938, the family fled and arrived in New York in 1939. Joseph attended the Bronx High School of Science and earned an undergraduate degree in math and physics from City College of New York.

He enrolled at Columbia in 1954, intending to become a theoretical physicist. But a friend urged him to visit a building with a computer on 116th ^{Street, near the Columbia campus, then the home of the IBM Watson Labs.}

“I got hooked on computing,” Professor Traub recalled in an interview in 2011 for the Computer History Museum. “I left physics, but I didn’t enroll in a department, because there were no departments of computer science, and I wanted to study computing.”

But Columbia did have a “committee on applied mathematics,” with faculty members from several departments. The committee allowed Professor Traub to do his Ph.D. thesis on computing calculations in quantum mechanics, which allowed him to run programs for hours on an IBM 650 computer. “I discovered that I loved programming,” he recalled. “That was the beginning.”

Professor Traub’s first marriage, to Susanne Traub, ended in divorce. In addition to Ms. McCorduck, he is survived by two daughters from his first marriage, Claudia Renee Traub and Hillary Ann Spector, and four grandchildren.

In the early days of computing, the machines were hulking but their capabilities — processing, storage and memory — were minuscule compared with today’s computers, so designing software algorithms to do the most work with the least use of computing power was crucial. Programmers always tried to do that, but there was no theory of optimal algorithms.

Professor Traub focused on that challenge — first at Bell Laboratories, which he joined in 1959, and later, starting in 1970, in academia. This study of the minimal resources required to solve computational problems was called “computational complexity.” And Professor Traub added the insight that the optimal algorithm design depended on the information it was working on, which became known as “information-based complexity.”

He also helped develop software tools used to value financial derivatives.

Over the years he worked with a number of collaborators, most notably Henryk Wozniakowski, a professor at Columbia and the University of Warsaw.

Prabhakar Raghavan, a computer scientist at Google, called Professor Traub “a hard-core theorist,” but one with insights that apply to areas attracting huge investment today by technology companies, including cloud computing and machine learning, a branch of artificial intelligence.

In recent years, Professor Traub had been doing research to apply similar principles to a new terrain of complexity, quantum computers. In theory, quantum computers would be unconstrained by the on-off switches, the 1’s and 0’s, of digital computing.

Professor Traub pursued the quantum research despite long odds of seeing it to completion.

“I’ve always admired him for tackling such a hard research area,” said Jeannette M. Wing, a computer scientist in Microsoft’s research laboratory, “especially knowing that he would be unlikely to see a practical quantum computer in his lifetime.”