Eugene Lawler

Eugene Leighton (Gene) Lawler (1933 – September 2, 1994) was an American computer scientist, a professor of computer science at the University of California, Berkeley.

Lawler came to Harvard as a graduate student in 1954, after a three-year undergraduate program at a southern university. He received a master's degree in 1957, and took a hiatus in his studies, during which he briefly went to law school and worked in the U.S. Army, at a grinding wheel company, and as an electrical engineer at Sylvania from 1959 to 1961. He returned to Harvard in 1958, and completed his Ph.D. in 1962 under the supervision of Anthony G. Oettinger with a dissertation entitled Some Aspects of Discrete Mathematical Programming. He then became a faculty member at the University of Michigan until 1971, when he moved to Berkeley. He retired in 1994, shortly before his death.

At Berkeley, Lawler's doctoral students included Marshall Bern, Chip Martel, Arvind Raghunathan, Arnie Rosenthal, Huzur Saran, David Shmoys, and Tandy Warnow.

Lawler was an expert on combinatorial optimization and a founder of the field, the author of the widely used textbook Combinatorial Optimization: Networks and Matroids and coauthor of The Traveling Salesman Problem: a guided tour of combinatorial optimization. He played a central role in rescuing the ellipsoid method for linear programming from obscurity in the West. He also wrote (with D. E. Wood) a heavily cited 1966 survey on branch and bound algorithms, selected as a citation classic in 1987, and another influential early paper on dynamic programming with J. M. Moore. Lawler was also the first to observe that matroid intersection can be solved in polynomial time.

The NP-completeness proofs for two of Karp's 21 NP-complete problems, directed Hamiltonian cycle and 3-dimensional matching, were credited by Karp to Lawler. The NP-completeness of 3-dimensional matching is an example of one of Lawler's favorite observations, the "mystical power of twoness": for many combinatorial optimization problems that can be parametrized by an integer, the problem can be solved in polynomial time when the parameter is two but becomes NP-complete when the parameter is three. For 3-dimensional matching, the solvable parameter-2 version of the problem is graph matching; the same phenomenon arises in the complexities of 2-coloring and 3-coloring for graphs, in the matroid intersection problem for intersections of two or three matroids, and in 2-SAT and 3-SAT for satisfiability problems. Lenstra writes that "Gene would invariably comment that this is why a world with two sexes has been devised."

...
Wikipedia