Short and Sweet

In computer science, a huge range of problems can be represented as “graphs”—mathematical abstractions consisting of nodes, depicted as circles, connected by edges, depicted as line segments. The nodes can represent almost anything—routers in a communication network, airline flights, movie titles.

Analyzing such graphs—finding a route through a communications network, or predicting a Netflix subscriber’s movie preferences—often requires something called a graph Laplacian, which is a big grid of numbers (a matrix) representing the strength of the connections between nodes. Each row of the grid can also be treated as a separate equation using the same variables; solving the Laplacian means solving all the equations simultaneously.

For years the best algorithms for solving graph Laplacians had an execution time that increased exponentially with the number of variables, but a paper published in 2004 presented the first of several “nearly linear time” algorithms. In June, at the ACM Symposium on the Theory of Computing, MIT researchers will present a new algorithm that’s drastically simpler than its predecessors.

“The 2004 paper required fundamental innovations in multiple branches of mathematics and computer science, but it ended up being split into three papers that I think were 130 pages in aggregate,” says Jonathan Kelner, an associate professor of applied mathematics, who led the new research. “We were able to replace it with something that would fit on a blackboard.”

One way to think about graph Laplacians is to imagine the graph as an electrical circuit and the edges as resistors; solving the Laplacian tells you how much current would flow between any two points in the graph.

The MIT algorithm’s first step is to find a “spanning tree” for the graph. A tree is a graph that has no closed loops; a spanning tree is one that touches all of a graph’s nodes.

The algorithm then adds back just one of the graph’s missing edges, creating a loop. On the circuit analogy, the net voltage around the loop must be zero, so the algorithm plugs in values for current flow that balance the loop. Then it adds back another missing edge and rebalances.

Remarkably, this simple, repetitive process converges on the solution of the graph Laplacian. Demonstrating that convergence didn’t require any sophisticated mathematics: “Once you find the right way of thinking about the problem, everything just falls into place,” Kelner explains.