Biological system used to solve an NP-complete problem

Ars Technica : NP-complete refers to a type of mathematical problem that can only be solved by exploring all possible solutions simultaneously. Because standard computers are limited to sifting through solutions sequentially, quantum computers had been proposed as one way of achieving the parallel computation required. Now a team of researchers has taken a different route by using a biological system to solve an NP-complete problem. The researchers created a grid of channels connected by junctions that either forced the contents to continue down the same channel or evenly split the contents between two other channels. The team then coated the channels with motor proteins, which are used by cells to move things internally along networks of protein fibers, and loaded pieces of those protein fibers at the top of the grid. The action of the motor proteins quickly moving the fiber fragments down to the bottom of the grid demonstrated all the possible solutions to one of the important problems in complexity theory, the subset sum problem . The system has limitations, however. About 0.2% of the time, the fiber fragments took the wrong path at the straight junctions, which resulted in statistical noise at the bottom of the grid. The researchers think that the noise would be dampened if the grid were scaled up.