We're used to thinking of silicon as being the go-to material in computer architecture. But now a team of scientists has shown that a random assembly of cold nanoparticles can evolve to perform computational tasks too.
A team of researchers from the University of Twente in the Netherlands has been experimenting with tiny gold particles to investigate if they can be encouraged to perform calculations. New Scientist explains how the team took a selection of 20-nanometre gold particles, arranging them in a circle just 1 nanometre from each other. Then the team placed eight electrodes around the circle's circumference, and applied a series of voltages — and found they were able to form a small network of transistors that was able to perform some basic calculations.
That didn't happen by magic, though: the researchers had no idea what kinds of voltages to apply, or where to apply them, in order to make it happen. Instead, the team used algorithms borrowed from evolutionary theory, testing out which voltages might have the right kind of effect, and then carefully honed the best to evolve a set-up that seemed promising. And it worked: the team turned the collection of gold particles into a set of six logic gates that were able to perform some simple tasks.
The researchers reckon that the nature of this kind of computing approach could make it well-suited to some kinds of calculations — like pattern recognition — that are best done by computational cores that run in parallel rather than performing tasks one after the other. It's promising, then, though it may just be a theory for now: These experiments were performed at just 0.3 degrees Celcius above absolute zero, so there's some progress to be made before gold nanoparticles are powering your laptop.
Image by Stinging Eyes under Creative Commons licence.