Researchers built the fastest random number generator ever made, using a simple laser. It exploits fluctuations in the intensity of light to generate chance – a coveted resource in applications such as data encryption and scientific simulations – and could lead to devices small enough to match one computer chip.
Really randomly surprisingly hard to come by. Algorithms in conventional computers can produce sequences of numbers that seem random at first, but over time these tend to show patterns. This makes them at least partially predictable, and therefore vulnerable to decoding.
To make encryption safer, researchers have turned to quantum mechanics, where the laws of physics guarantee that the results of certain measurements – for example when a radioactive atom decays – are truly random.
A popular way to exploit quantitative chance is to exploit fluctuations in how photons are emitted by the materials used in lasers. Typical laser devices are designed to minimize these fluctuations to produce light of constant intensity: they cause the light waves to bounce inside the material to force its atoms to emit more and more photons synchronized with each other.
But for a random number, researchers aim for the opposite. “We want the intensity to shift randomly, so we can digitize the intensity to generate random numbers,” says Hui Cao, an applied physicist at Yale University in New Haven, Connecticut.
Cao and her team made their laser material – a translucent semiconductor – in the shape of a bow tie. Photons bounce between the curved walls of the bow tie several times, before appearing as a scattered beam. The researchers can then capture the light with an ultra-fast camera. They recorded the light output of 254 independent pixels, which together produced random bits at a rate of about 250 terabits per second, or 250 terahertz. That’s several sizes faster than previous such devices that recorded only one pixel at a time. Their results were reported in Science the 25 of February.
The invention “represents a major leap in performance of random number generators,” says Krister Shalm, a physicist at the U.S. National Institute of Standards and Technology in Boulder, Colorado.
The fastest existing computers have clock speeds measured in gigahertz, which is far too slow to fully exploit the full power of Cao’s device. The layout could be scaled down with simpler light detectors instead of a high-speed camera. This could eventually give practical devices small enough to fit one computer chip, Cao says. These could have useful applications, such as encryption technology on mobile phones.
This article is reproduced with permission and was first published on March 2, 2021.