MIL-OSI USA: NIST and Partners Use Quantum Mechanics to Make a Factory for Random Numbers

Source: US Government research organizations

Randomness is incredibly useful. People often draw straws, throw dice or flip coins to make fair choices. Random numbers can enable auditors to make completely unbiased selections. Randomness is also key in security; if a password or code is an unguessable string of numbers, it’s harder to crack. Many of our cryptographic systems today use random number generators to produce secure keys.

But how do you know that a random number is truly random? Classical computer algorithms can only create pseudo-random numbers, and someone with enough knowledge of the algorithm or the system could manipulate it or predict the next number. An expert in sleight of hand could rig a coin flip to guarantee a heads or tails result. Even the most careful coin flips can have bias; with enough study, their outcomes could be predicted. 

“True randomness is something that nothing in the universe can predict in advance,” said Krister Shalm, a physicist at the National Institute of Standards and Technology (NIST). Even if a random number generator used seemingly random processes in nature, it would be hard to verify that those numbers are truly random, Shalm added.

Einstein believed that nature isn’t random, famously saying, “God does not play dice with the universe.” Scientists have since proved that Einstein was wrong. Unlike dice or computer algorithms, quantum mechanics is inherently random. Carrying out a quantum experiment called a Bell test, Shalm and his team have transformed this source of true quantum randomness into a traceable and certifiable random-number service. Their results were just published in Nature. 

“If God does play dice with the universe, then you can turn that into the best random number generator that the universe allows,” Shalm said. “We really wanted to take that experiment out of the lab and turn it into a useful public service.”

To make that happen, NIST researchers and their colleagues at the University of Colorado Boulder created the Colorado University Randomness Beacon (CURBy). CURBy produces random numbers automatically and broadcasts them daily through a website for anyone to use.

At the heart of this service is the NIST-run Bell test, which provides truly random results. This randomness acts as a kind of raw material that the rest of the researchers’ setup “refines” into random numbers published by the beacon. 

The Bell test measures pairs of “entangled” photons whose properties are correlated even when separated by vast distances. When researchers measure an individual particle, the outcome is random, but the properties of the pair are more correlated than classical physics allows, enabling researchers to verify the randomness. Einstein called this quantum nonlocality “spooky action at a distance.”

This is the first random number generator service to use quantum nonlocality as a source of its numbers, and the most transparent source of random numbers to date. That’s because the results are certifiable and traceable to a greater extent than ever before.

“CURBy is one of the first publicly available services that operates with a provable quantum advantage. That’s a big milestone for us,” Shalm explained. “The quality and origin of these random bits can be directly certified in a way that conventional random number generators are unable to.”

NIST performed one of the first complete experimental Bell tests in 2015, which firmly established that quantum mechanics is truly random. In 2018, NIST pioneered methods to use these Bell tests to build  Really Random Numbers”>the world’s first sources of true randomness.

However, turning these quantum correlations into random numbers is hard work. NIST’s first breakthrough demonstrations of the Bell test required months of setup to run for a few hours, and it took a great deal of time to collect enough data to generate 512 bits of true randomness. Shalm and the team spent the past few years building the experiment to be robust and to run automatically so it can provide random numbers on demand. In its first 40 days of operation, the protocol produced random numbers 7,434 times out of 7,454 attempts, a 99.7% success rate.