Qubits And Bell Test Researchers Generate "Perfect" Random Numbers for the First Time

Related Vendors

FAULHABER Drive Systems

SIMTOS (Seoul International Machine Tool Show)

Researchers at ETH Zurich have presented a quantum experiment that they claim can be used to generate perfect and certifiable random numbers.

The background to this is a practical problem with many random number generators: although they provide sufficiently good values for most applications, they are not completely free of systematic deviations. In security-critical applications such as cryptography, such distortions can be relevant because encryption methods are based on random numbers that are as unpredictable as possible.

"It may seem strange, but making a perfect coin or a perfect die is practically impossible," says Renner. No matter how symmetrical and smooth a die is made, one of its six faces will always point upwards slightly more often after a throw. "Even modern random generators, which are based on quantum mechanical effects such as the reflection of photons on beam splitters, are not entirely immune to such a systematic error or bias," adds Wallraff. But now Wallraff, Renner and their teams have found a way to generate perfect random numbers from non-perfect randomness after all. The results were published in the journal Nature.

Random Amplification With Bell Test

The ETH researchers refer to their method as random amplification. First, a non-perfect random generator is used to determine measurements in a quantum experiment. A more random bit sequence is then derived from the measurement results using a special algorithm.

The experiment is based on an improved Bell test with high measurement quality and a high data rate. Such tests examine quantum mechanical correlations between entangled systems and, under suitable conditions, can allow statements to be made as to whether the measured results cannot be explained by classical information transfer.

The setup consists of two superconducting chips that are cooled close to absolute zero. Each chip forms a qubit that can assume the states 0 and 1 as well as superpositions of these states. The chips are connected via a 98 feet-long tube, which is also cooled and through which microwave photons are transmitted.

The spatial separation is intended to ensure that no information can be exchanged between the qubits during the measurement, even at the speed of light. A non-perfect random generator was used to select the respective measurement basis. The team then calculated the certified random sequence from the correlated measurement results.

Relevance for Secure Digital Systems

According to the researchers, the resulting sequence of zeros and ones can be certified as perfectly random. The decisive factor here is not only the statistical distribution of the bits, but also the proof that the randomness is not due to hidden systematic influences.

In the long term, such a procedure could serve as a physically verifiable source of randomness for digital security systems. The researchers compare the role with that of atomic clocks in time measurement: other systems could rely on a defined and verifiable reference.

Possible applications include the encryption of sensitive communication, digital identities, public random services, lotteries and blockchain systems. Certifiable random numbers could also become relevant for quantum-safe communication methods, as the security of cryptographic systems depends largely on the quality of the random sources used.(sg)