Fundamental Research Division
The DRF at the CEA assemble approximately 6,000 scientists since January 2016.
Découvertes et avancées | Scientific result | Computing | Quantum Physics
Theoretical physicists at the IPhT have shown that by equipping a quantum processor with quantum memory, it is possible to reduce the number of quantum bits needed to solve an emblematic problem by three orders of magnitude.
While conventional computers contain a processor and memories, the scientific community has imagined until now the quantum computer as a processor without any memory and containing quantum bits (qubits) situated on a plane. This processor performs operations on the qubits and corrects the inevitable errors generated by the intrinsic fragility of their state.
Researchers at the IPhT asked themselves what a new architecture would offer by combining a quantum processor with quantum memory. To evaluate this idea, they chose Shor's algorithm, which, if implemented, would make it possible to break common encryption systems such as the RSA cryptosystem. This algorithm, designed for a quantum computer, has in fact an exponential advantage in computing time in comparison to current computers. In particular, Google estimated that a 20 million-bit quantum processor could solve this problem in a few hours, which is otherwise unsolvable with a conventional computer.
The physicists have shown that with the processor-memory combination, this problem now requires only 200,000 quantum bits, which is a gain of two orders of magnitude! By adding optimized error correction procedures, the number of qubits can decrease by another order of magnitude to 13,000.
"It was quite the surprise that this gain was so strong," recalls Nicolas Sangouard, author of the study. "Researchers working at Google and professional hackers were immediately interested in our result."
This gain of three orders of magnitude is indeed very important as it removes a barrier: under current techniques, the creation of 20 million qubits at low temperature would lead to insurmountable volumes to be cooled.
The computing time, on the other hand, increases from a few hours to a few months, although the lifespan of confidential data that may be divulged is even longer!
The trade-off for the reduction in processor size is a decrease in the parallelism of operations, which now need to be carried out more in series. The quantum computing power based on the superposition of quantum states is not affected at all.
This new configuration also requires a very faithful memory since the quantum states of the qubits are transferred from the memory to the processor and then vice versa, each time they are required to perform an operation.
"The idea for this work came to me following discussions with Patrice Bertet, a researcher at Iramis," explains Nicolas Sangouard. "His team knows how to extract the state of a superconducting qubit in the form of a microwave photon and keep it intact for nearly 100 ms. It is then necessary to successfully carry out the reverse operation in order to transfer the quantum states to the qubits of a processor and show that it is possible to manipulate this information using a large number of logical operations, by performing a memory-processor round trip between each operation."
Factoring 2 048 RSA integers in 177 days with 13 436 qubits and a multimode memory, PRL
CEA is a French government-funded technological research organisation in four main areas: low-carbon energies, defense and security, information technologies and health technologies. A prominent player in the European Research Area, it is involved in setting up collaborative projects with many partners around the world.