Q-score is Atos’ new announcement, the world’s first universal quantum metric. It is applicable to all programmable quantum processors regardless of make and origin. The new metric measures effectiveness in tackling real-life problems that classical computers cannot solve. This is in contrast to only measuring theoretical performance. Atos is committed to delivering early and concrete benefits of quantum computing through Q-score after working with others in industrial and academic partnerships for five years to develop use-cases that employ quantum computing to solve.
Today, how quantum systems are measured is how many qubits they possess, but because qubits vary in quality from maker to maker and are volatile, this is an imperfect way to measure effectiveness. By focusing on how every system solves complex problems, Atos Q-score will provide objective and reliable results to research centres, universities, businesses, and technological leaders who require accurate information.
Q-score measures the actual performance of the quantum processors when they solve an optimisation problem, representative of near-future Noisy Intermediate Scale Quantum (NISQ) times. This will provide a reference frame when providing performance scores and to maintain uniformity of measurement. Q-score will use a standard combinatorial optimisation problem in all assessments, the Max-Cut problem. The calculated score will be based on how many variables a system can optimise in a set amount of time. For example, 40 variables means a 40 Q-score/40 Qs.
Each year, Atos will be publishing a list of the most powerful quantum processors found around the world based on their Q-scores. Due in 2021, manufacturers will provide the scores themselves.
The 3 pillars Q-score (based on an open access software package) has are:
Application driven: Q-score is the only one of its kind based on near-term quantum algorithms, and it will measure the effectiveness of quantum systems when they solve problems.
Openness and ease of use: Q-score will be universal and free, benefitting from the company’s technology-neutral approach. The package will contain all the necessary tools and methodology and is easy to run.
Objectiveness and reliability: Atos will combine a hardware-agnostic, technology-agnostic approach with strong expertise in algorithm design and optimisation. The technology is the fruit of working with industry leaders and clients and will be open to assessment to all.
In 2021, a free software kit allowing Q-score to be run on any processor will be available, and manufacturers are encouraged to publish their results.
Because of Atos’ Quantum Learning Machine, estimates for various platforms’ qubits are available due to the QLM’s powerful simulation abilities. The estimates factor in manufacturer-provided information and results are around 15 Qs. One year ago the scores were all around 10 Q, and projected results will be 20 Qs a year from now.
The Atos Quantum Advisory Board, a group of international experts, mathematicians, and authorities in various physics fields have given Q-score the green light after meeting on December 4, 2020.
Today, quantum computing’s most promising application is to solve large combinatorial optimisation problems. The Travelling Salesmen Problem and Max-Cut Problem are among the most notorious examples. The former involves the question of what is the shortest route a salesman must take when visiting X number of cities once given the distances between the cities. While simple, the problem is made complex when variables pop up and have to be taken into account. Max-Cut is a more generic problem, dealing with a broad range of applications such as optimising electric boards or positioning 5G antennas in various locations. Q-score will produce a score after a quantum processor solves these problems.
The strongest of High Performance Computers (HPC) worldwide to come in the near-term (exascale) can reach the equivalent of 60 Qs, according to estimates. The same estimates also say that the best Quantum Processing Units (QPU) will achieve 15 Qs. Recent progress in the field allows Atos to be optimistic that 20 Q-scores next year will be a reality.
Q-score can be measured for QPUs having more than 200 qubits, an it will be the perfect metric to measure quantum superiority. Atos estimates that quantum superiority in context of optimisation problems to be reached is more than 60 Qs.
Read more from the Original Release