There are various fields of quantum computing. The first field involves quantum accelerators. This is a way to use quantum-based algorithms to accelerate specific calculations. Another way to apply quantum computing is to use it on sensors, for example to enhance the precision of atomic clocks. The third way is what is called the universal computer or the general-purpose quantum computer. In fact, Atos is working in this field by creating a kind of computer which could at a certain point in time replace the traditional computers.
Since quantum computers do not exist today, Atos decided to create a simulator of such a quantum computer. Atos designed an appliance and built a hardware, firmware and software layer with the so-called Atos quantum assembly language, giving the opportunity to the users to program using quantum gates instead of Boolean gates. Atos is delivering this, targeting first the universities, the academic world and the labs. Atos is starting with a very attractive price in order for everybody to be able to program quantum algorithms and benefit from the opportunity of this new paradigm.
Atos is delivering from 33 to 44 qubits of quantum capabilities in its simulator. The user doesn't have quantum acceleration because there is no quantum accelerator embedded in the system but you can program it and validate the functionalities of quantum algorithms. Quantum programmation and quantum computation are an exponential function. It cannot be compared to traditional HPC which is linear. If you give, for instance, something to compute in 20 seconds and you double the complexity, you will need 40 seconds to compute it on a traditional computer. On a quantum computer, if you have a quantum algorithm, you can accelerate this exponentially.
Arnaud Bertrand gave an example. If you have an assembly of 1000 people and you want to detect the people taller than 1,8 meters with brown eyes, in traditional computing you will come and see each person in the assembly, taking five seconds to measure the person and check whether the eyes are brown. Then, you pass to the next one. You will need to do this 1000 times. In quantum computing, there is an opposite way of processing the data. You will ask the assembly who is taller than 1,8 meters and has brown eyes. The persons who will answer to the description will raise their arm. The computation power is exponential because if you have an assembly of three people, the traditional way of calculation is good and even better than quantum computing. If, however, you have an assembly of 1000 persons, quantum computing and physics are much more efficient. If you have an assembly of 500.000 people, you cannot even compute it on a traditional computer. It has to be done on a quantum computer.
The only problem is, because of probability, you will have a quantum computer to ask the question twice, or even three, four or five times, just to be sure you read the appropriate information. A quantum computer can be much more efficient than a traditional computer but in very specific use cases and depending on the size of the data. In fact, the efficiency cannot be compared but if you take, for example, one of the quantum algorithms which is Shore's algorithm, a way to factorize prime numbers, it is quite impossible to do it on a traditional computer. It can take perhaps 1000 years to do it with a 1000-bit number. If you want to do this with Shore's algorithm on a real quantum computer, it takes something like one hour instead of thousands of years.
Primeur Magazineasked Arnaud Bertrand whether there are any benchmarks to measure speed or performance for quantum computing, as there are plenty for traditional computing.
Arnaud Betrand answered that today, you don't have any benchmarks on performance. What you have, is the size of the qubits. With the qubits, you know how much data you can process simultaneously. On the 30-qubit Atos system, if you are searching one qubit, it is twice the size of 30 qubits. The complexity is exponential. From 30 qubits to 40 qubits, it is 1024 times more powerful.
The machine that was introduced on July 4, 2017 is not a quantum-accelerated computer. It is a simulator. Atos built a simulator because there is a need to understand what one can do with this new paradigm. The goal is to provide a way to develop quantum algorithms and to understand the way to program quantum devices and quantum algorithms, and to validate the functionalities of those algorithms. Another reason is because the labs and the academic world are working on physical quantum computers. They are modelling and treating qubits in diamonds or in trapped ions. They need something to see whether their calculation is good. Because there is no quantum computer, they don't know. With the Atos system, they have a reference and they can use the Atos system as a reference for the physical qubit experiment.
Primeur Magazineobserved that the simulator can also be used for educational purposes at universities and in research centres.
Arnaud Bertrand confirmed this. Because of Shore's algorithm, the simulator can change the way we are teaching. Probably, the people who do research on cybersecurity will be interested to test the Shore algorithm in order to find algorithms which are quantum safe and which can resist quantum computing. Currently, the machine is not fast enough yet. You can validate the functionality but you won't have the quantum acceleration because there is no quantum accelerator in the system.
Primeur Magazineunderstands that the simulator is a preparation for the real quantum machines and wanted to know whether Atos already is working on the real quantum computers.
Arnaud Bertrand stated that Atos is working with the labs today because it is not industrialized yet. It is still at the lab level. The readiness is not yet good enough to be industrialized. Atos is working with labs like CEA but also with others in Europe. Arnaud Bertrand said that to be honest, he didn't know when the first quantum computers will be on the market. It can be 10 years, it can be 20 years or it can be never. The Atos team thinks it will be there one day. Arnaud Bertrand estimated that there will sooner be quantum accelerators than a general-purpose computer. What is positive about the Atos appliance is that you will be able to put quantum accelerators in this appliance and benefit from quantum acceleration before having a full general-purpose quantum computer.
Primeur Magazineobserved that it is similar to GPU accelerators today, then.
Arnaud Bertrand confirmed this. Because of the language that Atos developed, it will have just to replace the subroutine simulating the effect of Shore's algorithm, for example, with the driver to the Shore board in the quantum accelerator. All the development you are able to do today can be continued to be used in the future, if quantum accelerators go live and are utilized.