What is quantum computing, and what problems will it help us solve? David Sadek, Doctor in Computer Science, and VP, Research, Technology & Innovation at Thales, specifically in charge of Artificial Intelligence & Information Processing, outlines the progress made and the possibilities ahead.

© Thales

Quantum computers exploit the quantum properties of matter to carry out computing tasks, and they're expected to revolutionise the world of computing as we know it. But in terms of practical applications, what impact will these new machines have in the coming decades?

Quantum computing is more than just a technological breakthrough – it’s a completely new way of seeing computing in the conventional sense of the term and will give us unprecedented problem-solving and data analysis capabilities. The supercomputers currently used by mathematicians carry out processing tasks on the basis of bits, in other words series of 1s and 0s. A bit has either the value 1 or 0, but cannot have both values at the same time. Quantum computers use quantum bits, or qubits, units of information which, according to the laws of quantum physics, can be in both states – 1 and 0 – simultaneously. Like Schrödinger’s cat*, this state of quantum superposition allows computation to be performed simultaneously for all of the "superposed" states of a qubit. Given that the number of computations that can be carried out grows exponentially with the number of qubits used, we can see that a computer capable of handling a large number of qubits would have unprecedented power to address the kind of problems that were previously unsolvable using a conventional approach, or that could only be partly solved after years of computation. Quantum supremacy – the point at which a quantum computer becomes capable of solving problems that no conventional machine could ever solve – will be the next paradigm shift. It will be a real game changer in the world of computing, and like the digital revolution that preceded it, it will open the door to applications that simply didn't exist before.

Given that quantum computers capable of handling a large number of qubits are still at the prototype stage, how do we prepare for this transformation? 

Scientists established the mathematical principles behind this new type of computer long before quantum computers themselves were developed. The first quantum algorithms, designed to leverage the properties of qubits, have been in existence since the 1980s. The Deutsch algorithm, for example, provided the first indication, on a theoretical level, that a quantum computer could one day outperform a conventional computer. Shor's algorithm, which is used for finding the prime factors of an integer, was developed in 1994. Since then, it has come to be seen as a threat – theoretical for the time being, yet nonetheless critical – to public key cryptography systems, which are routinely used in cybersecurity applications, often without users being aware, to protect banking transactions such as credit card payments.

To be part of the coming quantum revolution, we first need to understand the fundamental science behind it. This is why Thales is adapting to this new paradigm in computing to produce a new generation of algorithm developers who are capable of comprehending the IT world in quantum terms – something that requires a fundamental change of mindset. It’s also important to understand that quantum computers capable of handling a sufficiently large number of qubits to achieve quantum supremacy – the Holy Grail of the quantum revolution – may only be at the prototype stage today, but machines with the capacity to run quantum algorithms already exist, so the first concrete use cases are well and truly on the table.

Which technologies are giving us a foretaste of the new world of quantum computing? 

© Thales

They fall into two main categories. First, there are the machines and technologies that exploit the quantum phenomena themselves, such as cold atoms, ion traps, superconducting loops, photon polarisation, etc. Even though the computers which harness these phenomena are currently only capable of handling a small number of qubits, they can be used to run quantum algorithms, so the first proofs-of-concept** (PoC) for real use cases can be developed straight away, and applications could emerge by 2030. For example, a French start-up called Pasqal, which is one of the many companies Thales is working with in this field, is developing a PoC using an actual quantum computer based on cold atom technology, which could soon provide a practical solution for banking security applications.

The second category is emulators, which simulate the operation of a quantum computer on a conventional machine so we can implement quantum algorithms and study their performance today. The problem is that emulating a qubit on a conventional machine requires a significant amount of computing power, and the power requirements will rise exponentially as more qubits are added. 

Even though these two types of technology only provide a foretaste of the capabilities of future quantum computers, they give us important insights into how the machines will work. That means we can already imagine new uses – which will clearly evolve over time – and get to grips with a technology that will go mainstream in the not-too-distant future. We’re a bit like skiers practising on an artificial slope while we're waiting for the start of the winter season so we can get out onto some real snow!

How will Thales position itself in the race to develop quantum computers? 

We’re not aiming to become a manufacturer of quantum computers. But we have significant know-how in key enabling technologies, and this could potentially position us in the supply chain for the future machines. The challenge for a company like ours is to develop practical solutions within the extremely broad field of possibilities opened up by quantum computing. Our objective is clear: we aim to be a pioneer in the development of applications that harness the new possibilities of quantum computing at scale, for both civil and military users, and a leader in the design of the quantum algorithms needed to implement these applications. To achieve this, we will be expanding our capabilities in three key areas simultaneously: the algorithms themselves, by producing a new generation of algorithm developers with expertise in the specific approaches and formalisms used in quantum computing; deployment of actual quantum machines, where Thales is already recognised as one of the only companies in France that can help to benchmark the various quantum computers already in existence; and the software tools that will be needed to take quantum computing into the mainstream and make it accessible to users with less expert knowledge of the underlying theory. By ramping up our capabilities in all these areas and making them available to all the Group's businesses, we will be able to use quantum computing to help solve practical problems in real-world applications that cannot be solved by conventional machines and algorithms.

What type of problems could be solved by quantum computing? 

At Thales, we take a pragmatic approach. Our first step has been to identify the practical challenges to applications in our areas of business that could be overcome by quantum computers when conventional supercomputers reach the limits of their capabilities. Then we classified those practical challenges into six categories of problems that quantum computers could solve: (1) combinatorial optimisation, which involves finding the best option from a large number of possibilities, a classic algorithmic conundrum frequently illustrated by the “travelling salesman problem”***); (2) resolution of linear systems (including differential equations), which could have potentially beneficial applications in the field of electromagnetism; (3) so-called Monte Carlo methods, which are based on the principles of random sampling and can be used for large-scale testing or probabilistic simulations; (4) quantum machine learning, which is at the intersection of artificial intelligence and quantum technology and involves using quantum computers to optimise neural networks; (5) testing cryptographic resistance to decryption by quantum algorithms; and (6) the highly promising field of quantum simulation of matter at molecular level, for example to determine the behaviour of a given chemical or to synthesise new molecules. 

In all these categories, whenever quantum algorithms and quantum computing could provide new ways of solving problems, we are exploring possible applications in all our areas of business. 

Can you give us any concrete examples of applications that Thales has started to explore using quantum algorithms and quantum computing technologies? 

Combinatorial optimisation algorithms are among the most likely to benefit from quantum technology, and we recently launched a PoC relating to a mission planning solution for satellite constellations. To coordinate the movements of satellites, operators have to manage large numbers of parameters and interactions. Even with a small group of satellites, this can quickly create so-called NP-hard problems****, i.e. problems that cannot be solved in a reasonable timeframe by a conventional algorithm, no matter how much computing power is used. In a simulation combining a quantum algorithm with conventional techniques, we have already demonstrated the feasibility of the solution for a small number of satellites, which suggests we could solve the mission planning problem for larger constellations. We are also testing quantum approaches to electromagnetic simulation in radar antenna design, for example using an HHL algorithm (named for its inventors, Harrow, Hassidim and Lloyd) to solve linear equations. Another example is the use of "quantum machine learning" algorithms to expose cyberattacks and for anomaly detection in images.

Will quantum computing eventually replace conventional IT as we know it today? 

It's hard to be sure. What we do know is that a new family of computers will be needed to solve certain problems, such as quantum simulation of matter, which are beyond the computation capabilities of our current machines. We also know for certain that quantum computers will not take over from traditional computers overnight. There will be a long transition, and during this intermediate phase computation tasks will be performed using hybrid solutions that combine high-performance computing (HPC) by conventional supercomputers with the use of quantum processor units (QPUs). In conceptual terms, this hybrid approach is anything but simple, because the tasks to be assigned to the quantum processor and the conventional computer will have to be identified for each operation. It will also be a challenge in terms of the range of skills that the engineers working on these systems will require. But we believe a hybrid approach offers the best opportunities for using quantum algorithms and quantum computing techniques to develop solutions at scale, probably for engineering in the broad sense of the term in the first instance, and then for the development of real-time applications. This is the route that Thales has decided to take.

 

*Schrödinger's cat is a thought experiment in quantum mechanics that was proposed by physicist Erwin Schrödinger in 1935. The experiment involves a cat in a sealed box with a vial of poison that may or may not be released depending on the state of a subatomic particle. If the subatomic particle is in a superposition of states, the cat would be both alive and dead until the box is opened and the state of the particle is determined.

**A Proof of Concept is aimed at demonstrating that a new idea or product is feasible and can be implemented in practice.

***The travelling salesman problem, a classic mathematical conundrum, has been the focus of extensive research over the years. It continues to be used today, for example as an introduction to computational complexity theory, which focuses on the time and memory required for an algorithm to solve a given problem. The travelling salesman problem can be summarised as follows: given a list of cities to be visited by a salesman, with the distances between each pair of cities being known, what is the shortest possible route that visits each city once and returns to the point of departure? Despite this simple definition, the problem is a complex one: the number of possible routes rises exponentially as the number of cities on the salesman’s route increases. To visit 7 cities, for example, there are a total of 360 possible routes. With 15 cities, the number rises to around 43 billion. If we apply the question to a list of 71 cities, the total would be 5*1080, roughly equivalent to the number of atoms in the universe. A conventional computer would quickly be overwhelmed by this phenomenon, known as a combinatorial explosion, and would be forced to suggest approximate solutions, due to the vast amount of time that would be required to compute all the different possibilities. A number of algorithms developed in recent years have proposed using quantum computing to solve the problem more efficiently. Running such algorithms on a suitable machine would finally enable our travelling salesman to find his ideal route – but they could also be used to optimise telecom networks, bus routes or logistics operations.

****In computational complexity theory, NP (non-deterministic polynomial time) is a complexity class used to classify decision problems. NP-hard problems are particularly difficult to solve.