Quantum computing: what it is and what it's not
Quantum computing is slowly making its way into mainstream, but will it make your job obsolete?
Since its conception at the turn of 1960s and 1970s quantum computing has come a long way. What started as highly theoretical work has now a chance to become a reality, offering a diametrically new paradigm for computation. The potential behind this breakthrough technology has spawned an industry projected to be worth north of 45 billion US dollars by 20401, with tens of billions invested to date in public and private funds2. Thus it shouldn’t come as a surprise that hundreds of research groups, start-ups, and established tech players have now joined the race and get a slice of this cake.
All that commotion around quantum computing resulted in the concept being recognized by the general public. However, most publications are either highly technical research papers or popular science articles and YouTube videos. The problem with the former is that it requires a lot of background knowledge, while the latter distorts reality by grossly oversimplifying or misrepresenting facts. As a result, the image of quantum technologies can be quite distorted for people not intimately familiar with the industry - I have been asked about it as if it were some sort of magic, which is certainly not the case.
What is quantum computing, then?
Simply put, quantum computing leverages the laws of quantum mechanics to perform useful computations. In contrast, classical computers rely solely on the laws of (no surprise here) classical physics and mathematical logic. Because the quantum world is so vastly different, it allows for phenomena that don’t have a counterpart in the classical world. Two of the most prominent examples are superposition and entanglement, both of which are crucial for a quantum processing unit (QPU) at the heart of a quantum computer to function.
At this point, most popular science articles mention that quantum computing differs from classical by using qubits instead of the traditional bits and that while a standard bit can only be in a 0 or 1 state, a qubit can be in both states at the same time. Both statements are as far away from the truth as possible. While qubits are indeed the most popular building blocks of quantum computers, it’s possible to use others, such as qutrits3. Furthermore, a quantum system (such as a qubit) can be in a superposition of states, but never in two states at the same time. Quantum mechanics is still physics, not magic.
Are quantum computers better?
The answer is “it depends”, and hinges on what we mean by a better computer.
Let’s start with a computability theory approach and say that computer A is better than B if A can solve every problem that B can plus some extra, without worrying about the time it takes to get the solution. Funnily enough, it turns out that in this sense quantum computers are not better or worse than classical ones, but completely equal! This may come as a surprise, but think of it this way: a quantum computer can be simulated on a normal one, and vice versa. Thus if something can be solved on a quantum computer, it must also be possible to solve it using traditional computers.
Everything changes if we take computing time and economic factors into consideration. The amount of computational resources required to simulate a quantum computer doubles with every additional qubit. Thus, we quickly hit a wall when we increase the size of the QPU. So quickly, in fact, that the world record for the biggest simulated quantum computer stands just shy of 50, and was achieved using the K and Sunway TaihuLight supercomputers. Given that commercially available quantum chips have already surpassed that limit45 and with roadmaps aiming for hundreds of qubits in a few years67, quantum hardware has progressed beyond the simulation capabilities of classical computers.
But besides exponential increase in simulation difficulty, expanding the size of a quantum computer by one qubit also doubles its computing power. Hence algorithms that scale efficiently with the number of qubits are absolutely intractable by mere simulation of the QPU dynamics. Such problems include quantum chemistry or integer factorization, both of which have very real applications. And that, in fact, is the real power of quantum computers. Of course, it is still possible that we find classical algorithms for solving those problems, but despite decades (or in some cases, centuries) of development, we have not been able to do so yet.
Will quantum computers ever replace classical computers?
Simply put, no. And not because of a technological hurdle that we are not yet sure how to overcome. The limitation is far more fundamental and stems from the actual laws of physics. As it turns out, quantum mechanics prohibits copying an unknown and arbitrary state of a quantum system. This restriction is known as the no-cloning theorem and prevents the existence of “quantum programs”. In other words, the description of what operations a QPU is supposed to perform cannot be stored on a “quantum drive” as a quantum state. It must be provided by classical computers!
Interestingly, even without the no-cloning theorem quantum computers would still require classical computers and electronics to operate. Why? The reason is that the result of a quantum computation is obtained via a measurement. This operation is rooted in an inherently classical world, and as such requires (or at least to the best of our current understanding of the problem) a classical system to be performed.
What role will classical computers play?
So if quantum and classical computers must coexist together, and a quantum computer can execute classical algorithms, will the role of good old Macs and PCs be reduced to just controlling the quantum hardware? Here is where I need to get a little speculative, but I think the answer is again negative.
First is the engineering aspect. Contemporary QPU architectures are not necessarily what I call portable. A typical quantum computer, including all the control electronics, can easily fill an entire room and can weigh several tons in total. It also needs significant electrical power to operate, in the range of tens of kilowatts. Of course, some degree of optimization and miniaturization is to be expected over the years, but I really don’t expect a significant reduction in the overall package size, especially for superconducting QPUs with their associated cryogenic hardware. But even if we somehow manage to achieve that, mobile applications will still be highly unlikely, as mechanical vibrations are a known source of errors in QPUs8.
Second, there’s the economic aspect. Quantum computers are expensive to design, produce, and operate, hence it only makes sense to use them for problems that we cannot solve otherwise. It defies logic to use a quantum device to tackle problems that can easily be solved using traditional hardware. So don’t expect quantum Minecraft or quantum Excel anytime soon. That will (most likely) always be the responsibility of classical computers.
What about the role of quantum computers?
As I mentioned already, the allure of quantum computers is to solve problems that are intractable otherwise, either due to insufficient computational power or economic reasons. Hence the biggest market for quantum computers is the high-performance computing (HPC) industry, with QPUs acting as hardware accelerators910. This usage also addresses the high acquisition and maintenance costs, which can be offset by a high device utilization rate.
It’s crucial to emphasize that when realized, quantum computers have the potential to be a literal quantum leap in technology. The impact on humanity that this technology can have is hard to overestimate. Every field of human activity relying on high computing power will benefit from it: from drug design to new materials science, from more efficient aircraft design to better combustion engines, and from transport route optimization to climate modeling.
Advances in those fields will almost certainly lead to even more progress and new discoveries, improving quality of life, and perhaps helping us fight the issues we face as humanity: pandemics, global warming, and many more. And let’s not forget the revenues generated along the way!
Should I start learning quantum algorithms, then?
Companies all over the world have already started to recognize the potential of quantum computing in industry and are clearly interested in reaping its benefits11. Yet the job market is rather unlikely to change because of that, at least not in the short term. After all, quantum computing is just a tool for speeding up calculations, so if your job has not become obsolete because of traditional computers, I think you are safe for the time being.
Unless, of course, you want to work with quantum computers directly, but that’s a topic for a whole new discussion.