RIO DE JANEIRO, BRAZIL – In the 1980s, physicist Richard Feynman thought about the advantages of simulating nature with a quantum computer instead of a conventional computer. He thus initiated a process that is beginning to yield results almost 40 years later.
This week Google announced that one of their quantum computers is far superior to the world’s fastest supercomputer. The comparison is not entirely fair, since a specific issue tailored to the way a quantum computer works was selected for the power test.
Nevertheless, the announcement has attracted a great deal of attention worldwide and has raised concerns that quantum computers could soon crack the conventional encryption systems.
We are still a long way from achieving this. We are going to explain what a quantum computer is, where its strengths and weaknesses lie, and what can realistically be expected of it in the coming years.
What is a quantum computer?
A quantum computer uses individual particles that obey the laws of quantum physics. These can be, for instance, electrons, charged atoms (ions) or quantum light.
These quantum objects behave in a way that is not known to classical physics. Thus an atom can be in two places at the same time, or propagate like a wave. Parts of this wave can superimpose and extinguish each other.
The researchers have been trying for 25 years to implement the concept of using the effects of quantum physics for computers that work faster than any supercomputer. They use the fact that quantum mechanical particles can be in two different states at the same time.
If these objects are turned into information carriers, these so-called qubits can simultaneously store the values 0 and 1. This distinguishes them from conventional bits, which are either in the 0 or 1 state.
Each additional qubit doubles the number of simultaneously storable values. Approximately 300 qubits are enough to absorb more numbers than the number of particles in the universe.
In order to use a quantum computer to perform calculations, the qubits must be selectively addressed and linked according to an algorithm. This way, one can process countless stored values simultaneously, i.e. many calculation paths can be run in parallel.
The problem: An observation of the qubits arbitrarily selects one of the alternatives. Therefore, a quantum computer produces a meaningful result only if the algorithm controls the calculation methods so well that the wrong variants, such as waves, cancel each other out and only the right one remains before they are measured.
What can (not) be done with a quantum computer?
The quantum computer displays its advantages when the number of possible solutions grows immeasurably.
The question of which round trip passing through five cities is the shortest, can be solved with paper and pencil. With 15 cities, there are some 87 billion possible routes, and each station multiplies this number even more.
A conventional computer, including future ones, will sooner or later stop. However, a quantum computer can potentially solve such problems. With each additional qubit, its computing power doubles and grows exponentially.
Similarly complex tasks like the problem of the traveling salesman can be found in machine learning or the breaking of encryptions. Experts believe that quantum computers will quickly search huge amounts of data for hidden patterns.
The problem is that programmers have to find a trick for each new application in order to superimpose the solutions in such a way that the optimal solution is filtered out. So far, only a few such algorithms exist. It is therefore difficult to assess the actual potential of this new type of computer.
However, most experts agree that one application will soon emerge: the simulation of molecules or crystals. Classical computers have a hard time doing this because, according to quantum physics, they can only simulate the many possibilities that are found in a molecule by successive calculation steps.
This is why even supercomputers fail to simulate molecules with over 50 atoms. Sufficiently large quantum computers will probably be able to simulate much larger compounds, such as those found in biology. This could, for instance, reduce the cost and speed up the search for active ingredients in the pharmaceutical industry.
Even if in decades to come there will be a quantum computer that can be programmed flexibly, it will probably not serve as a tablet in everyday life. For tasks such as word processing, the classical computer will continue to be best suited.
Is the quantum computer a danger to Internet security?
In 1994, the American mathematician Peter Shor provided the motivation to develop quantum computers.
The algorithm named after him cracks common encryption methods, provided a quantum computer with several thousand qubits is available. This concerns so-called asymmetric crypto-methods, in which the partners do not have to exchange a common key.
They are widely used on the Internet, in online banking, for instance, or for digital signatures that verify authorship.
To crack the code, the quantum computer finds out which prime numbers have to be multiplied to obtain a certain number with several hundred decimal digits. A classical computer would, roughly speaking, have to test all possibilities – more than there are particles in the universe. A personal computer would need billions of years for that.
A powerful quantum computer can do this within minutes. Experts are now warning that data that is to remain confidential for more than a decade should no longer be backed up using the relevant methods.
Post-quantum cryptography is currently being developed. It uses mathematical methods that quantum computers are unlikely to crack. There are already ready-to-use alternatives for digital signatures, and encryption methods are under development.
Due to this post-quantum cryptography, the quantum computer should pose a much smaller threat than initially thought.
Why is it difficult to build a quantum computer?
A qubit is more sensitive than any prompter: it reacts to the slightest environmental impact. Even the impact of an air molecule or heat radiation can turn the superimposed state of a quantum bit into a conventional state.
Therefore, researchers protect their qubits with vacuum, very low temperatures, and shielding. Nevertheless, qubits usually only maintain their superposition state for fractions of a second. An algorithm that lasts for minutes cannot be completed in this way.
Researchers are succeeding in keeping the qubits intact for increasingly longer. But errors in the execution of calculation steps still occur far too often. In the future, they will, therefore, be recognized and corrected by the quantum computer itself.
However, in order to do this, it needs several additional correction qubits for each calculating qubit. Researchers fear that this issue could blow up a powerful machine to millions of qubits.
Currently, nobody knows how to build an error-corrected, powerful quantum computer. This is basic research.
When will the breakthrough happen?
Experts agree that a flexible, i.e. freely programmable quantum computer should not be expected before 2030.
Some believe that it will take much longer. If there will be a major breakthrough at all. Evolution is also conceivable, starting with small quantum computers that solve narrowly defined partial tasks and leave the remainder to a classical computer, to those that master individual tasks such as optimization well, to the freely programmable quantum computer.
The quantum computer that Google has now presented marks the beginning of this development. We are now in the era of medium-sized quantum computers with qubits that have not yet been error corrected.
Such qubits could be just stable enough to perform limited calculations, such as simulating parts of molecules. It is therefore conceivable that such quantum computers will initially function as coprocessors for classical computers.
This could be achieved in a few years – provided that the quality of qubits with no error correction is sufficient. If this is not the case, it would also be conceivable to start with a small number of error-corrected quantum bits and gradually increase the number of bits.
Thus, it could result in a series of smaller achievements that will start in a few years. The one big breakthrough may never come.
Who is developing quantum computers and who is ahead?
According to the media, the big American companies Google and IBM are ahead. These companies are also leading in terms of the sheer number of qubits.
IBM offers a cloud service that allows researchers to test their algorithms on quantum processors. Google and IBM want to find practical applications for their computers as quickly as possible. There is also a startup in the US that is developing software and hardware for quantum computers with hundreds of millions of dollars of venture capital.
Europe is a leader in academic research. To maintain its status, physicists from the old continent called on the EU Commission to promote the field more strongly in 2016.
The Commission heard the call and two years later set up a flagship quantum technology project with a budget of €1 billion. Within this framework, two European quantum computers are to be built by 2021.
One of them is similar to Google and IBM computers, the other uses ion qubits. Physicists from Innsbruck (Austria) are the leaders in this type of quantum computer.
China is also investing billions of dollars in the construction of a national laboratory for quantum technology near the city of Hefei. The online retailer Alibaba, a kind of Chinese Amazon, is developing quantum computers and offers access via the cloud to a quantum computer with 11 qubits.
Whether the USA, Europe or China are in the lead is hard to say. It is still too uncertain what the future holds. Researchers often stress that individual institutions will find it difficult to cope with the technology’s complexity.
According to researchers, a functioning ecosystem of physicists, engineers, software developers, and users is required to achieve this goal. Whoever is the first to achieve it has a good chance of taking the lead.