The race is in full swing. The world's leading companies are trying to create the first quantum computer, based on technology that has long promised scientists to help develop wondrous new materials, perfect data encryption, and accurately predict changes in the Earth's climate. Such a machine will probably appear no earlier than ten years from now, but this does not stop IBM, Microsoft, Google, Intel and others. They literally piece quantum bits - or qubits - onto a processor chip, literally. But the path to quantum computing involves much more than manipulating subatomic particles.
A qubit can represent 0 and 1 at the same time, thanks to the unique quantum phenomenon of superposition. This allows qubits to perform a huge amount of computation at the same time, greatly increasing computational speed and capacity. But there are different types of qubits, and not all of them are created the same. In a programmable silicon quantum chip, for example, the value of a bit (1 or 0) is determined by the direction of rotation of its electron. However, qubits are extremely fragile, and some need temperatures as high as 20 millikelvins - 250 times colder than in deep space - to stay stable.
Of course, a quantum computer is not just a processor. These next-generation systems will require new algorithms, new software, connections, and a bunch of yet-to-be-invented technologies that benefit from colossal computing power. In addition, the results of the calculations will need to be stored somewhere.
"If it hadn't been that hard, we'd have done one already," says Jim Clark, director of quantum hardware at Intel Labs. At CES this year, Intel unveiled a 49-qubit processor, codenamed Tangle Lake. A few years ago, the company created a virtual environment for testing quantum software; it uses the powerful Stampede supercomputer (at the University of Texas) to simulate a 42-qubit processor. However, to actually understand how to write software for quantum computers requires simulating hundreds or even thousands of qubits, Clarke says.
Scientific American interviewed Clarke about the different approaches to building a quantum computer, why they are so fragile, and why the whole thing takes so long. It will be interesting for you.
How is quantum computing different from traditional computing?
Promotional video:
A common metaphor that is used to compare the two types of calculations is a coin. In a traditional computer processor, the transistor is either heads or tails. But if you ask which side the coin is facing when it is spinning, you will say that the answer can be both. This is how quantum computing works. Instead of the usual bits that represent 0 or 1, you have a quantum bit that represents both 0 and 1 at the same time until the qubit stops spinning and enters a resting state.
State space - or the ability to iterate over a huge number of possible combinations - is exponential in the case of a quantum computer. Imagine that I have two coins in my hand and I toss them in the air at the same time. As they rotate, they represent four possible states. If I toss three coins in the air, they represent eight possible states. If I toss fifty coins in the air and ask you how many states they represent, the answer is a number that even the most powerful supercomputer in the world cannot calculate. Three hundred coins - still a relatively small number - will represent more states than atoms in the universe.
Why are qubits so fragile?
The reality is that coins, or qubits, eventually stop spinning and collapse into a certain state, be it heads or tails. The goal of quantum computing is to keep it spinning in superposition in a set of states for a long time. Imagine that a coin is spinning on my table and someone is pushing the table. The coin may fall faster. Noise, temperature changes, electrical fluctuations, or vibration can all interfere with the operation of the qubit and lead to the loss of its data. One way to stabilize certain types of qubits is to keep them cold. Our qubits run in a 55-gallon barrel-sized refrigerator and use a special isotope of helium to cool them to near absolute zero.
How do different types of qubits differ from each other?
There are no less than six or seven different types of qubits, and about three or four of them are actively being considered for use in quantum computers. The difference is how to manipulate the qubits and make them communicate with each other. You need two qubits to communicate with each other in order to perform large "entangled" calculations, and different types of qubits get entangled in different ways. The type I have described that requires extreme cooling is called a superconducting system, which includes our Tangle Lake processor and quantum computers built by Google, IBM and others. Other approaches use oscillating charges of trapped ions - held in place in a vacuum chamber by laser beams - which act as qubits. Intel does not develop trapped ion systems because it requires deep knowledge of lasers and optics,we can't do it.
However, we are studying a third type, which we call silicon spin qubits. They look exactly like traditional silicon transistors, but operate on a single electron. Spin qubits use microwave pulses to control the spin of an electron and release its quantum force. This technology is less mature today than superconducting qubit technology, but it is arguably much more likely to scale and become commercially successful.
How to get to this point from here?
The first step is to make these quantum chips. At the same time, we have performed simulations on a supercomputer. To run Intel's quantum simulator, it takes about five trillion transistors to simulate 42 qubits. It takes a million qubits or more to reach commercial reach, but starting with a simulator like this can build the basic architecture, compilers, and algorithms. Until we have physical systems that will include from a few hundred to a thousand qubits, it is not clear what kind of software we can run on them. There are two ways to increase the size of such a system: one is to add more qubits, which will require more physical space. The problem is that if our goal is to build computers with a million qubits, mathematics will not allow them to scale well. Another way is to compress the internal dimensions of the integrated circuit, but this approach would require a superconducting system, which must be huge. Spin qubits are a million times smaller, so we are looking for other solutions.
In addition, we want to improve the quality of the qubits, which will help us test algorithms and build our system. Quality refers to the accuracy with which information is communicated over time. While many parts of such a system will improve quality, the greatest gains will come from developing new materials and improving the accuracy of microwave pulses and other control electronics.
Recently, the US Subcommittee on Digital Commerce and Consumer Protection held a hearing on quantum computing. What do legislators want to know about this technology?
There are several hearings associated with different committees. If we take quantum computing, we can say that these are computing technologies for the next 100 years. It is only natural for the US and other governments to be interested in their opportunity. The European Union has a multi-billion dollar plan to fund quantum research across Europe. China last fall announced a $ 10 billion research base that will focus on quantum informatics. The question is, what can we do as a country at the national level? A national strategy for quantum computing should be run by universities, government, and industry, working together on different aspects of the technology. Standards are definitely necessary in terms of communications or software architecture. Labor is also a problem. Now, if I open a vacancy for a quantum computing expert, two-thirds of the applicants are likely to be outside the US.
What impact can quantum computing have on the development of artificial intelligence?
Typically, the first proposed quantum algorithms will focus on security (e.g. cryptographic) or chemistry and material modeling. These are problems that are fundamentally insoluble for traditional computers. However, there are tons of startups and groups of scientists working on machine learning and AI with the introduction of quantum computers, even theoretical ones. Given the time frame required for AI development, I would expect traditional chips optimized specifically for AI algorithms, which in turn will have an impact on the development of quantum chips. In any case, AI will definitely get a boost from quantum computing.
When will we see working quantum computers solve real-world problems?
The first transistor was created in 1947. The first integrated circuit was in 1958. Intel's first microprocessor - which contained about 2,500 transistors - didn't come out until 1971. Each of these milestones has been separated by more than a decade. People think that quantum computers are just around the corner, but history shows that advances take time. If in 10 years we have a quantum computer with a few thousand qubits, it will definitely change the world just as the first microprocessor did.
Ilya Khel