Ten years later no one knows when

To a less informed person who has read a whole bunch of publications about quantum computers, one might get the impression that these are “off-the-shelf” machines that work in the same way as ordinary computers. Nothing could be more wrong. Some even believe that there are no quantum computers yet. And others wonder what they will be used for, since they are not designed to replace zero-one systems.

We often hear that the first real and properly functioning quantum computers will appear in about a decade. However, as Linley Gwennap, chief analyst at the Linley Group, noted in the article, “when people say that a quantum computer will appear in ten years, they don’t know when it will happen.”

Despite this vague situation, the atmosphere of competition for the so-called. quantum dominance. Concerned about quantum work and the success of the Chinese, the US administration last December passed the National Quantum Initiative Act (1). The document is intended to provide federal support for the research, development, demonstration, and application of quantum computing and technologies. In a magical ten years, the US government will spend billions building quantum computing infrastructure, ecosystems, and recruiting people. All the major developers of quantum computers - D-Wave, Honeywell, IBM, Intel, IonQ, Microsoft and Rigetti, as well as the creators of quantum algorithms 1QBit and Zapata welcomed this. National Quantum Initiative.

D-WAve Pioneers

In 2007, D-Wave Systems introduced a 128-qubit chip (2), is called world's first quantum computer. However, there was no certainty whether it could be called that - only his work was shown without any details of his construction. In 2009, D-Wave Systems developed a "quantum" image search engine for Google. In May 2011, Lockheed Martin acquired a quantum computer from D-Wave Systems. D-wave one for $ 10 million, while signing a multi-year contract for its operation and development of related algorithms.

In 2012, this machine demonstrated the process of finding the helical protein molecule with the lowest energy. Researchers from D-Wave Systems use systems with different numbers qubits, performed a number of mathematical calculations, some of which were far beyond the capabilities of classical computers. However, in early 2014, John Smolin and Graham Smith published an article claiming that the D-Wave Systems machine was not a machine. Shortly thereafter, Physics of Nature presented the results of experiments proving that D-Wave One is still ...

Another test in June 2014 showed no difference between a classic computer and a D-Wave Systems machine, but the company responded that the difference was only noticeable for tasks more complex than those solved in the test. In early 2017, the company unveiled a machine ostensibly consisting of 2 thousand qubitswhich was 2500 times faster than the fastest classical algorithms. And again, two months later, a group of scientists proved that this comparison was not accurate. For many skeptics, D-Wave systems are still not quantum computers, but their simulations using classical methods.

The fourth generation D-Wave system uses quantum annealingsand the qubit states are realized by superconducting quantum circuits (based on the so-called Josephson junctions). They operate in an environment close to absolute zero and boast a system of 2048 qubits. At the end of 2018, D-Wave introduced to the market BOUNCE, that is, your real-time quantum application environment (KAE). The cloud solution enables external clients to access quantum computing in real time.

In February 2019, D-Wave announced the next generation  Pegasus. It was announced to be "the world's most extensive commercial quantum system" with fifteen connections per qubit instead of six, with over 5 qubits and turning on noise reduction at a previously unknown level. The device should appear on sale in the middle of next year.

Qubits, or superpositions plus entanglement

Standard computer processors rely on packets or pieces of information, each representing a single yes or no answer. Quantum processors are different. They don't work in a zero-one world. elbow bone, the smallest and indivisible unit of quantum information is the described two-dimensional system Hilbert space. Therefore, it differs from the classic beat in that it can be in any superposition two quantum states. The physical model of a qubit is most often given as an example of a particle with spin ½, such as an electron, or the polarization of a single photon.

To harness the power of qubits, you must connect them through a process called confusion. With each added qubit, the processing power of the processor doubles themselves, since the number of entanglements is accompanied by the entanglement of a new qubit with all the states already available in the processor (3). But creating and combining qubits, and then telling them to perform intricate calculations is not an easy task. They stay extremely sensitive to external influenceswhich can lead to calculation errors and, in the worst case, to the decay of entangled qubits, i.e. decoherencewhich is the real curse of quantum systems. As additional qubits are added, the adverse effects of external forces increase. One way to deal with this problem is to enable additional qubits "CONTROL"whose only function is to check and correct the output.

3. Symbolic representation of the 50-qubit IBM system

However, this means that more powerful quantum computers will be needed, useful for solving complex problems, such as determining how protein molecules fold or simulating the physical processes inside atoms. many qubits. Tom Watson of the University of Delft in the Netherlands recently told BBC News:

-

In short, if quantum computers are to take off, you need to come up with an easy way to produce large and stable qubit processors.

Since qubits are unstable, it is extremely difficult to create a system with many of them. So if, in the end, qubits as a concept for quantum computing fail, scientists have an alternative: qubit quantum gates.

A team from Purdue University published a study in npj Quantum Information detailing their creation. Scientists believe that kuditsunlike qubits, they can exist in more than two states—for example, 0, 1, and 2—and for each added state, the computational power of one qudit increases. In other words, you need to encode and process the same amount of information. less glory than qubits.

To create quantum gates containing qudits, the Purdue team encoded four qudits into two entangled photons in terms of frequency and time. The team chose photons because they don't affect the environment as easily, and using multiple domains allowed for more entanglement with fewer photons. The finished gate had a processing power of 20 qubits, although it only required four qudits, with added stability due to the use of photons, making it a promising system for future quantum computers.

Silicon or ion traps

Although not everyone shares this opinion, the use of silicon to build quantum computers appears to have huge benefits, as silicon technology is well established and there is already a large industry associated with it. Silicon is used in Google and IBM quantum processors, although it is cooled to very low temperatures in them. It's not ideal material for quantum systems, but scientists are working on it.

According to a recent publication in Nature, a team of researchers used microwave energy to align two electron particles suspended in silicon and then used them to perform a series of test calculations. The group, which included, in particular, scientists from the University of Wisconsin-Madison "suspended" single electron qubits in a silicon structure, the spin of which was determined by the energy of microwave radiation. In a superposition, an electron simultaneously rotated around two different axes. The two qubits were then combined and programmed to perform test calculations, after which the researchers compared the data generated by the system with data received from a standard computer performing the same test calculations. After correcting the data, a programmable two-bit quantum silicon processor.

Although the percentage of errors is still much higher than in so-called ion traps (devices in which charged particles such as ions, electrons, protons are stored for some time) or computers  based on superconductors such as D-Wave, the achievement remains remarkable as isolating qubits from external noise is extremely difficult. Specialists see opportunities for scaling and improving the system. And the use of silicon, from a technological and economic point of view, is of key importance here.

However, for many researchers, silicon is not the future of quantum computers. In December last year, information appeared that the engineers of the American company IonQ used ytterbium to create the world's most productive quantum computer, surpassing D-Wave and IBM systems.

The result was a machine that contained a single atom in an ion trap (4) uses a single data qubit for encoding, and the qubits are controlled and measured using special laser pulses. The computer has a memory that can store 160 qubits of data. It can also perform calculations simultaneously on 79 qubits.

4. Scheme of the IonQ ion trap

Scientists from IonQ conducted a standard test of the so-called Bernstein-Vaziranian algorithm. The task of the machine was to guess a number between 0 and 1023. Classical computers take eleven guesses for a 10-bit number. Quantum computers use two approaches to guess the result with 100% certainty. On the first attempt, the IonQ quantum computer guessed an average of 73% of the given numbers. When the algorithm is run for any number between 1 and 1023, the success rate for a normal computer is 0,2%, while for IonQ it is 79%.

IonQ experts believe that systems based on ion traps are superior to the silicon quantum computers that Google and other companies are building. Their 79-qubit matrix outperforms Google's Bristlecone quantum processor by 7 qubits. The IonQ result is also sensational when it comes to system uptime. According to the creators of the machine, for a single qubit, it remains at 99,97%, which means an error rate of 0,03%, while the best results of the competition averaged about 0,5%. The 99,3-bit error rate for the IonQ device should be 95%, while most of the competition does not exceed XNUMX%.

It is worth adding that, according to Google researchers quantum supremacy – the point at which a quantum computer outperforms all other available machines – can already be reached with a quantum computer with 49 qubits, provided that the error rate on two-qubit gates is below 0,5%. However, the ion trap method in quantum computing still faces major hurdles to overcome: slow execution time and huge size, as well as the accuracy and scalability of the technology.

Stronghold of ciphers in ruins and other consequences

In January 2019 at CES 2019, IBM CEO Ginni Rometty announced that IBM was already offering an integrated quantum computing system for commercial use. IBM quantum computers5) are physically located in New York as part of the system IBM Q System One. Using the Q Network and Q Quantum Computational Center, developers can easily use the Qiskit software to compile quantum algorithms. Thus, the computing power of IBM quantum computers is available as cloud computing service, reasonably priced.

D-Wave has also been providing such services for some time now, and other major players (such as Amazon) are planning similar quantum cloud offerings. Microsoft went further with the introduction Q# programming language (pronounced like) that can work with Visual Studio and run on a laptop. Programmers have a tool to simulate quantum algorithms and create a software bridge between classical and quantum computing.

However, the question is, what can computers and their computing power actually be useful for? In a study published last October in the journal Science, scientists from IBM, the University of Waterloo and the Technical University of Munich attempted to approximate the types of problems that quantum computers seem best suited to solve.

According to the study, such devices will be able to solve complex linear algebra and optimization problems. It sounds vague, but there may be opportunities for simpler and cheaper solutions to issues that currently require a lot of effort, resources and time, and sometimes are beyond our reach.

5. IBM quantum computer

Useful quantum computing diametrically change the field of cryptography. Thanks to them, encryption codes could be quickly cracked and, possibly, blockchain technology will be destroyed. RSA encryption now seems to be a strong and indestructible defense that protects most of the data and communications in the world. However, a sufficiently powerful quantum computer can easily crack RSA encryption through Shora's algorithm.

How to prevent it? Some advocate increasing the length of public encryption keys to the size needed to overcome quantum decryption. For others, it should be used alone to ensure secure communications. Thanks to quantum cryptography, the very act of intercepting the data would corrupt them, after which the person interfering with the particle would not be able to get useful information from it, and the recipient would be warned about the eavesdropping attempt.

Potential applications of quantum computing are also frequently mentioned. economic analysis and forecasting. Thanks to quantum systems, complex models of market behavior can be expanded to include many more variables than before, leading to more accurate diagnoses and predictions. By simultaneously processing thousands of variables by a quantum computer, it would also be possible to reduce the time and cost required for development. new drugs, transport and logistics solutions, supply chains, climate modelsas well as for solving many other problems of gigantic complexity.

Nevena's law

The world of old computers had its own Moore's law, while quantum computers must be guided by the so-called Nevena's law. He owes his name to one of the most prominent quantum specialists at Google, Hartmut Nevena (6), which states that advances in quantum computing technology are currently being made in double exponential speed.

This means that instead of doubling performance with successive iterations, as was the case with classical computers and Moore's law, quantum technology improves performance much faster.

Experts predict the advent of quantum superiority, which can be translated not only into the superiority of quantum computers over any classical ones, but also in other ways - as the beginning of an era of useful quantum computers. This will pave the way for breakthroughs in chemistry, astrophysics, medicine, security, communications, and more.

However, there is also an opinion that such superiority will never exist, at least not in the foreseeable future. A milder version of skepticism is that quantum computers will never replace classical computers because they are not designed to do so. You can't replace an iPhone or a PC with a quantum machine, just as you can't replace tennis shoes with a nuclear aircraft carrier.. Classic computers let you play games, check email, surf the web, and run programs. Quantum computers in most cases perform simulations that are too complex for binary systems running on computer bits. In other words, individual consumers will get almost no benefit from their own quantum computer, but the real beneficiaries of the invention will be, for example, NASA or the Massachusetts Institute of Technology.

Time will tell which approach is more appropriate - IBM or Google. According to Neven's law, we are only a few months away from seeing a full demonstration of quantum superiority by one team or another. And this is no longer a prospect “in ten years, that is, no one knows when.”