| 2020-03-31 18:21 |
Lead: Reading from Tencent quantum laboratory.
Finishing | Jiang Shang Bao
2019 was a year of quantum computing occupy the hot news forum, calculated from January IBM announced the world's first commercial quantum prototype to September Google declared for the first time in the world "Quantum hegemony."
This year is a milestone in the history of on quantum computing, because it means that quantum computing era will come.
Quantum computing in the end how far away from us? Large-scale commercial but also how long? The current state of the art in what era?
March 28, at the headquarters of CCF YOCSEF AC members, the headquarters of any academic secretary Hou, director of the Center for Internet Security industry, Beijing Jiaotong University, Tao Yaodong, CCF YOCSEF headquarters AC members, Professor of Computer and Information Technology, Beijing Jiaotong University, Wang Wei presided, CCF YOCSEF (China computer Federation Youth computer Technology Forum) held a "quantum computers far away from us?" online forum for this topic, Dr. Zhang Hui origin of quantum computing from Hefei Science and Technology Co., Ltd. made a "quantum computer development situation and prospects, "the report. Dr. Zheng Yarui from Tencent quantum laboratory made a report "quantum computing, we also need to do?", And Tsinghua University tenured Associate Professor Kihwan Kim (Jin Qi Huan) did a "Quantum Computation with Trapped Ions" report.
In addition, Zhongshan University professor Lee Green Week Data and Computer Science, National University of Defense Technology Dr. Xiaogang strong as a special guest also took the floor. Tianjin University of Technology Professor Luo training, Posts & Telecom Press senior editor Herui Jun Planning chaired line.
AI Technology comment on the report, "Overview of the development of quantum computers and prospects" of the sort has been released, now Tencent quantum laboratory of Dr. Zheng Yarui "quantum computing, we also need to do?" After finishing the report has now released, Please enjoy ~
Zheng Yarui: Dr. graduated from the Institute of Physics. The main field of a superconducting quantum computing. In the design and simulation of superconducting quantum chip, a quantum microfabricated chip, high-precision control of qubits, qubits accurately read, quantum couplers, hybrid aspect quantum system has many years of research experience. Hardware is currently head of Tencent quantum laboratory is committed to promoting the development of quantum and landing technology.
In the report, Dr. Zheng Yarui introduces the development of key issues precision measurement and control qubits, quantum error correction and NISQ such as quantum computers, and shared his views on sustainable development of quantum fields.
When put to practical quantum computer?
Dr. Zheng Yarui think quantum computing, it may take a decade to realize valuable commercial floor at least, Google declared a goal of 1 million qubits in 10 years in fact, some radical, but given the strength of Google's powerful, does not rule out 10 years possibility of realization. In 2018, IBM declared that their "Quantum volume" may be doubling every year, Quantum volume corresponding to the effective number of bits of quantum bits, double the annual task of comparing Google's decade one million bits very close. Honeywell ten times each year to enhance the performance of propaganda than Google, IBM more radical, if ten dimensions to calculate, Honeywell is about 10 billion times the performance, which is a bit too much publicity.
The computer calculated using classic low voltage represents 0, 1, in which the quantum bit, the computing quantum states. Qubit itself to follow the superposition principle of quantum mechanics, the quantum bits can be in a superimposed state 0 and 1, N qubits can be in fact on the 2 ^ N quantum superposition states.
Therefore, the number, and the number of quantum states of qubits growth was exponential relationship. In 1980, Professor Feynman would ask: With the growth in the number of quantum bits, its complexity grows exponentially, can directly use some of the problems of quantum mechanical calculations of quantum mechanics? The concept of quantum computing from that time gradually evolved into a more systematic discipline.
How to make a quantum computer
How to make a quantum computer it? There are three requirements, first of all have qubit; then there is a high fidelity quantum operations, called quantum operations, which can be likened to a non-classical computer doors, doors with these basic NOT gate or door operation; the last to perform algorithm requires enough computing quantum bits.
These three issues seem simple, but doing it is not simple. First of all, what materials can be used as quantum bits? Now humans have very many excavated quantum computing vector comprising ultracold atom, ion trap, photons, superconducting qubits, can be used for quantum computing. These systems have their advantages and disadvantages, it is not known what kind of system must be able to ultimately quantum computation, but relatively fast development of superconducting qubits and ion trap.
Superconducting Qubit most important advantage lies in its configuration similar to the circuit, it is possible using the traditional large-scale integrated circuit technology allows fast qubit system manufacturer.
This technique is particularly fast development in recent years, borrowing IBM and Google study, the left can be seen from the above superconducting qubit decoherence time is about two to three years will be doubled, and now the performance of the entire quantum bits than ten years ago a lot of good. In 2005, the decoherence time is 1 microsecond level, the qubit decoherence time now has hundreds of microseconds. Increase the number of bits is also very fast.
2019 Google released a chip with 53 qubits, in fact, in 2017, when Google has 72 bits of quantum chip, 53 compare the number 72 does not appear to increase, in fact, the complexity of a lot more than before.
To some extent, the qubit 53 at the chip level can be achieved more than 100 bits. In addition to the decoherence time and the number of qubits, fidelity superconducting qubit gate operation is very high, in 2014, has reached 99.4%, Google the latest quantum chip fidelity will be higher, there may be 99.6% more than the correction threshold.
Looking back, how to build a quantum computer? First we have a quantum bit, and then to have high fidelity quantum operation, a sufficient number of quantum systems. Superconducting now generally able to meet these three requirements, this technology has reached a threshold.
Quantum computer development path
Figure from IBM, describes the possible paths behind the development of quantum computing. From the bottom up the entire box superimposed, the lowest level for the Physical Quantum Processor, is the concept of the physical level.
First of all have a quantum qubit chip, with qubits, and how the control, followed by a quantum bit reading. After reading the control and achieve high fidelity, and do quantum system Quantum error Correction, i.e. correction quantum, the main purpose of this operation is to further enhance the control accuracy of the quantum system.
In fact, more than 99% of the control accuracy is not enough to achieve practical application of quantum algorithms. But scientists are very smart, since direct door to make high fidelity is not easy, so why not use a lot of qubits it? The so-called quantum error correction is to learn from the classic concept of computer error correction to ensure that the final to reach the total equivalent of quantum operations, can achieve high fidelity.
On this basis, can rely on a large number of qubits to achieve very high fidelity quantum operations, with more qubit quantum algorithm to achieve valuable floor, went to the Logical Quantum Processor, is the logical level of quantum processor. In the same logic level required to break through the technical and physical aspects of quantum processors, the need to control and read, and finally to achieve valuable quantum algorithms.
From the point of view the ultimate goal, to achieve the landing of quantum computing applications, in theory, need to reach about one million qubits, quantum error correction to do after this number, more than 1,000 will form about the size of a logical qubits, this scale sufficient floor applications.
From the data, single-bit gate 99.99% accuracy required, two bits door needs 99.9% accuracy, reading need to achieve very high fidelity. Currently two-bit gate has to be 99.4%, fidelity has been very close to the target algorithm landing. Next important issue is that while increasing the number of bits, make sure the door fidelity quantum bits will not fall, which is the main focus of the current academic circles.
How to solve the problem of double the number of bits and fidelity?
How to solve these two problems a technical point of it? Start angle superconducting qubits, superconducting circuit was observed conversion process. FIG upper left is a simple diagram, the lines can be considered a superconductor lines, there is a ring of the following circuit, there are two small squares yellow rings above, two small blocks called Josephson junctions, it can be used composition SQUID, called SQUID.
FIG rightward on two small peaks can be seen, the particles can be considered as a peak in the inside of the rolling system, then the quantum system, the energy level can be split into a series of sub-particles, use can be these quantum level operation. Quantum necessary to control the operation of reading.
The figure is read yellow mark (Re-out), it is considered that an input line and an output line to the state of the qubit can be read. Videos single bit circuit configuration is easy, but if there is a row, only the reference printed circuit structure, each of the quantum calculations performed by designing a row of qubits.
Obviously a row structure is not suitable for practical applications, but also two-dimensional array structure, but a two-dimensional array structure qubit control line and reading difficult. This is also the reason the number of qubits upgrade will encounter bottlenecks. The basic idea to solve this problem is to increase the likelihood of realization dimension of wiring.
In 2014, Google already had this idea, by drawing on traditional semiconductor processes (Flip-Chip technology, flip-chip) chip docking the two buckles together to achieve wiring. On the top right corner is the Google Sycamore processor, which is composed of two chips bonded to each other, is a bit above the chip, the chip is below the wiring.
Google implements control and reading two-dimensional bit array in this way. This concept can be further developed, not only limited to two chips, the chip can be a lot of layers bonded together, this is a multi-layer stack technology, MIT Lincoln Lab has achieved the technology. Therefore, in such a technological breakthrough, based on the wiring might not be a stumbling block hindering the expansion of the number of bits.
Another problem is how to improve bit precision operation of the door. At present, the operation is limited to the accuracy bit gate decoherence time, simply, how this problem has been converted into the number of bits while lifting, it is possible to enhance the coherence time of the qubit.
According to studies, it is considered Superconducting Qubit decoherence mainly due to the influence TLS, TLS is a two-stage system (Two level System), the substrate and the surface of the superconducting quantum defect of the chip may have some impurities, or capable of forming a metastable semi-level system, can interfere with the operation of the superconducting qubits. Lifting superconducting qubit decoherence defects need to remove impurities, the SQUID chip circuit made more clean decoherence performance becomes better.
So, from the roots, it needs to be improved from both materials and processes, superconducting quantum computing from the perspective of the development of the material industry has done a lot of innovation and improvement in two to three years has doubled the decoherence time.
In the superconducting qubit just appeared when the material did not realize problems. Because superconducting qubits appear very late, so far, it was only 21 years old. This year is the superconducting qubit discovered 21 years, its performance decoherence from less than a microsecond, the rapid increase to two or three hundred microseconds, must rely on renewable materials and processes.
Early qubits are used in very traditional materials, semiconductor materials are also frequently used qubits choice, but relatively poor performance decoherence, less than a microsecond. For this reason, researchers at the time of the superconducting quantum computing does not have high hopes.
So, the idea is to do basic research, basic research problems of quantum mechanics, quantum computing if it be used or there is a gap. In 2007, researchers change a capacitor structure. 2013, 2014, Google has made a lot of technology innovations, including the design also made some updates. Decoherence time to elevate to the level of one hundred microseconds.
In this, scientists have developed new materials, such as tantalum metal which may be the bits to further enhance the decoherence time of the order of several hundred microseconds. This is very encouraging for researchers, superconducting quantum computing development time is short, but during the 21 years of development, has maintained a very fast rate of progress. Currently, the new technology, new materials has not reached the bottleneck, such as ultra-high vacuum can be packaged so that more clean the surface circuit.
For lifting the superconducting qubit decoherence time, researchers optimistic expected to enhance the level of no problem in milliseconds. Ms is what concept? Means a quantum bit, a single bit can be five nines (99.999%) fidelity, it is possible to achieve double-bit 4 9 (99.99%) of fidelity.
According to quantum algorithms currently known, really need to have a real value scale of 1 million bits, 1 million bits too far, at least ten years to achieve this goal, counting on decades of technological breakthroughs, industry development very negative.
So researchers consider: 100-1000-scale quantum chip, can achieve valuable applications it? FIG blue upper region, has been achieved in determining the quantum Google supremacy when 3 come from 2 Google may be the way to go next. Between 2 and 3 Google defined as recent applications, the main purpose is in the middle of 2 and 3, try to find valuable applications.
Quantum Computation and classical computation or synergistic
People often spend diagram to calculate the quantum and classical computing comparison, Google sometimes fried concepts such as quantum computer challenge traditional supercomputers. At present, the traditional supercomputers is very powerful, much stronger than we currently have the capability of quantum computers. From the point of view of confrontation, a quantum computer within a decade may be hopeless.
If you change a point of view, the purpose of the advantages of quantum computing is not a challenge classical computer, but synergistic. Ability to overcome difficulties before simply relying on classical computers, the direction is really should think about.
How to further development of quantum computing?
First, the core key issue is to enhance the accuracy of measurement and control qubit;
Second, quantum error correction step using a quantum computer is more critical;
Third, to achieve one million before the qubits, the quantum scale 100-1000 bit of research to do NISQ, this time inspired by the prototype needs;
Fourth, the issue of sustainable development of quantum field should pay attention, because some of the view that a quantum computer is certainly not made out, this will damage the field; can not be too optimistic about the company's commitment to investors or government investors too, if the actual not so fast technological development, may make investors disappointed, it will destroy this area;
Fifth, the calculation of quantum computation and classical relations relationship is non-confrontational, quantum computing technology from many classic computing;
Sixth, quantum computing technology in the development process, companies have a very important role, as businesses and practical applications closer together, may bring a different idea of quantum computing to help find the true value; seventh, businesses and research institutes should be more open, quantum computing may not be able to do valuable commercial landing in the short term, at this stage, too much emphasis on technological protection may not be conducive to the development of technology.
