After announcing quantum superiority last year, Google finally ushered in another major development – the first time to simulate chemical reactions using quantum computers. < / P > < p > in order to achieve this latest achievement, the researchers used the sycamore processor to simulate the isomerization of a diazene molecule consisting of two nitrogen atoms and two hydrogen atoms. Finally, the quantum simulation is consistent with the simulation carried out on the classical computer, which verifies their work. < / P > < p > it is worth mentioning that the sycamore processor used in this new research made a worldwide sensation in 2019, and in October last year, it helped Google quantum team’s research to appear on the cover of the 150th anniversary edition of the magazine. < / P > < p > in the paper on sycamore, it completed the target computation in 200 seconds, which required 10000 years of continuous computation on the world’s fastest supercomputer at that time. As a result, Google has announced the quantum advantage of proving that quantum computing can handle some problems better than classical computers. The 76 authors of the paper say it is only a step away from the valuable short-term applications of quantum computers. < / P > < p > after the publication of the paper, Google CEO petraey said in an exclusive interview, “this incident is like the Wright brothers invented the airplane. Although the first test flight of the plane lasted only 12 seconds, which seemed useless, it proved the possibility of the plane flying. ” In fact, according to the existing physical theory, the universe in which we live follows the quantum laws at the most fundamental level. Therefore, the early application of quantum computing can help us better understand the way the universe works, and gradually realize the accuracy of molecules and intermolecular interactions according to quantum physics Simulation plays an important role in important research fields involving chemistry, such as medicine and carbon emission control. < p > < p > 10 months later, Google used quantum computers to simulate chemical reactions for the first time, which is also the initial landing of Pichia’s original vision. < p > < p > for the results, Google researcher Ryan babushi said that although the chemical reaction may be relatively simple and not impossible for non quantum computers, the work is still a big step for quantum computing. “We are now doing quantum calculations of chemical reactions on a completely different scale,” he said. The previous calculation work can be done by hand with pencil and paper, but the demonstration we see now must be completed by computer According to babushi, it should be easy to extend the algorithm to simulate more complex reactions, which require only more qubits to simulate reactions in macromolecules, and then fine tune the calculation. In the future, he said, we can even use quantum simulation to develop new chemicals. According to the quantum mechanical law of chemical reaction process, accurate calculation and prediction of chemical reaction process can unlock new chemical field and improve existing industry. Unfortunately, due to the exponential expansion of the number and statistics of quantum variables, except for the smallest system, the exact solutions of all other quantum chemical equations are still unable to be obtained by modern classical computers. At the same time, atoms and molecules are controlled by quantum mechanics, so quantum computers are expected to be the best way to simulate them accurately. That is to say, by using quantum computer and its unique quantum mechanical properties to deal with the calculation which is difficult to deal with by classical computer, the simulation of complex chemical reaction process can be realized. < / P > < p > today’s quantum computers are powerful enough to gain significant computational advantages in some tasks, but previously, it was difficult for quantum computers to achieve the accuracy required to simulate large atoms or chemical reactions. In other words, whether the quantum computer can be used to accelerate the current chemical reaction quantum simulation technology is still an open question. < / P > < p > in this latest experiment, the Google team unlocked the app. They use a noise robust variational quantum eigenvalue algorithm to directly simulate the chemical reaction mechanism. < / P > < p > although the calculation focuses on the Hartree Fock approximation of a real chemical system, it is twice as large as the previous chemical calculation with quantum computers and contains more than ten times the quantum gate operation. It is worth mentioning that the current quantum algorithm can achieve the accuracy of experimental prediction, and open the way to the real simulation of quantum chemical system. In addition, the Google team has released the code for the experiment, which uses openfermion, an open source library for chemical quantum computing. < / P > < p > this experiment is based on the sycamore processor, which demonstrated the quantum superiority of Google last year. < / P > < p > although this latest experiment uses fewer qubits, a higher fidelity of quantum gates is needed to solve the problem of chemical bonds. This also promotes the development of new, targeted calibration techniques that can best amplify errors for diagnosis and correction. < / P > < p > errors in quantum computing may come from various sources in the quantum hardware stack. The sycamore processor has 54 qubits and consists of more than 140 independent tunable elements, each of which is controlled by high-speed analog electrical pulses. In order to realize the precise control of the whole device, more than 2000 control parameters need to be fine tuned. Even if there are small errors in these parameters, they will be quickly added to the final calculation and accumulate into large errors. < / P > < p > in order to accurately control the device, the Google team uses an automated framework to map the control problem to a graph with thousands of nodes, each node representing a physical experiment to determine a single unknown parameter. With this diagram, the team can move from a priori knowledge of the device to a high fidelity quantum processor, which can be completed in a day. < / P > < p > left: the energy of the linear chain of hydrogen atoms increases with the bond distance between each atom. The solid line is simulated by Hartree Fock with a classical computer, while the point is simulated by a sycamore processor. < / P > < p > right: each point calculated with the sycamore processor has two precision measures. Raw is non error mitigation data from the sycamore processor. “+ PS” is an error reduction type of data from the corrected number of electrons. “+ purification” is an error mitigation measure for the correct state. “+ vqe” is the combination of all error elimination and circuit parameter variation relaxation. Experiments on H8, H10 and H12 show similar performance improvement after error mitigation. < p > < p > in October 2015, the University of New South Wales in Australia used silicon to make quantum logic gates for the first time, nearly five years ago. This time is not short, quantum computers should have more explicit development, especially when electronic calculators have pointed out the direction for the development of quantum computers. < / P > < p > the core arithmetic logic unit design, control unit design, chip instruction system, compiler, programming language, and even software ecology are all ready-made. Quantum computers only need to follow the previous rut and copy operations step by step. In five years, it may not be realistic for everyone to have a quantum computer, but it should not be a problem for small-scale commercial use. < p > < p > Real qubits are far less stable than conventional silicon-based circuits. The qubits used by Google, IBM and rigetti are composed of micro nano resonant circuits etched by superconducting metals, although this hardware scheme is easier to manipulate and integrate than other types of qubits. Each quantum circuit has two definite energy states, which we can record as 0 and 1 respectively. By applying microwaves to the circuit, researchers can put it in one of the States, or any combination of the two states – say, 30% of the zeros and 70% of the ones. However, these “intermediate states” will disperse, or “decoherent,” in a very short time. Even before decoherence occurs, the noise may “collide” and change these quantum states, making the calculation result “derailed” and evolving in an unwanted direction. < / P > < p > in this case, when Google published a paper last year claiming quantum superiority, Greg kupperberg, a mathematician at the University of California, Davis, who is also an expert in the field of quantum computing, thinks that the goal set by Google is not the core of the problem. The physicist Chad rigetti, founder and CEO of rigetti, gives a vivid example of this. If you spend 100 million dollars to build a 10000 qubit computer, when the error correction problem is solved, it has great power. On the contrary, it is a random noise generator. In theory, there are many ways to use quantum computers to simulate the ground state energy of molecular systems. In this study, the Google team focused on the “building blocks” or primordial circuits of quantum algorithms and improved their performance through vqe. In the classic setup, the original circuit is equivalent to the Hartree Fock model and is an important part of the algorithm developed by the team to optimize chemical simulation. < / P > < p > this allows the Google team to focus on scaling up without spending a lot of simulation costs to validate the device. Therefore, robust error mitigation on this component is essential for accurate simulation when extended to the “beyond traditional” range. The error in quantum computation results from the interaction between quantum circuit and environment, which leads to wrong logic operation: even a small temperature fluctuation may lead to quantum bit error. The algorithms for simulating chemical reactions on quantum computers must solve these errors at a low cost, including the number of quantum bits and additional quantum resources, such as the implementation of quantum error correction codes. < / P > < p > at present, the most popular way to solve errors is to use vqe. The Google team chose vqe, which was developed a few years ago. It regards quantum processors as neural networks and tries to optimize the parameters of quantum circuits by minimizing the cost function and solve the noisy quantum logic. Just as the traditional neural network can solve the defects in data through optimization, vqe can dynamically adjust the parameters of quantum circuit to solve the errors in the process of quantum computation. < / P > < p > in a study published on June 8, Professor Andreas wallraf of the Federal Institute of technology in Zurich and his collaborators demonstrated that they could detect errors in a 4-bit square grid coded logical qubit with three auxiliary bits. As soon as the paper was published, it raised doubts that “manipulating each independent qubit will introduce a certain error. Unless the error can be lower than a certain threshold, entanglement of the initial bit with more bits will only increase more noise”, said Maika Takita, a physicist from IBM Say, “you have to try to get below that threshold before you demonstrate anything.” < / P > < p > auxiliary bits and other error correction devices