This article is from WeChat public account:Quantum (ID:QbitAI), author: policy side, chestnuts, fish, sheep, Lee, cover: Google blog
Nature’s latest cover story
Google has broken a small step and human science has made a big step.
This breakthrough was personally announced by the CEO, and the paper was published on the 150th Anniversary of Nature, the headlines of major mainstream media, and the hottest first in the whole network. Even Trump’s eldest daughter Ivanka could not bear it. Live the first time to send a congratulatory message:
“Official Announcement. US RealizationQuantum hegemony!
Yes, that is the quantum computing 200 seconds = the world’s strongest super-calculation of 10,000 years of breakthrough, now Google in the most solemn form of the official announcement, more than the computer recognition of the cat, the limelight over the AlphaGo, turned out.
And Google’s CEO is still excited to introduce, this is like the moment when the plane was first invented – the Wright brothers’ plane flew for only 12 minutes for the first time, but it proved the possibility of the plane flying.
This is a historic moment, and Google has revealed for the first time that it has been burying itself for 13 years.
The first realization of quantum superiority
Quantum Supremacy, the “quantum hegemony” in Ivan’s mouth, Google is more inclined to translate into quantum superiority.
Intuitively, quantum superiority means that at some point in the future, a powerful quantum computer can accomplish tasks that are almost impossible to accomplish with a classic computer.
For example, in a day, crack the password that was originally tens of thousands of years to crack, implement general artificial intelligence, and quickly simulate molecular models.
The previous milestone breakthroughs were at the stage of envisioning and never been realized. But a month ago, Google’s draft “unexpected” was released on NASA’s official website. The result was very shocking, saying that the 200-second quantum calculation achieved the strongest super-calculation of 10,000 years.
Unexpectedly, the paper was rushed off the shelf, but it caused more attention.
The public opinion was blown up at the time. Some people thought that Google was a “god stick”, and some thought that they were removed from the issue of US national security…
More competitors attacked directly, and IBM specifically published a paper questioning Google’s “misleading the public”. And now, the paper is positivePublished in Nature, Google CEO Wei Chai Ge proud official Xuan Google AI team achieved quantum superiority, also highlighted in the blog:
“Just like the first rocket successfully escaped from the gravity of the earth and flew to the edge of space. This breakthrough shows us what is possible and pushes things that seem impossible to achieve. This is the significance of this milestone to the world of quantum computing – a moment full of possibilities.”
Pui Chau also said, Google has been working hard for 13 years, and the company was frustrated because of limited progress.
Thirteen years ago, in 2006, Google scientist Hartmut Neven began exploring a new idea—using quantum computing to speed up machine learning and spawning the Google AI Quantum team.
Next, in 2014, the American Physical Society academician John Martinis joined Google as the chief scientist of Google Quantum Hardware, leading the work of building quantum computers.
Two years later, Sergio Boixo, chief scientist of quantum computing theory, published a paper on Nature Communications, which eventually focused the team’s work on quantum dominance computing tasks.
This is a marathon for research, everything starts from scratch. Even for Google’s star team, such a job is a huge challenge.
In fact, before October last year, Google’s progress in quantum superiority has been limited.
However, I never imagined that the wildfires in California could not be extinguished in October 2018. For security reasons, Google had to close the lab in Santa Clara for a short time, and a group of scientists were also forced to take vacations.
But during this periodInstead, it gave birth to new ideas and then realized a real transition.
Pui Gu also feels that quantum computing is not a clear future. It is not easy to believe and firmly believe that it can be achieved.
But Google has always believed that quantum computing can accelerate the resolution of some of the world’s most pressing problems. Quantum computing provides an unprecedented opportunity for humans to understand and simulate nature on a molecular scale.
Pui Chau also said that quantum computing will be a great complement to the work people do on classic computers, and quantum brings a complete cycle to computing.
However, even now that the milestone has arrived, Pigues has warned that this is only proof of the feasibility of the direction, not the moment when mass commercialization is effective, we need to continue to attack and need to continue Go ahead.
“Unintentional” leaks
As mentioned at the beginning, Google’s research on quantum superiority appeared briefly on NASA’s official website before Nature’s publication.
So the question is coming, why is NASA released? Why did you withdraw it soon after it was put on the shelves?
According to the Financial Times, Google has cooperated with NASA last year and set up flag: The quantum superiority of 2019 is to make the computing power of quantum computers far exceed that of classical computers, and to complete calculations that classical computers cannot.
The plan is to compare the results of running on a quantum computer with the classic simulation (simulating quantum circuits on a classic computer).
The quantum chip used by the two parties is called Bristlecone, which has 72 qubits. Bristlecone had to keep the superconducting circuit near absolute zero, so it was impossible to move away from Google’s labs. NASA researchers could only remotely connect chips via the Google Cloud API.
According to the agreement, the two parties coded the software needed to run the simulation on NASA’s strongest supercomputer Pleiades in early 2019, and compared the results of quantum circuit simulation and Google quantum computer hardware in July this year.
Google and NASA have always been optimistic, but some people in the industry think that this flag will fall. Researchers at Alibaba’s Data Infrastructure and Search Technologies division have published a paper suggesting that achieving quantum superiority may require quantum chips with lower error rates.
Daniel Lidar, director of the Center for Quantum Information Science and Technology at the University of Southern California, also expressed doubts. In his comments on MIT Technology, he said: “ (to achieve quantum superiority) seems to need other ways to suppress errors.”
If peer review is passed, it means that the flag has not fallen, and quantum computing will enter a new era.
So why did NASA release the paper soon after it was released? Fortune magazine reported that the research results have not been fully peer reviewed.
Now only a month later, the paper has been on the Nature, the credibility has greatly improved, and the new era of quantum computing has really come.
So what is the specific paper content?
Details of the paper
After NASA withdrew the paper for a month, Google finally published the paper on Nature.
Google said in the abstract:
“We use a programmable processor with 53 superconducting qubits, occupying a state space of 253≈1016. The results of repeated experiments will sample the corresponding probability distribution.”
The bits in a classic computer can only be in 0 or 1 state, and Schrödinger tells us that cats can be superimposed in both dead and live states, as are qubits, and can be at 0 and 1 at the same time. Two states.
One qubit can only represent two states, two qubits can represent four states, three qubits can represent eight states, and so on.
Because the state of an object in quantum mechanics evolves in this superposition state space, coupled with the coupling between different qubits, more states can be simulated.
So only 53 qubits can simulate 1016 states, and this number is beyond the computing power of today’s supercomputers.
After finishing the basic concept of quantum computers, let’s take a look at the hardware of Google Quantum Computer.
Google calls this quantum processor that achieves quantum superiority as Sycamore. It consists of a two-dimensional array of 54 transmon qubits, each bit withThe surrounding 4 bits are coupled.
The above image shows the layout of the Sycamore processor, which contains an array of 54 qubits (indicated in gray ×), each rectangle Connect to its four nearest neighbors via the coupler (indicated by blue squares).
The appearance of the entire processor is very similar to that of a normal CPU chip.
The processor is fabricated from aluminum to achieve a Josephson junction in low temperature superconducting and uses indium to create bumps between two silicon wafers. The chip is wire-bonded to the superconducting circuit board and cooled to less than 20 mK in the dilution refrigeration unit.
This temperature is only two percent higher than absolute zero. The reason why it is so cold is to reduce the ambient thermal energy to much lower than the quantum potential energy to prevent external heat from interfering with the quantum processor.
The processor is connected to the room temperature electronics via filters and attenuators that synthesize control signals. The state of all qubits can be read by techniques that utilize frequency multiplexing at the same time.
In order to fully control this quantum processor, Google has also designed 277 digital-to-analog converters.
So what is the problem with Google, using quantum mechanics, and such a super-complex quantum hardware?
It is precisely a quantum circuit sampling problem that is not well solved by classical calculations. On this issue, the computing power of classical computers is stretched.
Each time a random quantum circuit is run on a quantum computer, a bit string is generated, such as 0000101. Due to quantum interference, like the speckle formed by the laser after passing through the slit, (sampling), some bits Strings are more likely to appear than other bit strings.
However, with the number of qubits n(width) and the number of gate loops m (depth), it has become increasingly difficult to find the most likely bit strings for a random quantum circuit with a classic computer.
In the experiment, Google first ran a random simplification circuit of 12 to 53 qubits to keep the circuit depth constant.
After verifying that the system is up and running, Google runs a random hard circuit with 53 qubits and increasing depth. When the depth m is increased to 20, the classic simulation becomes completely unusable.
It takes 200 seconds to get a million samples on a quantum processor, and the classic fidelity of equal fidelity on one million cores will take 10,000 years, and the verification of fidelity will cost hundreds of Million years.
Google only shows an application of quantum computers in the paper, which can be used in the future to solve problems including quantum physics and quantum chemistry simulation.
The breakthrough in quantum computing can also lead to new applications of machine learning that accelerate the resolution of some of the most pressing and complex problems the world is facing. For example, simulations of climate change, such as exploring which molecules can make more effective drugs.
IBM dissatisfied, accusing Google of flicker
Many technology companies are engaged in quantum computer research, including traditional IT giants such as IBM and Microsoft, and Internet companies like Alibaba.
On the night before Google officially published the paper, IBM chose to fight with Google.
The Blue Giant said that Google’s claim about quantum superiority is flawed. Because Google is actually competing without fully utilizing the full capabilities of supercomputers.
For this, Google declined to comment.
IBM’s words can be interpreted in a colloquial manner, that is, Google adjusted the baseline. In the original paper, Google’s quantum computer solved the problem that the original supercomputer would take 10,000 years to solve in just 200 seconds.
But IBM said that this problem is not as difficult as Google’s publicity. If there is time to optimize and improve, then it will take only 2.5 days to solve the problem.
Although this time is still much longer than that required by quantum computers, it is far from being out of reach.
In IBM’s view, the so-called quantum superiority is to do things that classic computers can’t do, and Google obviously didn’t do it.
Jay Gambetta, an IBM quantum computing researcher, said the company was not trying to confront Google, but to avoid over-promoting the term “quantum superiority.”
Some people think that Google achieves quantum superioritySex is not important, and it is not important for IBM to argue with Google. What is important is that under the competition of these giants, quantum computing technology is rapidly developing in a process beyond our imagination.
Two chief scientists lead
Sergio Boixo is the chief scientist of quantum theory at Google AI Quantum Labs and teaches at the Department of Electrical Engineering at the University of Southern California.
His research areas are Bose-Einstein Condensation, Quantum Information, Quantum Computing, Quantum Communication, etc. He has published 84 papers in his field, totaling 3259 times.
John Martinis from Google AI Quantum Lab is the chief scientist of quantum hardware. He was a Fellow of NIST and a Fellow of the American Physical Society and currently works at the University of California, Santa Barbara.
John Martinis graduated from the University of California at Berkeley where he earned two bachelor’s and doctoral degrees in physics.
He later joined the National Institute of Standards (NIST), during which he also invented a series array of superconducting quantum interferometers. (SQUID) amplifier.
In 1993, he set about building a high-resolution X-ray microcalorimeter based on a superconducting sensor and a tandem array SQUID. This work has been developed to include applications in X-ray microanalysis and astrophysics, as well as optical and infrared astronomy.
In 2010, he was awarded the “Scientific Breakthrough of the Year” award for his first demonstration of the quantum ground state in a mechanical oscillator system.
In 2014, he was awarded the London Superconducting Quantum Bits Low Temperature Physics Research Award, and in the same year joined Google’s leadership in quantum hardware research.
One more thing
At the end of the last, at this historic moment, let’s end with some answers from the MIT Technology Review interview:
I am excited about such a milestone, but I really want to remind everyone that this is just the beginning of a possibility to be verified. It takes a long time, even 10 years, from a real large-scale change, but we can I am optimistic about the speed of technological development.
When we look back at Deep Blue in 1997, defeating Kaparov and AlphaGo to defeat Li Shishi in 2016, the time required for a breakthrough is not short, but when such a time comes, more people will join and join. We humans are progressing like this.
And it’s so exciting now, we’re in the technological cycle of time, AI will accelerate quantum computing, and quantum computing will accelerate AI, the big challenges and big problems that humans encounter, Now there is a chance to be solved.
Transportation:
Google Blog: https://www.blog.google/perspectives/sundar-pichai/what-our-quantum-computing-milestone-means/
ThesisOriginal: https://www.nature.com/articles/s41586-019-1666-5
IBM papers: https://arxiv.org/pdf/1910.09534.pdf
This article is from WeChat public account:Quantum (ID:QbitAI), author: policy side, chestnuts, fish, sheep, Lee