【Facebook Fans Collection】2021 Season 3

I finally avoided the hassle of having to move in early April, so I was free to start sorting through Facebook news and reviews. I have always thought that long texts are a more concise form than voice and video (of course not to the density of mathematical expressions XD), so it is necessary to organize scattered posts into articles. It's amazing that I don't have any crypto-related posts this season at all LOL

In addition to chronological order, I will also classify this type of records into different chapters for the convenience of readers. The content will be copied and pasted with fan-posted text, and reviews, comments and outlooks will be added (displayed in different fonts for everyone to read). As usual, everyone is welcome to share, forward, leave a message, applaud (even donate ).

Quantum Computing and Physics

Quantum hegemony experiment cracked again?

You may have the impression that at the beginning of the year, Pan Jianwei successfully performed an experiment of 76 photons in the quantum optics group of the University of Science and Technology of China ( Note 1 ), claiming to be the second quantum hegemony after the Google 2019 experiment. The name of the experiment is Gaussian Bososn sampling. (There is a Chinese team that takes the GPU to crack Google's quantum supremacy claim that we said skip).

Let’s talk about the theoretical background first. This experimental theory is better to explain (compared to Google). We need to imagine that stacks of beamsplitters are arranged in rows, and all photons will produce various reflections and transmissions after entering the beamsplitter array. The phenomenon of interference, interference, and finally the statistical properties of photons leaving the beamsplitter array can be very difficult to calculate with classical computers: because in principle any boson can be used to do this experiment, so this problem is called Boson Sampling (by Scott Aaronson proposed). The complexity of this calculation is lower than that of the famous Shor algorithm (it can be seen from the absence of a bunch of entangled states), but it is still beyond the scope of classical calculations. An important premise is to use "single-photon light sources for experiments", which basically Difficulty is guaranteed.

In practice, it is a very challenging experimental setup to have several single-photon light sources, so some people have turned the above Bososn sampling into Gaussian Bososn sampling that can be replaced by squeezed light. Note 1 The Science is of course using the latter for experiments. , it turns out that the complexity of Gaussian Bososn sampling is not as complicated as the predecessors imagined, and a theoretical challenger appeared immediately. A Russian team published a preprint in June ( Note 2 ), pointing out that if the squeezed light accuracy of the actual experiment is considered, They found a way to use a laptop to simulate seemingly "unsimulated" experimental photon distribution patterns.

In fact, I wrote these things just to share with you that quantum information is still in a very early stage, and it is often possible to claim a slight advantage, and then inspire information scientists or theoretical physicists to improve existing algorithms, and classical computing does not Not the oranges that have been squeezed out, the quantum supremacy/quantum superiority that can be achieved in the NISQ era will be much more difficult than everyone thinks.

In many cases, whether quantum computing is really speeding up is a very subtle issue. In addition to the classic events such as HHL, which can be clearly compared from the complexity of the algorithm, the "quantum optimization" and "quantum machine learning" that are expected in the NISQ era are very much. It's more like learning by doing and trying. It is more difficult to say that there is a rigorous speedup. Although it is not impossible to have a miraculous effect, it is more of an accident than a rigorous result.

40 years (or 25 years?) of quantum computing

Recently, John Preskill, a major in the field of quantum computing (who defined NISQ after 2018 and influenced the research direction of the entire field with one person) wrote an article "Quantum computing 40 years later", starting from Feynman to write the entire historical development of quantum computing, which is This is a popular science-level article with no mathematical difficulty and is very fluent and easy to read (mainly talking about concepts and reviews, and a little mention of quantum Fourier transform QFT). It is recommended that interested friends can read it to build a framework (see link ).

However, I personally always think that "quantum computing has a history of 40 years" is a very recent historical invention propaganda, although Feynman has called for discussion, quantum communication was theoretically constructed in the 1980s, and quantum Turing machines and some interesting algorithms are already in It was proposed before 1995, but without Shor's two major contributions, quantum computing should have been just a niche toy and will not become one of the mainstream of physics. Perhaps it can be said that quantum computing has been in 25 years (laugh. Shor not only used QFT to construct the Shor algorithm in 1995, but also quickly proposed the earliest quantum error correction system (Quantum Error Correction). It can be said that the two major problems of quantum computing are in He unraveled, 1) What interesting and great problems could we solve if we had a perfect large-scale quantum computer? 2) Can we build perfect large quantum computers from imperfect physics experiments?

"Shor is not born in the sky, and eternity is like a long night." In a sense, after Shor answered the above two questions, even if there are still many interesting problems related to quantum information and quantum computing in basic physics, the commercialization of quantum computing is hampered. It has become only the challenge of applied physics and various engineering (it can be said to be the main theme of the research direction that makes the government pay).

I still think that the history of quantum computing should be 25 years (from 1995), Feynman, the famous physicist, can understand that everyone wants to cling to the story, but the actual quantum algorithm and quantum error correction It is the core of defining this field (not speaking and telling stories. But Feynman himself said that "history of physics" and "history learned by physicists" are two sets of things, so later generations invented his quantum The calculation of prophetic status is not surprising.

On possible fundamental obstacles in quantum computing

This time I will share a less talked-about aspect. In the theoretical challenges of quantum computing, people may often hear various discussions about exponential acceleration (Shor) and polynomial acceleration (Grover), but in fact, these two algorithms are both They have quite good properties: the initial quantum state (input quantum state) required by their algorithm is easy to handle, and the efficiency of similar computing will not be inferred. A quantum computer starts with all qubits in the ground state (white paper). Before starting to run various algorithms, quantum state preparation must be done first.

In fact, this step can also be regarded as a kind of quantum logic gate, but there is a huge difference between theoretically being able to prepare and being able to "efficiently prepare". If the complexity of preparing the quantum state is too high, the computation may not make sense if it is fast (how much water a barrel can hold is determined by the shortest board). Even if QRAM (quantum RAM) exists in theory, it is uncertain how long it will take for this field, which is far more primitive than the quantum computer itself, to mature and be implemented. The rapid and efficient fabrication of arbitrary quantum states is by no means an easy or obvious challenge. Although this challenge does not affect Shor and Grover, these two classical algorithms will not be practical within 10 years. I personally estimate that quantum simulation may require various magical initial quantum states. In this case, logic gates may not be used. (too slow), an adaibatic approach (adiabatic process) or a VQE evolutionary approximation of quantum machine learning may be more promising.

It is really not easy for the great Shor and Grover algorithm to use a simple initial state as the initial state of the algorithm. In the field of quantum machine learning, complex initial quantum states can be said to be swept under the carpet. However, even if everyone is tacit, there will still be reviewers who are very thoughtful (or deliberately trying to poke you) when submitting papers to mention the problem of quantum state preparation. At this time, you can only increase your politeness and rhetoric ability by 10. times, so that the reviewer is happy to have the opportunity to publish the paper smoothly.

The biggest news in physics this week is the death of Nobel Laureate Weinberger

He is famous for his electroweak theory in quantum field theory (that is, the electromagnetic force and the rhombic force that unify the four forces), and the pursuit of beauty and symmetry, harmony and unity after quantum field theory can be said to start from here , although we still do not see a grand unified theory (unified electroweak and strong force), supersymmetry (unified fermions and Posons), final theories (like string theory hopes to unify the four major forces plus supersymmetry), nature It may not be ultimately described by the above-mentioned dream, but Weinberg's contribution is definitely a "profound greatness" in Landau's 1st grade.

Then I think Weinberg is also worth thinking about his discussion of science and value, and his views on religion, science is not only technical work, because even if science cannot affirm value, so cannot deny it, it does not represent scientists/science Civilization cannot be valuable. In the book "Science vs. Cultural Rivals", I think his arguments are well written. And Weinberg also has a deep textual research on the difference between what actually happened and what people learned from textbooks. This part can be seen in "The Great Discovery: A Journey to Shape the World with Science". It can be said that my past The most exciting and well-documented work in the history of science I've seen in years (never outdated). Finally, I share two famous words from Weinberg, "Regardless of whether religion exists or not, good people in this world will do good, and evil people will do evil. But when good people do bad things, it has to rely on religion."" Even if it is impossible for science to make sensible people unreligious, at least one of its greatest achievements is to make it possible to become unreligious, and we should never shrink from that achievement".

Weinberg has made a great contribution within physics (the unified theory of electroweakness) and has shaped a whole generation/school of discourse (elegant symmetry). There are views on scientific civilization. As an atheist, Weinberg is not as aggressive as Dawkins, and his position that scientists and scientific civilization should deal with value issues is calm and firm, and he is indeed a great man.

Optical quantum computing startup PsiQuantum gets $450 million infusion

PsiQauntum, a quantum computing hardware company, has two special features. The first is that he plans to bypass the entire NISQ (medium-sized, noisy quantum computer stage) directly and move directly to the Holy Grail of error-correctable quantum computing. It is expected to be directly Millions of qubits (currently, ion traps and superconductor giants both plan to have 1,000 qubits in 5 years), which is basically an unbelievable declaration, and besides the fact that the pie is very large, their hard Bulk technology is optical, photons that are very difficult to have nonlinear interactions. Although cooperation with Global Foundary can obtain the top process technology of semiconductors, I don’t think this can make up for the shortcomings of optical quantum computing (I hope they can show a first-hand boson sampling to prove quantum hegemony is fine). In any case, PsiQauntum raised US$450 million in the D round (the latest investment units include M$ and the super-rich Temasek), and the total raised funds since its establishment in 2016 has exceeded US$660 million, and then I really haven’t seen it. To what black technology came out. I feel extremely unknown about such fundraising, and my boss feels it is like the first half of the Internet technology bubble. There is an old saying in the market that the quantum industry may not be immune: "Markets are always born in despair, and grow in dubiousness. , mature in vision, and perish in hope.” I can only hope that PsiQauntum really has a killer product, otherwise he should be a milestone in this bubble cycle, and there will be no bigger investment cases (single startups are injected with capital , rather than a valuation of $600 million is no joke).

PS: Compared with the information released by PsiQauntum, I think Xanadu has more specific progress

I now really think that PsiQuantum will become an iconic bubble event like "Anlong Fraud", "Subprime Mortgage Crisis", "Bad Blood", and "Nicola Hydrogen Truck". "stage, beginning or already entering the "vision" stage, I sincerely recommend that you don't see the media publicity and feel that this industry is about to mature, and quantum computing must enter the next cycle (or at least the next decade) for it to mature.

Construction of the Abu Dhabi Quantum Computing Laboratory appears to continue in full swing

After reading this news, I want to emphasize the importance of small circles and networking in academia. In addition to being the chief scientist of the Abu Dhabi Quantum Research Center, Jose Ignacio Latorre is also the director of the Singapore Quantum Technology Center where I am currently studying. , the Barcelona cooperation case mentioned in the video is also related to him (Jose Ignacio was a professor at the University of Barcelona in Spain before arriving in Singapore, and he also opened his own startup company), so he traveled around Europe, the Middle East and Southeast Asia. The ability is really beyond ordinary people, probably only the United States, Japan and China can't get in LOL

It is said that Jose and the previous CEO Artur style is completely different. Artur is a very chill British professor. He came to Singapore for vacation, surfing and relaxation, nothing to worry about. Jose is interested in the research direction of each research group to know the progress, and calls different government departments and business units all day long, and participates in various conference activities, etc., so he can be the director of two quantum centers. I think it is completely possible to do it and deserve it, not to mention that although his new venture in Spain is said to be letting go, he is still the founder after all.

NSA's latest QA on quantum computing

I saw the latest document released by the NSA at the beginning of the month, and I read it later to share with you how the world and intelligence agencies are facing the latest progress in quantum computing. First of all, the NSA defines Cryptographically Relevant Quantum Computer, which is a quantum computer that is meaningful in cryptography, because there is no need to discuss the stuff in small laboratories. The second is that the NSA does not hold a position on how long the CRQC will exist or whether it will exist. The NSA is because the national security system must be kept secret for more than ten years, and it must be prepared in advance before investing in this issue (of course I estimate that they even The same is true when the large quantum computer is covered. LOL)

For QKD (Quantum Key Transmission): The NSA report believes that this thing is basically useless, and may have academic value, but it cannot be used for the protection of national security systems (even if it is theoretically safe, there must be loopholes in experiments) , and a completely different hardware infrastructure).

For existing cryptography: it is recommended to use symmetric encryption AES-256 for confidentiality preservation, and asymmetric encryption algorithms such as RSA/ECDSA are not quantum safe. (It is also recommended that you do not use unconventional encryption methods). In terms of digital signature, you can continue to use the hash base method of SHA-384 first.

For PQC (post-quantum cryptography): NSA is waiting for NIST to announce the final runoff algorithm, but NSA thinks there is a high probability that it will be using lattice encryption. (This is not surprising, it seems that PQC is lattice encryption will become a big winner)

To put it simply, I also believe that quantum communication (QKD) is basically a subfield of low commercial value. If there is no QKD, quantum satellites may still be able to do quantum sensing applications such as telemetry. The size of the cake is completely different. ． As for quantum computers, I think that after this wave of quantum computing hype passes (because NISQ cannot be used for large-scale applications) and PQC is mature and implemented, the entire quantum computer trend may be quiet for a while, and then the king will return in the next ten years? (There will be government intelligence agencies, big tech companies and high-quality new innovations in the middle of the winter, but it will not be possible to reach the sky in one step)

About the "Second Quantum Revolution"

Since backreaction wrote an article about the second quantum revolution, I'll share my thoughts on this discourse. First of all I have to say that this argument is very successful and important. Science may objectively rely on mathematics and experiments to create structures, but scientists are people and need stories to understand.

It can be said that before the 1990s, the heirs of quantum physics were quantum field theory and particle physics, and the standard model, symmetry and beauty seemed to be unquestioned. It can be said that quantum physics has been regarded as a solid and uninteresting cornerstone, and the original controversy has become an ancient past. If it weren't for John Bell's turnaround, I don't think quantum computing would be around until the 21st century (the physicist I admire most is Bell). After effectively attracting everyone's attention to entanglement, coupled with the evolution of experimental technology (A Aspect should have a Nobel Prize), quantum information only sprouted in 1980 (Quantum Turing Machine, QKD BB84), 1990 Breakthrough (Shor/Grover /Quantum Error Correcting Code). When the whole community began to pay attention to the intertwined integration of physical thinking and information science thinking, a new paradigm was finally born. The most abstract way of saying that a bit is a more pure and fundamental existence than atoms, or even space-time. Of course, specifically, quantum measurement and quantum communication technologies are mature (whether they are useful or not), quantum simulation and quantum computing cakes are very large, and resources can be temporarily injected from the government and enterprises. Quantum algorithms and complexity theory also feel that they can make Information science theory has played for several generations. It can be said that with the cooperation of all aspects, quantum computing finally has the opportunity to claim itself as the true heir of the golden age of quantum physics. 1920 was the first wave of revolution, and now it is the second wave of revolution (the image of particle physics is not in this historical narrative. pieces).

I later realized that "history of how physics actually happened" and "history learned by physicists" are probably very different things, and even if the progress of objective knowledge makes everyone forget the detours and detours, history has always been Not in a straight line. In the past half century or so, it can be said that the generation of particle physics that started with Weinberg shines brightly. The unity and symmetry of the field have become the keynote of the entire history of physical thought, as if it were the end of history (the first time I saw it). "Elegant Universe" and other books really think so, it can be said to be influenced). However, after the university began to study quantum physics/quantum computing, the framework of the history of physics in my mind underwent a paradigm shift. I later thought that the integration of the most important physical and information fields in the 21st century no longer takes the framework of the integration of the four major forces. XD

life in singapore

Singapore Epidemic Progress and Singpass Introduction

Singapore's National Day is today, and Singapore will also enter the phase of gradual unblocking from tomorrow, and restaurants will be allowed to use. At the same time, the scale of unblocked gatherings and activities will be distinguished according to whether the participants have been fully vaccinated with two doses of the vaccine (if both are vaccinated, more people can gather for dinner). At the same time, in order to avoid possible cluster outbreaks of the epidemic, NUS will require students who have been vaccinated to quickly screen themselves every month (if they have not yet been vaccinated, they will need to quickly screen each week), take photos and upload records, etc. All in all, this is a lot of personal health information scattered in different places, and it is very challenging how to process and present a bunch of privacy-sensitive information for verification (how does the restaurant verify that the guests at this table have received two doses of vaccines? ?).

Singapore has Singpass, a one-stop personal digital information station. After binding and logging in, you can see your personal ID card/visa, medical records (vaccination administration, clinic appointments), education records, pension records (CPF), immigration management Information (ICA), housing records (HDB flats), tax records (it is very inconvenient to pay taxes without Singpass), and there is a notice board for locals to find jobs (Singapore has regulations that many jobs must be opened to locals first 30 days to hire foreigners). Such an app is indeed very convenient, but I haven't applied for the past three and a half years because of laziness and avoiding a single point of failure for privacy explosion reasons (there are always some options to bypass Singpass), and until recently, the chain of vaccine data seems to have to be Through Singpass, I had no choice but to spend half a day going to the counter to apply for an account opening on work days. It's also interesting to see a novice challenge unlocked after completing 70% of the PhD career in Singapore.

Singapore's border control measures have finally been significantly loosened after the eight months written above. Except that the number of people in restaurants has been increased to 10, and the wearing of outdoor masks has been changed to an encouragement system, the most important thing is that border control measures are basically complete. Lifted, all long-term residents and short-term visitors will no longer be disturbed by annoying travel restrictions (depending on the regulations of all relevant countries, of course). As for Singpass, I really haven't used it, it's an app that I haven't used since I applied for it.

The original link is a physics apprentice staring at the singularity