19w4THdZDXm0mIOaKw2Jcfg

From test tube computers to qubits in the cloud

Carly Anderson
  • There are some applications beyond breaking internet encryption that quantum computers will be good at. We know that quantum computers will be particularly simulating nature (chemistry, materials, complex physical systems). Additionally, there are optimization problems that are “classically hard but quantumly easy”, and identifying more is an active area of research.
  • The first quantum computers that appeared in the 2000s were simply chemicals in a test tube, programmed with radiofrequency pulses (similar to an MRI scan).
  • Since then, scientists. engineers and researchers have made qubits (quantum bits) out of many types of particles — single atoms, single ions, electrons, or photons (particles of light), and from the relationships between particles. Quantum computers based on all of these are being developed established companies (Google, IBM, Microsoft, Honeywell) and startups (e.g. Rigetti, D-Wave, IonQ, Xanadu, Atom Computing)
  • Quantum computing will be primarily offered as a cloud service, at least at first. Major quantum computing companies, including Microsoft and IBM are already offering cloud-based quantum development environments based on their hardware. Meanwhile, Amazon has created a platform to enable access to a range of quantum hardware companies’ technologies.
  • Thinking back on our limited expectations for how classical computers would be used in the early days of the transistor, we should expect that the applications of quantum computers will be much MORE useful than we currently imagine!

In my first post on quantum technologies, we covered a few immediate applications of quantum technology: cybersecurity and efforts to build the quantum internet. In this post, we’ll look at a bigger and longer term area of quantum technology— quantum computing. Although they may still be decades away, quantum computers have the potential for the biggest impact on human life of any quantum technology.

Why should we get excited about quantum computers? What applications (outside of breaking internet security) are they useful for?

Where will quantum computers have an advantage when the technology matures? In our previous post on quantum technologies, we talked about internet security and how quantum computers will be able to break today’s encryption protocols. We know of other problems that are hard for classical computers, but that we have already found quantum algorithms that could solve these problems easily (if a powerful enough quantum computer existed). Identifying problems that are classically hard and quantumly easy is a very active area of research, and this question is far from being definitively answered!

In the fall of 2019, Google succeeded in defining a problem that quantum computers could solve faster than classical computers, and declared quantum supremacy. [1] While the problem they chose wasn’t that useful, the key point that I took from this announcement was that this was just the first example. More demonstrations of quantum computers solving specific problems faster than today’s computers will follow. At some point, it won’t be specific problems, but entire classes of computations.

In general, quantum computers are good at understanding nature. Experts generally agree that quantum computers will someday revolutionize chemistry and materials science by allowing scientists to predict specific properties of molecules and materials.

Example of a nitrogenase enzyme that converts nitrogen to ammonia (Image by Jjsjjsjjs)

One example given by the Microsoft Research team is unlocking new methods of fertilizer production by decoding biology. Fertilizer is currently made through the energy-intensive Haber-Bosch process, using natural gas as an input. Meanwhile, bacteria are able to make the same fertilizer with very little energy using a nitrogenase enzyme, which is too complex to simulate with classical computers. Quantum computers should be able to precisely simulate how atoms interact with the enzyme, showing us how to do this reaction more efficiently.

Perhaps the best argument that quantum computers will eventually become ubiquitous is looking back at how people (specifically leaders in the field!) viewed classical computers in the early days. The inventors of the transistor and early computer startups could not have possibly imagined how navigation apps, internet search, online banking, video streaming, e-commerce, and social media would infiltrate and change modern life. Similarly, it’s hard to imagine what people might create with quantum computers.

“I think there is a world market for maybe five computers.”
Thomas Watson, chairman of IBM (1943)

So what exactly is a quantum computer? While classical computers have converged on very standard designs, the field of quantum computers under development is extremely diverse. Any “computer” which uses the quantum states of subatomic particles to store information qualifies. Let’s take a brief tour of the landscape.

The First Quantum Computers

The idea of quantum computers has been around since the 1970s and 1980s. [2] In the 1990s, researchers felt comfortable enough with the idea of quantum computing to start developing clever algorithms (ways to solve problems) on them, if they existed. Shor’s famous algorithm describing how quantum computers could factor large numbers, breaking RSA encryption was proposed in 1994. It’s fair to say that the math preceded the engineering.

The first quantum computer (with just two qubits) was built by a UC Berkeley/MIT/Los Alamos team in 1998. In 2001, Shor’s algorithm was first demonstrated on a quantum computer was demonstrated by scientists and Stanford graduate students at IBM’s Almaden Research Center across the bay in San Jose. These early quantum computers were simply chemicals in a test tube! [3].

Between 2000 and 2010, research and development of quantum computers accelerated rapidly. Breakthroughs in controlling the quantum states of atoms, electrons and photons created more options for designing quantum computers. In the past decade, the range of technologies in development, the number of qubits achieved, and companies active in quantum computing technology has increased exponentially.

How to Make a Qubit

A physical quantum bit, or “qubit” can be made from many different types of particles. It can be a single atom, a single ion, a single electron, or a single photon (a particle of light). A qubit can also be made from a more complicated system, like a very cold superconducting electrical circuit in which many electrons are moving. (Preskill 2018) More exotic types of qubits called topological qubits don’t store the information in the particles, but in how the particles interact with each other — their topology.

It’s not that hard to change the quantum state of a group of particles — we do this when we get an MRI at the hospital, or when a scientist does an NMR experiment (the average synthetic chemistry grad student probably does 10 NMR experiments a day to confirm what chemicals they made.) The HARD part is changing the quantum states of individual particles independently to create distinct qubits.

Back in 2001, the IBM team needed 10¹⁸ molecules for a 7-qubit quantum computer. The Google quantum computer that demonstrated quantum supremacy, Sycamore, has 53 physical qubits — progress! However, experts say that a quantum computer will need at least 10,000 qubits to be useful, so we still have a ways to go in terms of the number of qubits needed.

Interference patterns, by vbsouthern under CC BY 2.0

The second hard part is keeping the qubits in the new states long enough to do a calculation. This property is called coherence. Almost anything can disturb a quantum state: temperature fluctuations, vibrations, looking at it the wrong way. Quantum computer designers have to go to great lengths to isolate the qubits from the rest of the world.

Some errors are inevitable, and error correcting is another critical area of quantum computing R&D. [4] Quantum computers need thousands of logical qubits to do meaningful calculations. If many of the quantum computer’s qubits are subject to error, or rapidly decohere and lose the information they stored, it could take 1,000+ physical qubits to make a single logical qubit.

Today researchers and companies are trying a range of strategies to make physical qubits that stick around long enough to be useful. All of them are really cool and deserve much more detailed descriptions, but here is a 160 km overview:

  • Superconducting qubits are the most common approach. These qubits consist of a pair of superconducting metal strips separated by a tiny nanometer-wide gap that electrons cross, leading to quantum effects. The superconducting qubits are manipulated with microwave pulses. (SPIE 2020) These quantum computers require huge dilution refrigerators (very different from your frig at home!) to keep the qubits near absolute zero. IBM, Google, Rigetti, Intel [5] and others are making quantum computers with superconducting qubits.
  • Ion trapping is an approach that appeared around 2010 to make qubits based on semiconductor manufacturing techniques. The ions are trapped in free space by electromagnetic fields, which are generated by chips with tiny etched features. Lasers are used to manipulate the ions. Honeywell, IonQ, and Universal Quantum are working on this approach.
  • Quantum Computers based on Silicon Photonics use single photons in silicon photonic waveguides (i.e. pipes for photons) and are then manipulated using networks of optical components. PsiQuantum and Xanadu are working on this approach.
  • Neutral atom quantum computing (example) uses light to contain and manipulate the quantum states of uncharged (neutral) atoms. Atom Computing is working on neutral atom quantum computers.
  • Quantum annealers are a specialized type of “quantum computer” that is good for solving specific optimization problems. It takes advantage of quantum fluctuations (specifically quantum tunneling) to find the optimal solution over a set of discrete values. D-Wave is a notable company in this space, with the distinction of selling what many consider the first quantum computer in 2011.
  • Topological qubits are in theory more stable than qubits encoded in individual particles, but are also farther from commercialization (and more mind-blowing [6]). In topological quantum computers, the qubits are anyons (a little-known quasi-particle). Observing these quasiparticles was a huge physics breakthrough, and won the 2016 Nobel Prize in Physics (Washington Post explainer). Topological qubits also require the chips to be held at low temperatures. Microsoft is working on quantum computers with topological qubits.
“A full wafer of Intel quantum computers” by jurvetson under CC BY 2.0

How to Use a Qubit (Building Quantum Computers)

Coherent, high-quality qubits by themselves do not make a quantum computer. Quantum computing companies need to both build amazing qubits, and then build the infrastructure to interact with them. Below are the four basic things needed for a functional quantum computer:

  1. The qubits and things to contain them
  2. Hardware to control and measure the qubits
  3. The processor that determines the sequence of qubit operations and measurements, based on the instructions of an algorithm
  4. A non-quantum (classical) computer to allow users to interact with the quantum computer, store intermediate and long-term results, and handle access to networks.

All quantum computers are essentially quantum-classical hybrids — quantum hardware and a quantum processor that are controlled and programmed by a classical computer.

Notably, quantum computing will be primarily offered as a cloud service, at least at first. In the last few years, there’s been a race to provide public cloud-based quantum services. You can actually use one today, from the comfort of your living room. For example,

  • Microsoft has been building quantum computing services for over a decade, including a quantum computing service in Azure that can be accessed remotely by users. (The service currently simulates 30–40 qubits.)
  • IBM launched its “Q Network” in 2016 to provide access to the company’s experts, developer tools and cloud-based quantum systems for organizations to trial. According to the company, 130 billion executions have now been run on the Q Network, and that 100 commercial partners, including Delta Airlines, Goldman Sachs and ExxonMobil, have signed up for membership.
  • Rigetti launched it’s cloud computing service in 2018, along with a $1mm prize for the first conclusive demonstration of quantum advantage on their platform. (Note that this is currently a quantum simulator, rather than access to a full device.)
  • D-Wave has launched cloud services, as well as selling quantum annealing hardware.
  • Amazon (AWS) is offering access to others’ quantum computing services through its quantum solutions platform.

Software development for quantum computers is also a growing area, with well over a dozen startups focusing on providing software services (notably Cambridge Quantum Computing, 1QBit, Strangeworks, and Zapata.)

As an example of how these services are starting to be used: Accenture (which is developing hardware agnostic quantum platforms) is collaborating with quantum software maker 1QBit and Biogen to develop a program to help with drug discovery.

Developing quantum computers involves beautiful math, breakthrough physics and creative engineering design, and enough uncertainty to capture the imagination. The key challenge for early-stage and growth investors alike is that the applications for quantum computers are still largely undefined. Meanwhile powerhouse companies like Microsoft, IBM and Amazon have begun to build out quantum service platforms in the cloud. This puts these megacompanies in a strong position to capture a large part of the value created, regardless of which quantum computing approach turns out to be the most powerful, useful and/or widely applicable.

So far the private sector has put “table stakes” money into quantum computing products and services- spending is expected to reach $830mm in 2024, up from $250mm in 2019. (For context: roughly $1B went into e-Sports in 2019, and $5.3B went into Foodtech investments. Quantum computing is still a nascent space.) There haven’t been major public acquisitions in this space for investors to queue off of (other than Rigetti picking up a smaller quanutm software developer), and it is doubtful that there will be major exits any time soon even in the current SPAC craze.

It also isn’t too late for new startups to make a splash. University spinouts are common in the field quantum computing startups, given that much of the IP generated in this space comes from a relatively small set of physics departments around the world. Big names such as IBM, Google, Alibaba, Hewlett Packard, Tencent, Baidu and Huawei are all also doing their own research on quantum computing. (Nature) The US government and other national programs around the world are pouring significant funding into a variety of quantum computing R&D initiatives as well.

With these considerations, and given that most experts in the industry predict that useful quantum computers are still 15+ years away, this is an especially risky space for traditional venture capital investment.

As a final caveat: it’s hard to predict the rate of future progress in areas like quantum computing, where a significant breakthrough could quickly accelerate progress. This is an area that we are continuing to track with interest.

“Prediction is very difficult, especially about the future.” –Niels Bohr, Nobel laureate and quantum physicist

Want to learn more about quantum computing? Stay tuned for information about our upcoming Quantum Technologies webinar on 3/17/2021 at 12pm PT/3pm ET!

“Circuit” by Yu. Samoilov under CC BY 2.0
  1. To demonstrate quantum supremacy, Google’s Sycamore quantum computer solved a random number generator problem very quickly. It showed that the random numbers generated by a quantum computer are actually random. (As mentioned in the previous post on quantum computing, random numbers are used in computing A LOT, so this experiment wasn’t as “random” as it may seem.) Still, they came up with this problem specifically to show that their quantum computer could do something faster… the problem they solved isn’t one that would come up in real life. IBM, whose supercomputer they used as a benchmark, argued that they didn’t use their supercomputer correctly. This kicked off a lively debate on quantum computing and its merits (or lack of) from many corners of the internet. Quantum supremacy, but not yet quantum usefulness. (Forbes)
    The hardware and work to set up the experiment itself is still impressive — read more about it in the Nature paper published by the research team.
  2. Pioneers of quantum theory include the late Richard Feynmann of Cal Tech; Paul Benioff of Argonne National Laboratory in Illinois; David Deutsch of Oxford University, and Charles Bennett of IBM’s Watson Research Center, N.Y.) (Source)
  3. For the chemists: the researchers who first demonstrated Shor’s algorithm at IBM’s Almaden lab in 2001 used a billion billion (10¹⁸) custom-designed molecules in a test tube as a 7-qubit quantum computer. How? These new, custom molecules had seven nuclear spins — the nuclei of five fluorine and two carbon atoms — which can interact with each other as qubits. To program the molecules and read the outputs, the researchers used radio frequency pulses like those used in nuclear magnetic resonance (NMR) experiments and for medical MRIs. (Science Daily)
  4. Underscoring these difficulties: The National Academy of Sciences recently published a report that measures progress in quantum computing by two key metrics. The first metric is the number of physical qubits, without error correction. The second metric is progress in creating quantum computers where error correction is effective.
  5. “[A topological quantum computer] employs two-dimensional quasiparticles called anyons, whose world lines pass around one another to form braids in a three-dimensional spacetime (i.e., one temporal plus two spatial dimensions). These braids form the logic gates that make up the computer.” (Wikipedia: anyons) I consider any definition that includes “quasiparticles”, “world-lines” and “three-dimensional spacetime” mind-blowing!

Source link

%d bloggers like this: