kentoh - Fotolia
What's the best itinerary for a Mars rover to follow tomorrow? How will caffeine interact with other drugs in your system? Until recently, answers to questions like these were difficult, if not impossible, for computers to answer. They simply contain too many variables.
But, now, the advent of quantum computers is bringing answers to those questions closer. And recent advances and investments in quantum technology are making some think that the use of quantum machines by mainstream customers in research and, yes, business is right around the corner.
"There is a lot of excitement in the field right now," said Seth Lloyd, director of the Center for Extreme Quantum Information Theory at MIT.
Jerry Chow, manager of experimental quantum computing for IBM Research, said, "It's the first technology to really change the game, to tackle a class of problems that have been neglected because they are too complex for any high-performance computing system."
Billions of investment dollars are adding to the excitement around quantum machines. Venture capitalists have backed quantum startups with hundreds of millions of dollars, while corporations have invested billions more in R&D, according to analyst Peter Rutten, research manager of server solutions at IDC.
The U.S. government is funding quantum information science to the tune of $200 million per year; the European Commission has announced a $1.18 billion initiative; while the U.K., Australia and Canada all are making significant investments, Rutten pointed out. The Chinese Academy of Sciences, meanwhile, is hard at work on what it said will be the world's first quantum computer.
If you are tempted to ask, "Are we there yet?" the answer is no, but it might be time to get your list of problems in order and assemble a team to explore how to put the power of quantum machines to work solving them. If quantum computing is getting real at long last, the solutions to many problems that are beyond the capabilities of classical computers -- and the answers to many more questions that people are just beginning to imagine -- might be just over the horizon.
"We're kind of where Intel was in 1964. We're at the dawn of the quantum computing age," said Vern Brownell, CEO of D-Wave Systems Inc., the maker of a quantum annealing machine that is being used by research institutions.
"It's like a grad student's dream is now becoming a reality," said David Moehring, CEO of IonQ Inc., a quantum computing startup.
Not bits -- qubits
If you're new to quantum computing, it won't be long before you feel like you're in a technological hall of mirrors. Nothing is quite like the world you're used to. Briefly: The building blocks of quantum computing are qubits, which are based on the quantum states of atomic particles. While the bits of classical computers are on-off switches greatly miniaturized, qubits can be in many states at once, not just on or off -- a property known as superposition. A condition called Entanglement occurs when qubits become intrinsically linked together. Together, superposition and entanglement create enormously enhanced computing power for solving certain kinds of problems.
For a general-purpose quantum computer, 49 qubits is the level at which most agree it would surpass the capabilities a classical computer for optimization problems. "It may not sound like very much, but a 49-qubit device crosses a threshold where you can't simulate and understand what's going on inside [a quantum computer] with a classical computer," MIT's Lloyd said.
However, getting a qubit ready to do work is no easy task. It may require exotic materials or supercooling to near absolute zero. In addition, a quantum machine must overcome some nettlesome problems, including decoherence and noise, both of which threaten the stability of qubits and their ability to process information.
There are several different quantum computing schemes being pursued by several major players as well as smaller companies, each intended to tackle these challenges in different ways, and it's not clear yet which will win out in the long run. [See sidebar.]
As to the problems that quantum technology can solve, the so-called traveling salesman problem is typical. A traveling salesman must visit the largest city in every state following the best route -- the shortest route that meets with each customer at the right point in each customer's buying cycle. Although seemingly straightforward, the problem contains too many variables for a classical computer to calculate efficiently.
And how does a relatively simple molecule like caffeine interact with the new drug your pharmaceutical company would like to introduce? It turns out the relatively simple caffeine molecule has far more permutations when it comes to interacting with other molecules than any classical computer can simulate. But a sufficiently powerful quantum computer could tackle such a problem.
The ability to solve multivariable problems lends itself to machine learning and the related field of deep learning in which artificial neural networks lend a hand, according to Lloyd. "With a small-scale quantum computer, you can do some pretty serious damage to machine learning problems," he said.
As to industries, the most fertile ground for quantum machines are expected to include drug discovery, materials discovery, financial services, artificial intelligence (AI) and security. The impact could be significant. "A big enough quantum computer, with several thousands of qubits, could decrypt the encryption we have today," said Brian Hopkins, vice president and principal analyst for enterprise architecture at Forrester Research.
Who's having all the fun?
Cloud-based services offer enthusiasts an avenue for testing quantum computing applications.
James Wootton, a research fellow at the University of Basel, Switzerland, was busy at work thinking up theoretical problems to be solved by theoretical quantum computers when he received a pleasant surprise: He realized that he could actually carry out one of these experiments. "I could do it myself that very day. I would not have to find someone to do it for me," he said.
Wootton and his students have been using IBM's cloud-based Quantum Experience service regularly, and Wootton has created a game that runs on Quantum Experience, called Quantum Battleships, which enables participants to experience the unusual behavior of qubits in the context of a game.
Quantum Experience is within the technical reach of many, according to Wootton. Using its Python-based API "is something that people with a knowledge of programming should be able to do," he said. If not, there is a drag-and-drop GUI version, he added.
"Having devices that actually exist has altered my research quite a lot. … I encourage people to experiment with it," Wootton said. So far, 52,000 users have done just that, running over a million executions on Quantum Experience and generating 25 papers, according to Chow of IBM Research.
Although some researchers at NASA Ames Research Center have tried using IBM Quantum Experience, Rupak Biswas, director of exploration technology at NASA Ames, said he prefers working with an actual quantum computer to better understand how it works, rather than with a cloud-based system.
Biswas and NASA Ames researchers have been experimenting with a D-Wave machine. Although the D-Wave machine that NASA is using has some 2,000 qubits, because the system is a quantum annealer, rather than a general-purpose computer, it is limited to certain kinds of optimization problems.
But some of these optimization problems are right up the alley of an organization concerned with space exploration. For example, according to Biswas, planning an itinerary for a rover on Mars is a variable-rich problem for which a quantum device like the D-Wave machine could be helpful. The rover must stay away from steep slopes and must orient its solar panels to receive sunlight. "It's a complex optimization problem," he said.
While a classical computer can provide answers to such problems, experts like Biswas wonder whether classical-computing-derived solutions are, in fact, optimal. Input from a quantum machine could shed new light on that. So, part of NASA Ames' task is to compare quantum results with classical results to see if a quantum machine is capable of coming up with a better answer. However, Biswas was quick to point out that the D-Wave system has significant limitations of its own. For instance, the D-Wave system that NASA Ames is using does not have error-correcting capability, as do classical microprocessors. "If a qubit is misbehaving, you would never know."
Also using a D-Wave machine is Daniel Lidar, professor of electrical engineering at the Viterbi School of Engineering at the University of Southern California. The school acquired a 128-qubit D-Wave machine in 2011 and has since upgraded it to 1,152 qubits. So far, Lidar and colleagues have been focusing on basic issues like making sure the machine is actually using quantum effects, benchmarking the quantum annealer against classical alternatives and implementing quantum error correction on the system.
Although excited by the possibilities of quantum work, Lidar has no illusions as to the state of current progress.
"There simply is nothing you can run on a quantum computer that you couldn't already do much better on classical computers. Quantum computing, at this point, is very much a research domain. That is a fact and it's going to take several years before that situation changes," Lidar said.
One thing that no one is predicting is the wholesale replacement of classical computers with quantum machines. The most optimistic quantum proponents say quantum machines will be a cloud-based problem-solving adjunct to classical machines. "Ultimately, it will be ubiquitous, but not a replacement for classical computing. We believe applications will be hybrid, with some quantum and some classical computers," D-Wave's Brownell said.
Classical computers, meanwhile, are hardly standing still. "Today's compute platforms are highly advanced and will continue to evolve for a long time. Quantum computing will not replace current compute technologies, but it will, at some point in the future, augment them to solve today's unsolvable problems," IDC's Rutten said.
Rutten's estimate of the total worldwide server market is $53 billion annually. "Quantum computing will not make a dent in the market in the next five years. It could perhaps make up a fraction of 1% of the total market in five years," he said.
Even so, for some businesses, it might make sense to assign a manager to become expert on the technology, to track advances and to give Quantum Experience a try. "You could get ahead and figure out how to get data out of it. It's something that takes time. Even for a programmer who knows about quantum computing, it takes a little bit of practice," Wootton said. That level of diligence could pay off if the excitement felt by quantum proponents proves well founded.
Hopkins of Forrester asserted that progress in quantum computing will be made exponentially, not linearly. "If it does move exponentially, this thing could bury you in just a few years," he said. That kind of rapid advancement could be game-changing and calls for due diligence on the part of those in industries on which multivariable problem-solving might have the most impact.
Biswas said, "In general, it's a very exciting place to be."
A quantum computing inflection point?
Although quantum computers have not yet surpassed their classical counterparts, there are many signs they could be nearing that threshold. Here is a rundown of companies (in alphabetical order) working on quantum computing.
D-Wave is the only company to be shipping a working quantum machine, but D-Wave systems are actually not computers, but quantum annealers. Some believe that because D-Wave systems are not general-purpose computers, they have unacceptable limitations, despite reaching the seemingly high level of 2,000 qubits. The machines have attracted the attention of NASA Ames and other research centers. "We have built our fourth generation processor and will come out with a new one every several years," said D-Wave's Brownell. The company doesn't quote prices, but its systems are thought to cost between $10 million and $15 million each.
Google has been working with D-Wave machines at NASA Ames and has created a 20-qubit microprocessor of its own, with the aim of creating a 49-qubit chip in 2017.
IBM, through its IBM Q initiative, has made a 5-qubit general-purpose system available via its Quantum Experience cloud service to all comers for free, enabling interested parties to try out basic problems and get the feel of interacting with a quantum computer. The Python-based software-development kit is available on GitHub under an Apache 2 license. Later in 2017, IBM said it will offer a 16-qubit version to the public and a 17-qubit version targeted to business users. IBM is pursuing gate model technology, also known as circuit model technology.
Intel is aiming to adapt silicon transistors to do the work of superconducting qubits.
IonQ is developing a general-purpose quantum computer using lasers to cool and isolate, or "trap" Ytterbium ions to create quantum states. This approach has the advantage of not requiring supercooling. IonQ is working on a 30-to-40-qubit ion trap computer. "The techniques are understood and the way to scale them is understood," said IonQ's Moehring.
Microsoft is working in the field of topological qubits, which may be able to hold its quantum state longer than is possible through other approaches.
Rigetti and Co. Inc. released for public beta in June 2017 Forest 1.0, a programming and execution environment for developing algorithms for quantum-classical hybrid computing. Users can simulate the algorithms they create using Rigetti's Quantum Virtual Machine in the cloud. Meanwhile, the company is developing a quantum-integrated circuit architecture.
Quantum computing could apply to AI and medicine
IBM working on building quantum machines
Quantum machines a threat to cryptosystems