Cloning Qubits: A New Way to Converse in Quantum Computing
Quantum computing represents an infinite realm of possibilities when it comes to sifting and sorting through incredible amounts of data. No small challenge when you consider that a 2013 report found that 90 percent of the world’s data was generated in just two years.
But one of the most complex issues in quantum computing isn’t the sheer volume of data –it’s a matter of translation.
“Programming for traditional computers is like a foreign language to quantum computers,” explained Steve Adachi, a senior researcher in Lockheed Martin’s Palo Alto, California, offices who has been exploring this field for four years. “Quantum computers require an entirely different approach to programming.”
Lockheed Martin is currently collaborating with D-Wave Systems and academia to advance our understanding of quantum computing.
“Traditional computers have bits that are limited to ‘thinking’ serially in zeroes and ones, while quantum computers have quantum bits or qubits that can ‘think’ in both zeroes and ones, and all of the points in between, all at once,” Adachi continued. “Combine qubits with the fact that quantum computing taps into the physics of entanglement and tunneling, and we’re talking about an entirely new computing language. Problems must be written in a completely different way, making it a true problem of problems.”
In the world of quantum computing, these problems are known as quadratic unconstrained binary optimization (QUBO) problems.
Solving the “Problem of Quantum Problems”
In 2014, Adachi and the Lockheed Martin team tapped Harvey Mudd College’s (HMC) clinic program for ideas on how to best approach the challenge of QUBO problems. The HMC clinic program, which is in its 52nd year, brings together junior and senior-level undergraduate students with industry to collaborate on science and engineering research.
“As the team took on the task of optimizing the programming of the D-Wave quantum computer for qubit communication, we witnessed spontaneous cross-disciplinary learning with our students,” said Richard C. Haskell, Professor of Physics and Director of the Physics Clinic, HMC. “This cross-disciplinary learning was a key factor of the success the students’ ability to help solve the issue of converting problems into the right format.”
Their solution: Cloning qubits.
Under the guidance of Adachi and their academic advisors, four HMC students created a software solution that translates problems into a form that the D-Wave computer can understand, by “cloning” qubits. The students’ areas of study include physics and computer science, which contributed to the diverse thinking needed to address this challenge.
“By cloning these qubits, you can realize all of the connections that are needed to solve real-world problems,” said Adachi. “Without this embedding, there would be no way to solve many QUBO problems on the D-Wave, which really limits what you can do.”
The Value of Industry–Academia Partnerships
This isn’t the first time Lockheed Martin partnered with HMC to research science and engineering challenges. The college recently presented Lockheed Martin with a Milestone Award recognizing the Corporation’s sponsorship of 35 clinic projects in computer sciences, engineering, mathematics and physics since 1965.
"Clinics are one of the most important educational programs at the College, and they wouldn't be possible without dedicated sponsors like Lockheed Martin,” said Maria Klawe, president, Harvey Mudd College. “Our students are afforded a unique opportunity to take responsibility for a challenging problem and to produce professional results.”
Adachi, who earned his undergraduate degree in Mathematics from HMC, couldn’t agree more.
“I still remember fondly my own Clinic experience over 30 years ago, where we worked on oil reservoir simulations using a CRAY-1 supercomputer, which was a state-of-the-art machine at the time,” said Adachi. “Now I’ve come full circle, helping to introduce the next generation of students to what could be the computing technology of the future.”
|Cloning Qubits: How Quantum Conversations Work|
With a quantum computer like the D-Wave machine, problems must first be translated into the QUBO format. This format uses an equation containing a specific number of binary variables, where each variable must be associated with a qubit on the D-Wave so they can “talk” to each other. Not all of the qubits on the D-Wave are actually connected, so it’s hard to have a full conversation.
The logical graph represents a problem we might want to solve, that has five variables with connections between every pair of variables. The physical graph represents D-Wave’s current capability which does not have some of these connections. But by cloning x3, x4 and x5, variables can now be chained together so that all of the required connections in the logical graph are present.
For certain types of graphs, this new technique offers a true advantage in real-world problems that Lockheed Martin is currently exploring using quantum computing.