Breaking Through The 49-Qubit Simulation Barrier: 56 Qubits!

Published by whurley (10/24/2017)

Mathematical Advances Postpone Quantum Supremacy. Wait . . . What?!?

Earlier this month IBM announced they had created a method for calculating quantum amplitudes, also called probability amplitudes, using far less memory than current methods. A probability amplitude is a numerical factor in a math equation. Specifically, the one used to calculate the probability that a quantum system in superposition will collapse to a particular state when measured. That’s a mouthful. At this point some of you are wondering why I’m excited about it. The answer: because it allowed IBM to break the 49 qubit barrier. They now have a classical computer simulating a 56-qbit quantum computer.

Edwin Pednault (Distinguished Research Staff Member, IBM Research) helps us put this feat into perspective. “At roughly 50 qubits, existing methods for calculating quantum amplitudes require either too much computation to be practical, or more memory than is available on any existing supercomputer, or both.” In other words, 49 qubits was the limit of what classical computers could simulate. Now, IBM has bested the 45-qubit simulation achieved by the Institute of Technology in Zurich in July, and done so using far less memory. Institute of Technology in Zurich’s 45-qubit simulation took 500 terabytes of memory, IBM’s 56-qubit simulation only 4.5 terabytes. Previous methods would have required 1 exabyte (1 billion gigabytes) of memory to simulate 56 qubits.

Moving the Goalposts

Itay Hen (Computer Scientist, University of Southern California) has identified an unintended consequence of this fantastic work: “IBM pushed the envelope. It’s going to be much harder for quantum-device people to exhibit [quantum] supremacy.” Quantum supremacy is commonly used to mean the point at which a quantum computer solves a problem that a classical computer can’t practically solve. In the race to achieve quantum supremacy, Google had plans to demonstrate a 49-qubit chip by the end of 2017. Theoretically, 49 qubits would have been enough, until IBM pushed the target back.

Still, kudos to IBM. These advances in our simulation capability will help drive the development of quantum computers and software. Now that we’ve got 56, let’s shoot for 100. 100 qubits could represent over one quadrillion squared values. Think about what all the talented scientists out there could do with that! I know I am. Let’s go get it.