# High Level Explanation of Measuring the Speed of Adiabatic Quantum Optimization Processors – Some Real World Benchmarking Shows 10,000 times faster

Dr. Suzanne Gildert is the author of the wonderful Physics and Cake blog. She has a high level overview of the Investigating the Performance of an Adiabatic Quantum Optimization Processor paper (10 pages) up at the Dwave blog. How do you measure the speed of an adiabatic quantum computer?

This is a tricky question to answer, but this paper gives it a shot. The paper focuses on a special kind of quantum algorithm, known as Adiabatic Quantum Optimization (AQO), which runs on a microprocessor built by D-Wave, and I’m going to try and give a high level overview of the main messages in this paper. The microprocessor itself has been built and tested using a superconducting electronics technology, but this paper concentrates on SIMULATING what it is expected to do. I’ll explain why, and what the results tell us about the promise of these processors.

Introduction

AQO processors behave more like analog computers than digital ones, so it is particularly tricky to measure their ‘clock speed’ in order to compare it to other processors.

The paper is mainly concerned with something known as the ‘minimum gap’. What is this? Well it is the thing that limits the speed of your computation. It is the ‘bottleneck’ in the computation, and it is what makes these problems very difficult to solve. As we turn the crank on our computation, there are points where the system can become confused and sometimes gets ‘stuck’ in the wrong answer. But quantum systems (like AQO processors) avoid this problem, as they naturally have a ‘gap’ between the right answer and the wrong answer at all times (the lowest energy state and a slightly higher energy state) during the computation. The gap is extremely important – if you built the same system of little spins without quantum mechanics, as you turn the crank the gap goes to zero at some points, and therefore it is highly likely that you’ll end up in the wrong answer. The computation has a chance of ‘following’ the wrong path!

Results of Standard Benchmarking Speed Tests

The AQO approach was compared to similar problem solving techniques already used to solve these types of hard problems in the real world. The AQO approach seems to be approximately 10,000 times faster for these optimization problems than the best approaches currently known.