Quantum Computing: Tackling the Readout Challenge

Novel technologies, novel issues.

As history has shown, disruptive technology breakthroughs are usually accompanied by drastic changes in many aspects.

One example is the invention of the automobile and how the world had to change to adapt to it. We needed to build streets to allow for smooth rides. In order to not interfere with the now accelerated traffic, people had to be moved off the streets, leading to the “invention” of the sidewalk and the playground. And not the last is the need to provide parking spaces en masse, (raise of the multi-storey car park) to avoid people losing too much time to look for a parking space.

A completely new peripheral infrastructure was required to unleash the benefits of a car. Without that it was pretty much useless. If you cannot expect to get a parking lot, you better go by foot or bike!

What does that have to do with Quantum Computing

With Quantum Computing we are (likely!) to see a similar transformation. Product roadmaps of Quantum Hardware providers project an exponential growth of computing power¹. In analogy to the automotive story: we can expect faster and faster cars in the near future. This requires us to have a growing number of roads to make effective use of the cars and that is a growing number of algorithms and concepts to make use of quantum computers. Together with PASQAL and the University of Exeter, we are exploring these. For example, we have developed a novel algorithm for solving partial differential equations that exhibits favorable scaling to industrial problem sizes (https://blogs.sw.siemens.com/art-of-the-possible/exploring-the-quantum-frontiers-of-differential-equations-in-engineering/, https://arxiv.org/abs/2404.08605)

However, there are still many technology challenges which prevent unleashing the full benefit of Quantum Computing. One of those technology hurdles is the fundamental readout problem.

The readout problem

The readout problem refers to the challenge of accurately measuring the state of a quantum system after it has undergone a quantum computation, i.e. to readout the final state. This final state contains all the information of the result of the computation.

The primary difficulties associated with the readout problem stem from the inherent properties of quantum mechanics. Measuring the state of a quantum system can disturb the system itself, causing the state to change in an unpredictable way, due to the uncertainty principle. Think of Schrödinger’s cat (see info box below). Additionally, quantum measurements are inherently probabilistic, leading to measurement errors where the measured state does not accurately reflect the true state of the system. Furthermore, quantum systems are highly fragile and can easily interact with their environment, resulting in a loss of the delicate quantum coherence that is essential for quantum computation

As the number of qubits, the fundamental units of quantum information, in a quantum computer increases, the readout problem becomes increasingly challenging. Reading the full classical solution from these quantum states with tomography techniques scales exponentially with system size.

Essentially that implies that all the potentially exponential speed of the quantum computation is actually wasted on reading the results.

It is like getting into your car for a 2 min ride but you have to circle the block for 15 min to find a parking space.

Distilling the relevant features

Together with our partners from PASQAL and the University of Exeter, we have shown that the readout problem can be addressed with quantum learning tools². Treating outputs of quantum differential equation solvers as quantum data, we demonstrate that low-dimensional output can be extracted using specific measurement operators. We demonstrated this quantum scientific machine learning approach to classify solutions for shock wave detection as well as turbulence modeling base on data samples coming directly from quantum differential equation solvers.

We could show that this novel approach significantly boosts classification accuracy. This brings us one step closer to an adoption of quantum computing.

However, there is still a long way to go to for quantum computers to be leveraged on a large scale. Although quantum hardware is advancing significantly, we have to take baby steps to build all the necessary peripheral infrastructure to really unleash its full power. Coming back to our initial analogy. We have cars, we have (some) roads, and the first parking garages are popping up. Since the timeline of technology evolution is hard to predict, our research efforts ensure that you will be able to benefit from quantum computing using our tools once it is ready for prime time.

1. In terms of number of qubits, number of logical qubits, and qubit volume. ↩︎
2. Williams, Chelsea A., Stefano Scali, Antonio A. Gentile, Daniel Berger, and Oleksandr Kyriienko. “Addressing the Readout Problem in Quantum Differential Equation Algorithms with Quantum Scientific Machine Learning.” arXiv preprint arXiv:2411.14259 (2024). ↩︎