by Christoph Sünderhauf and Oliver O’Brien
Every year, Riverlane opens applications for its Master’s and PhD internship schemes, giving students the opportunity to explore a career in quantum computing and helping them understand how their skills and expertise could be used in a commercial setting.
We’re constantly impressed by the quality of our interns and, in this post, we wanted to congratulate and highlight work from Oliver O’Brien, who’s now back at Cambridge University, for this work on a problem ubiquitous in quantum algorithms.
“I came into the internship hoping to learn more about early fault-tolerant quantum algorithms, and I left with a whole new perspective on algorithm design and resource estimation. Christoph was an incredible mentor—always open to discussion and generous with his time. I’m really proud of what we accomplished together in just four months, and to have the work published in Quantum is a fantastic outcome,” Oliver explained.
The paper “Quantum state preparation via piecewise QSVT” has just been published in Quantum Journal. It tackles the issue of loading classical data into a quantum computer, which is a problem ubiquitous in quantum algorithms that can easily cause loss of quantum advantage.
Efficient quantum algorithms are essential for unlocking the full potential of quantum computing and must be developed in tandem with quantum error correction (QEC) technologies. While QEC is Riverlane’s main focus, algorithms have a significant role in helping quantum computers reach computational advantage.
Sometimes, running a quantum algorithm on the data takes significantly less time than preparing the data in the first place. So, we need data loading to be as efficient as possible.
Christoph and Oliver developed a technique that could require 50 times fewer computational resources than the best previous data loading methods. “It was exciting to see my ideas come alive and be developed further in course of the internship,” Christoph said.
Now, let’s hand over to Christoph and Oliver to tell us more about their paper.
Taking a piecewise approach
Loading classical data into a quantum computer is a major bottleneck that limits the practical usefulness of many quantum algorithms.
While some classes of data can be compressed and loaded efficiently, many real-world datasets contain sharp discontinuities or singularities that make this much more difficult. In this work, we introduce a new method for efficiently loading such challenging data into quantum systems, enabling faster access to a broader range of inputs.
We achieve this by extending a powerful framework known as Quantum Singular Value Transformation (QSVT). QSVT allows polynomial transformations of matrices that are encoded into quantum circuits—a process known as block encoding. We develop a piecewise variant of QSVT that applies different polynomial transformations to different regions of the input, allowing us to better match the structure of complex data. To support this, we introduce a new block encoding that enables efficient sampling from piecewise polynomial transformations.
To demonstrate the impact of our method, we show how it can efficiently prepare a classical B-spline window function—valued for its excellent spectral properties. This function enhances the performance of Quantum Phase Estimation (QPE), a key subroutine in many quantum algorithms. Using our technique, this enhancement can be achieved with 50 times fewer resources than the best previous methods.
Our work provides a general framework for preparing a wider range of quantum states based on structured classical data, especially those with localized features or sharp transitions. We expect it to be broadly useful in early fault-tolerant quantum computing, where minimizing circuit costs is essential.
You can read the full paper here.