Last week, Riverlane opened its doors to 11 students at various stages of university education for our second Bertha Swirles Workshop. We had students who were at the start of their academic journey, in the early years of their university education, some who were completing an integrated master's and even a PhD student!
The students came from a mix of subjects, from chemistry to physics and philosophy. I should add that while some of the workshop attendees had some experience with quantum computing (perhaps a course at university or experimentation with Qiskit), the topic of quantum error correction (QEC) was new.
We are aware that women are significantly underrepresented in quantum computing, with only 1 in 10 working in this field. At Riverlane, one of the ways that we are trying to change this is by educating women at this early stage of their careers in quantum error correction, so that they might be inspired to choose this direction for their future careers.
The workshop kicked off with a foundation in quantum computing, including talks from Maria Maragkou, our VP Commercial, about the quantum computing ecosystem, as well as the importance of intellectual property from Olivia Buckingham, our in-house patent attorney. The students were fascinated to learn more about the world of IP and, in particular, Olivia’s role.
After setting the scene, we were ready to delve into the complex world of QEC. We wanted to understand if it would be possible to go from classical error correction all the way to logical operations with the surface code in the space of two hours. It turns out that Riverlane has the QEC expertise and knowledge to be able to deliver on this. The trick seemed to be starting with something classical (bits being encoded redundantly) and then gradually building on how that changes for a quantum system that is subject to decoherence. Once the ingredients of data qubits, auxiliary qubits and stabilisers were understood, a patch of surface code no longer seemed like an abstract mathematical object. In fact, one student commented on how they had seen a picture of the surface code before but never really understood it. Annie Ray brought everything together with a fun cross-stitching analogy to understand how patches of surface code are ‘stitched’ together in a process called lattice surgery.
On Wednesday, we were lucky enough for the students to have a Q&A with Steve Brierley, CEO of Riverlane. One student asked what incentivises Riverlane to have such a welcoming atmosphere and environment, to which Steve replied that he wants people to enjoy their work. In the afternoon, we began to introduce the concept of qLDPC codes through Vanthana Ganeshalingam’s talk about belief propagation (BP) decoders.
We had provided such a solid foundation in QEC that students were able to attend one of Riverlane’s weekly T- and bisqubits technical talks, where we invite external researchers to present cutting-edge research. This particular Thursday, we had Josu Etxezarreta Martinez presenting his work on the BP+OTF[1] algorithm as an almost-linear time decoder for qLDPC codes. The fact that we had taught students enough information about QEC that they were able to attend (and take notes in) this talk was a great achievement.
From a recruitment perspective, we understand the importance of having tangible projects to talk about during the interview process. We wanted to understand how much the students had really understood about decoders for qLDPC codes, so we set the challenge of comparing convergence rates for different BP decoders using Joschka Roffe’s open-source repository[2]. Some students commented on how this was their first experience using an open-source repository. In fact, for some workshop attendees, this was their first time really properly coding at all.
It was an incredibly challenging hackathon to attempt with only three days of QEC training, but every group produced plots and presentations exploring different types of BP decoders that could be applicable to qLDPC codes. One group referred to the underlying structure of the syndrome to explain the convergence rates at different code distances. We were impressed at how the students confidently used QEC terminology throughout the presentations, despite being new to the topic at the beginning of the week.
The week ended with a celebration, during which I was talking to one of the students about a university project they were now considering to combine quantum computing and chemistry. This goes to show the value of educating students about quantum computing at the early stages of higher education because it could inspire them to shift directions towards a career in quantum computing - and hopefully QEC!
References
[1] 
An almost-linear time decoding algorithm for quantum LDPC codes...
[2] 
GitHub - quantumgizmos/ldpc: Software for decoding classical and quantum codes