IEEE QCE impression

3 minute read


Some thoughts I was left with after this years IEEE quantum week, with extra focus on the numerical simulation and software architecture workshops.

Conferences like these are truly important to the community. What I particularly liked about the IEEE Quantum Week talks where how they for the most part were truly responsible and providing a nuanced view without the overhyping that one often sees for example at the more business oriented conferences. The program catered to many different interests and this resulted in some very interesting lunch-time conversations. Among attendees there were those who work in quantum software and hardware, but also the rest of the software stack, algorithms, applications and so on. There were plenty of attendees both from industry and academia and it’s always interesting to discuss and see how views differ.

I came to the conference largely with the intention try to probe the community on our current directions at Pasqal and use this to make some upcoming recommendations and decisions. It was interesting to see that much of what we do at Pasqal is indeed very different to the rest of the industry. Although what we consider best practices certainly aligns.

Advanced simulation for quantum computations workshop

Two day workshop dedicated to numerical simulations of quantum computing. I was presenting a talk and sat in a panel for this so I ended up spending virtually all my time the first two days with this. It was great to see what everyone was up to and to discuss where we should be heading.

Two things that I was already suspecting became very clear to me.

  1. We should leave performance tuning to the experts
    1. The state of the art libraries that are written with no other aim than to run ever bigger programs have made trade-offs that I simply couldn’t justify making both from a maintainability perspective but also from a user friendliness perspective based on how I see these codes used at Pasqal.
  2. When dealing with tensor network methods it pays off to be clever when setting up your simulation more so than spending time tuning your numerical code for general cases.
  3. See for example all re rebuttals of quantum advantage experiments.

Quantum software architecture

This was probably the highlight of the conference for me. We spent the the whole day together talking about the challenges of software architecture in the quantum computing area. The workshop was organised by members of the software engineering institute at CMU and as part of a ongoing project of theirs that had already received some thought.

The organisers included Rick Kazman obviously with such a guru in the workshop much of how we approached things were based on his previous work. After two quick presentations we started by analysing problems according to his principles of architectural quality attributes and after some discussion ended up forming three groups that would tackle specific problems. I focused on the “usability” track, where we had several interesting discussions not only on usability directly but how usability must always be a tradeoff with e.g. testing, security, or robustness of program. For example, we’ve seen at Pasqal, that sometimes you want to give the users what they want in terms of flexibility but you end up making it very hard for novices producing correct input and in some cases invite them into ways of thinking that we should discourage.