I am trying to improve on our current approaches to quantum circuit simplification. The circuit representation we use both for circuit design and optimisation poses indeed a few challenges. It can be hard – or unnatural – to squeeze a quantum primitive into a circuit: think of how a controlled unitary is decomposed into something unrecognisable in a typical universal gate set. On the user-facing design side, this might be patched by introducing new ad-hoc primitives as they are needed. On the optimisation side, however, this causes problems: it is very hard to resynthesise higher-level primitives from simpler ones, or to make use of ad-hoc primitives in a general way.
This means that we face difficulties in devising and using optimisation strategies that could leverage this higher-level structure. Beyond circuit optimisation performance, this will increasingly become an important issue if we wish to compile to architectures that might support more exotic gate sets than are currently typically available, such as multi-qubit operations on ion traps devices.
I am currently formulating a new presentation of quantum computation that aims to address some of these issues. Paper to come out on this very soon…
- Corinzia, L., Penna, P., Mondada, L., Buhmann, J. M. (2019). “Exact Recovery for a Family of Community- Detection Generative Models.” IEEE International Symposium on Information Theory, ISIT 2019, 415 – 419. [arxiv:1901.06799]
- Iten, R., Reardon-Smith, O., Mondada, L., Redmond, E., Kohli, R. S., & Colbeck, R. (2019). “Introduction to UniversalQCompiler.” preprint [arXiv:1904.01072]
- Mondada, L., Karim, M.E., and Mondada, F. “Electroencephalography as implicit communication channel for proximal interaction between humans and robot swarms.” Swarm Intelligence 10.4 (2016): 247-265 [pdf].