We recently hosted our first quantum computing hackathon here in Cambridge, UK, where over 20 keen quantum programmers were given exclusive cloud access to Rigetti’s 8-qubit Agave quantum computer. We had participants with various backgrounds, ranging from secondary school students to PhD students and experienced professionals, most of whom had no prior experience programming quantum computers.
Participants were given the challenge of finding the best solution to a quantum chemistry problem, by writing their own optimiser for the Variational Quantum Eigensolver algorithm (VQE). This involved running quantum circuits on Rigetti’s 8Q Agave QPU via the cloud to compute the energy of a small molecule, and then using their own classical optimisation strategy to find the quantum circuits that compute the ground state energy. It is expected that VQE can offer an exponential speedup over classical methods for these electronic structure calculations and, as hardware scales up, VQE should be capable of solving high-value problems in materials science and drug discovery. A wide range of creative ideas were submitted by participants, including methods based on Fourier analysis, genetic algorithms and simulated annealing.
Thomas Parks, a computer science graduate student at the University of Cambridge, explained his approach: “ I used a Bayesian optimization (BO) method to find the energy minima. BO methods aim to find the minima of black-box continuous functions, by updating an internal belief about the shape of the function. This belief, modelled as a Gaussian Process, is used to sample at function locations that are either uncertain or likely to be near the minima. The Bayesian method appeared especially powerful on the real QC due to its ability to efficiently make use of noisy observations.”
Thomas also added: “The hackathon was a wonderful introduction to the world of Quantum Computing. The advisors from River Lane Research and dividiti really helped to get everyone going rapidly, and access to the Rigetti quantum computer was really special. The atmosphere was very collaborative, with everyone discussing ideas and helping each other out!”
Participants ran their code using Quantum Collective Knowledge (QCK), a framework developed by dividiti and River Lane Research to benchmark existing quantum computing systems, pinpoint the state-of-the-art and forecast future developments. QCK builds upon Collective Knowledge, a universal open source framework for reproducible and collaborative R&D. Using QCK made it straightforward for participants to write their code, benchmark it on real hardware, and submit their results. The quality of submissions could then be compared fairly using a time-to-solution metric.
Anton Lokhmotov, co-founder of dividiti, commented: “We’ve had very positive feedback from attendees, who were surprised by how quickly they could get up and running on a quantum computer. It was encouraging to see initial results coming in under an hour.”
Full hackathon results, as a Jupyter notebook, can be found here.