UK Quantum Hackathon entries, 2024 and 2025
2024 Quantum copula
​Summary
We participated, for the first time, in the UK Quantum Hackathon organised by the National Quantum Computing Centre (NQCC) in 2024, a national competition alongside many UK entities (BT, NHS, UK Atomic Energy Authority, ...) to expand our quantum research activities and, at the same time, strengthen our profile and build new research collaborations. Our team of university researchers made a significant contribution, helping us to win the competition. As a follow-up to the competition, we extended the research conducted during the competition to secure a publication.
Problem
A central challenge in actuarial science is characterising the distribution of losses generated by an insurance portfolio. Copulas provide a natural framework for modelling joint distributions, but flexible high-dimensional copula modelling and sampling remain challenging in practice.
The project extends the work done in Nature's publication Copula-based risk aggregation with trapped ion quantum computers, which presents a method for evaluating risk aggregation for portfolios that include insurance losses. This estimate is critical for insurers to validate their cash availability against regulations, ensuring they can cope with large and simultaneous losses which may also trigger significant movements in financial markets (like in the case of a natural disaster). At the same time, these models are computationally expensive.
Solution
The key contributions and novel aspects of this work are:​
-
Insurance loss distributions are heavy-tailed and have a distinct behaviour with respect to stock market distributions, making it necessary to craft a specific embedding method to load the information on quantum hardware.
-
The data size is limited by its nature (1 data point per day, limiting its size to a maximum of the order of 10^5), generating issues in the training procedure for the computation of the KL-divergence. A new method is introduced that can be applied in the wider QML landscape.
-
Training models on noisy quantum hardware can be unreliable; to stabilise the procedure a hierarchical learning approach has been used: starting the training on a smaller ansatz on simulated hardware and then using the learned weights on a larger circuit to warm-start the learning process on real quantum hardware.
​Status
As of July 2026, the team is working to finalise the preprint review, with the aim of a publication in a quantum-sector journal. The algorithm has been trained and executed on real quantum hardware provided by Quantinuum, a partner on this project.
2025 Road maintenance
​Summary
We participated, after the success of the 2024 edition, in the UK Quantum Hackathon organised by the NQCC in 2025, with the aim of continuing to expand our quantum research area and, at the same time, strengthen our profile and build new research collaborations. Our team of university researchers made a significant contribution, helping us win the competition for the second time in a row. The work was the extension of a research project we ran in the previous months, co-funded by the NQCC.
Problem
Road networks consist of several asset classes, each with different maintenance and condition assessment requirements. As this critical national infrastructure ages, the monitoring and maintenance required to keep it functional goes up, while budgets for this activity remain static. This drives the need for these activities to become more efficient over time. One way to achieve this is by optimising the scheduling of maintenance that requires road closures, which is the most expensive kind.
The aim is to minimise disruption caused by maintenance activity in transport networks by optimising the scheduling of maintenance actions to reduce the overall number of closures and maximise the activity that takes place during those closures.
Solution
The team developed an alternative solution to the one we provided during the SparQ programme, co-funded by the NQCC, leveraging a QUBO formulation compatible with D-Wave's hardware, performing hyperparameter optimisation, and enabling testing on larger scenarios. An additional contribution consisted of reducing some simplifications of the problem formulation, making the algorithm more general and closer to the real framework. On top of that, the team implemented an additional classical solver, based on the Gurobi Optimizer, and benchmarked the performance of all quantum, hybrid and classical solvers.
​Status
The work of the research team was praised by the jury of the NQCC's Quantum Hackathon by awarding it the first place, and the jury particularly appreciated the awareness demonstrated regarding the ethical and societal impact of the proposed solution.