Ostrava, 8 September 2025 – Researchers from IT4Innovations National Supercomputing Center at VSB – Technical University of Ostrava have launched a three-year research project to develop a new adaptive quantum error correction system for solving optimisation problems. The project leverages reinforcement learning, one of the most advanced methods in artificial intelligence, enabling quantum computers to independently design optimal quantum correction codes tailored to specific tasks and current noise conditions. The research, funded by the Moravian-Silesian Region's Vouchers for Universities programme, received a grant of nearly CZK 4.7 million.
“Optimisation problems such as travelling salesman, protein folding, and financial portfolio management are extremely demanding in terms of computing power. Quantum computers offer the potential to solve these problems more efficiently. However, this requires effective error and noise suppression, with noise being currently an obstacle to their practical deployment,” explains project guarantor Marek Lampart, Head of the Quantum Computing Lab at IT4Innovations. Project investigator Ryszard Kukulski adds: “We will use reinforcement learning, which will function as an intelligent agent—based on feedback, it will gradually improve the correction codes specifically for the given type of problem and quantum computer architecture. This will maximise the accuracy of calculations and reduce the need to repeat experiments.”
The research focuses on solving optimisation problems using the Ising model, which forms the mathematical basis for many problems in industry and the natural sciences. These problems can be solved using modern quantum algorithms, primarily the Quantum Approximate Optimization Algorithm (QAOA). However, the results are often subject to errors due to noise in quantum systems.
The solution is reinforcement learning, a type of artificial intelligence that learns through trial and error. The reinforcement learning agent will generate topological quantum codes that minimise the impact of noise. The system will adapt to the specific architecture of the quantum computer—for example, VLQ, the superconducting quantum computer installed at IT4Innovations. It will take into account the type of error and the connectivity of qubits and select the optimal correction code. The solution will also include probabilistic error correction and the use of proven error mitigation methods. These use clever mathematical techniques to refine the results even when the quantum computer is still making errors – without the need for complex corrections for each one.
The project “Quantum error correction codes enhanced by reinforcement learning dedicated for the Ising model-based optimization” will be implemented between 2025 and 2028. It will significantly advance not only research into quantum algorithms but also bring quantum computing closer to practical application. As a result, quantum computing will become more accurate and reliable without the need for frequent repetition of demanding experiments.
“Vouchers for universities in the Moravian-Silesian Region are intended to support cutting-edge research and science. They serve to attract talented postdoctoral researchers from abroad and bolster research teams at local universities. In this way, we create jobs for graduates of doctoral programmes from abroad or for experts with PhD degrees and extensive experience worldwide. The Moravian-Silesian Region has approximately CZK 93 million for these scientists in the Just Transition Operational Programme, and all universities in our region can apply for funding until the beginning of November next year,” said Josef Bělica, Governor of the Moravian-Silesian Region.
Key Facts and Figures
Commencement: May 2025
Completion: May 2028
Funding: CZK 4,680,244
Funding programme: Vouchers for Universities in the Moravian-Silesian Region – 1st Call
Grant provider: Moravian-Silesian Region through the Just Transition Operational Programme
Principal investigator: IT4Innovations, VSB – Technical University of Ostrava