To develop algorithms for real-time scientific simulation and data mining. The seamless integration of scientific computing in complex engineering processes can yield predictive and diagnostic information to optimise outcomes. Computing with high-dimensional models and datasets in real-time poses tracktability challenges for conventional linear algebra and optimisation algorithms, and so we look for efficient probabilistic approaches that approximate these computations at a fraction of the time with moderate computational resources. Applications of interest arise in multiple engineeting applications including computational imaging, diagnostics for digital manufacturing and simulation for digital twins.
Our research interests are at the interface of computer science, engineering and applied mathematics. Currently we are working on:
Inverse problems: Model-based and data-driven image reconstruction and uncertainty quantification for large-scale problems in time-critical applications.
Data sketching: Model order reduction and data compression for mathematical modelling and data mining.
We are currently pursuing research on the following projects:
Optical, real-time imaging of gas dispersion: Funded by the
James Clerk Maxwell Foundation we research into efficient and robust algorithms for gas plume imaging from low-count backscattered lidar data.
Spectral computed tomography: We are exploring deep learning algorithms for high-resolution tomographic imaging and material characterisation. Partly funded by an Institutiuonal Strategic Support Fund from Welcome Trust in collaboration with colleagues at the medical school and MARS Bioimaging.
Sketched numerical simulation and imaging: We explore algorihms based on randomised numerical linear algebra that are suited to fast approximation of high-dimensional forward and inverse problems govened by elliptic and parabolic models.
In-process diagnostics for selective laser sintering: purpose and collabs, funding some more text here.
2.10 Alexander Graham Bell Building | School of Engineering | Kings Buildings Capmpus | EH9 3JL, Edinburgh, UK | Tel: +44(0)131-650-2769 | Email: n.polydorides_at_ed.ac.uk