Within the National Centre for HPC, Big Data and Quantum Computing (HPC) a research fellowship is funded.
In recent years, machine learning (ML) tools have been applied to a variety of scientific fields, including computational fluid and solid mechanics. Referring to microfluidics, we focus on a morphing environment induced by the actuation of compliant channels, by reactive phenomena at the fluid/solid interface (e.g., precipitation/dissolution processes), or by the flow of microparticles and/or in presence of multiple fluid phases.
Convolutional NN architectures have shown promising results in assessing the probability distribution function of parameters used for feeding Monte Carlo simulations. Starting from multi-fidelity datasets linked to experiments and simulations, a coarse-grained, data-driven surrogate would be adopted to avoid high computational costs. Powerful approaches have been recently proposed in this regard, resting on the so-called physics-informed NNs (PINNs) where model-based and data-driven approaches are integrated within a unique framework in a synergic way.
As Quantum systems produce atypical patterns that classical systems are thought not to produce efficiently, results are therefore foreseen also regarding Quantum principal component analysis.
The research fellow will have to collect information on the behavior of microfluidic devices, for the analysis of the various phenomena under study. He/she will have to develop quantum machine learning models, based on what is evidenced by the experimental tests. The ability to perform analyses on quantum computers or to learn how to operate in this field is therefore required.