ASML
Graduation assignment: Research the relation between frequency decomposition, information and causal
Introduction
This is a graduation assignment for a master student in Applied Mathematics, Electrical Engineering or a related discipline with knowledge of information theory and signal analysis who is interested in using simulations and real machine data to set up experiments.
Job Mission
The complex machines made by ASML produce enormous amounts of data. Some of if it might actually contain information about failures and deviations in the system's behavior. The first challenge of diagnostics is to narrow down the list of data to look at. We could call this a reduction of the search space.
ASML is focusing more and more on big data type of approaches. Some of these approaches require a lot of computational resources (e.g. machine learning). Information theory based approaches seem to be more promising in terms of computational power required and the involvement of domain experts. Part of the processing flow is the calculation of what we call features. Each signal is described by a number of features (an analogy is the decomposition of a signal into frequency components). It is believed that information is contained in frequency components and that due frequency dependent phase delays, causality might depend on the frequency. That is, the causal direction might reverse within a relationship as a function of frequency.
Your assignment will be to use both simulations and real machine data, to set up experiments to prove or disprove the above hypothesis. Secondly, you will devise methods using power spectral density (frequency domain) analysis to further narrow down/further reduce the search space and/or find ways to use this for validation of our information theory based diagnostic methods.