Reducing Uncertainty

Uncertainty is an unavoidable fact of life. Most processes, whether in science or society, are highly complex and hence hard to predict. At CWI, research is focused on investigating the nature of uncertainty inherent in complex processes of all kinds, covering a wide range of fundamental questions. Can we chart the various sources of uncertainty in a complicated process and how do they propagate towards the end result? More specifically, how do noisy input data, and approximations or even misspecifications of models, impact on the uncertainty of the outcome? What noise characteristics capture pre-defined properties, and how can we accurately compute them, either through analytical means or via simulation? What amount of variation is intrinsic to a process and therefore unavoidable, and closely related to this, when is a result unusual and indicative of aberrant causes? Or – cunningly turning the tables – can we exploit our understanding of the nature of uncertainty to create applications in which it is maximized, for example in order to devise secure cryptographic encodings.

As uncertainty can never be eliminated, these questions are highly relevant for a host of pressing societal concerns. Examples include the optimal deployment of limited resources when information is incomplete or noisy, risk assessment to safeguard critical infrastructures, and taking informed and rational decisions in the face of global challenges including resource security, climate change and extreme events such as tsunamis or nuclear accidents, that - although unlikely to happen - would have a huge impact if they do.

By framing these questions related to the nature and quantification of uncertainty in an overarching scientific knowledge context, precise answers become possible. As uncertainty affects every aspect of our lives, these investigations are therefore key to forging a future in which science and society can continue to thrive in the face of an unrelenting increase in complexity and its accompanying risks.