Simulation accuracy has seen a steady increase in the past decades, to the point where extremely complex processes in virtually every field of applied science and engineering can be efficiently modelled with great details.  However, the level of predictive accuracy of such models is hindered by the uncertainties that permeate the required boundary conditions and model parameters.

Uncertainty quantification is a recent research field at the boundary between statistics, applied mathematics and engineering that aims to quantitatively identify, represent and propagate through the computational model all the sources of uncertainty in a given problem (see Figure 1). This allows researchers and applied scientists to move forward from purely deterministic predictions to “predictions with confidence”.

Common questions in uncertainty quantification include: i) how to represent the sources of uncertainty in the system (e.g. through a joint probability density function), ii) how to efficiently propagate the uncertainty in the model inputs to the outputs (e.g. through Monte-Carlo simulation and surrogate models), iii) which input parameters affect the most the prediction uncertainty (global sensitivity analysis), iv) how to calibrate the parameters of a model so as to match measured experimental data (Bayesian model inversion), etc.

The UQLab computational framework (Marelli and Sudret (2014)) developed by the Chair of Risk, Safety and Uncertainty Quantification at ETH Zurich provides a model-agnostic (black-box philosophy) software tool to address in a single user-friendly framework all the above mentioned questions. It comprises state-of-the art uncertainty quantification algorithms that can be easily deployed by non-experts.

Since its public release in July 2015, over 1100 registered users from 57 countries have incorporated uncertainty quantification in their research through UQLab. To date, UQLab has been adopted in structural and mechanical engineering, geosciences, high energy physics, computational electromagnetism, applied mathematics, computational macroeconomics, medical engineering and many other fields. Due to its ease of deployment, comprehensive documentation (manuals and step-by-step examples) and intuitive language, UQLab allows users to acquire command over advanced applied uncertainty quantification tools with unprecedented efficiency.