The Dark Side of Object-Oriented Modelling: Numerical Problems, Existing Solutions, Future Perspectives.

Abstract:

Object-Oriented Modelling languages allow to tackle the system-level modelling of engineering system in a modular and convenient way, ideally focusing on clarity and generality of the code rather than on numerical robustness and computational efficiency. Symbolic and numerical methods are now available to generate efficient simulation code from this models; however, there are still some problems (most notably initialization) that are far from being solved in a completely satisfactory way from the end-user perspective. O-O models are now also starting to be used for optimization, which has very different numerical requirements than simulation; numerical and symbolic methods similar in purpose to those already existing for simulation are badly needed in order to automatically turn high-level O-O models into low-level equations that can be optimized reliably and efficiently, thus avoiding the manual tuning of the code for this purpose. The talk will present the current state of the art and point out the needs that haven't been satisfied yet, making the case for new cross-disciplinary research in the field.

Presentation Slides

Biography:Francesco Casella got his Master's and PhD degrees from Politecnico di Milano, where he is currently Assistant Professor and teaches courses of Automatic Control and Modelling and Control of Continuous Processes. His main research interests are model-based control of energy systems, with a strong emphasis on dynamic modelling, and object-oriented modelling methods, tools and applications at large. He is an active member of the Modelica Association since 2003, and a member of the board of the Open Source Modelica Consortium. In 2009, he has been the main organizer of the 7th Modelica Conference in Como.