Dr Cezar Ionescu
Cezar's research involves applications of concepts and methods from functional programming, type theory, and category theory to the analysis of a wide range of problems in artificial intelligence and data science, from correctness of software models, to the improvement of our understanding of social science concepts (needed, e.g., in order to ensure "fairness", "transparency", "accountability" of algorithms). The current domain of application is the correctness and explainability of machine learning algorithms, especially decision making and reinforcement learning. Additional interests are new approaches to the teaching of mathematics at undergraduate level (via domain-specific languages of mathematics) and the "computational thinking" aspects of the GCSE in computer science.
- Cezar Ionescu, Patrik Jansson, Nicola Botta. Type Theory as a Framework for Modelling and Programming. Leveraging Applications of Formal Methods, Verification and Validation. Part I, Modelling. LNCS 11244, Springer, 2018. Online at Springer
- Nicola Botta, Patrik Jansson, Cezar Ionescu. The impact of uncertainty on optimal emission policies. Earth System Dynamics, 9(2): 525-542, 2018. https://www.earth-syst-dynam.net/9/525/2018/
- Nicola Botta, Patrik Jansson, Cezar Ionescu. Contributions to a computational theory of policy advice and avoidability. Journal of Functional Programming, 27. CUP, 2017.
- Cezar Ionescu. Vulnerability Modelling with Functional Programming and Dependent Types. Mathematical Structures in Computer Science. Volume 26, Special Issue 01, January 2016, pp 114-128. doi Preprint
- Jochen Hinkel, Daniel Lincke, Athanasios T. Vafeidis, Mahé Perrette, Robert James Nicholls, Richard S. J. Tol, Ben Marzeion, Xavier Fettweis, Cezar Ionescu, and Anders Levermann. Coastal flood damage and adaptation costs under 21st century sea-level rise.PNAS, vol. 111 no. 9, March 4, 2014.
- Cezar Ionescu, Patrik Jansson. Dependently-typed programming in scientific computing: Examples from economic modelling. Implementation and Application of Functional Languages Lecture Notes in Computer Science 2013, pp 140-156. Preprint
- Nicola Botta, Cezar Ionescu, and Edwin Brady. Sequential decision problems, dependently typed solutions. In proceedings of PLMMS 2013.
- Cezar Ionescu, Patrik Jansson. Testing versus Proving in Climate Impact Research Proceedings of the 18th Workshop Types for Proofs and Programs (TYPES 11), 19 pp. 41-54.
- Nicola Botta, Antoine Mandel, Cezar Ionescu, Maureen Hofmann, Daniel Lincke, Sibylle Schupp, Carlo Jaeger. A functional framework for agent-based models of exchange. Applied Mathematics and Computation 218(8):4025-4040 (2011).
- Nicola Botta, Cezar Ionescu. Relation-based computations in a monadic BSP model. Parallel Computing, Elsevier, Sept. 2006.
- Cezar Ionescu, Richard J.T. Klein, Jochen Hinkel, K.S. Kavi Kumar, and Rupert Klein. Towards a formal framework of vulnerability to climate change. Environmental Modeling & Assessment, February 2009, Volume 14, Issue 1, pp 1-16.
- Reinforcement Learning for Functional Programmers. Meeting of IFIP Working Group 2.1 on Algorithmic Languages and Calculi, July 1-6 2018, Brandenburg. Abstract
- Folding Directed Acyclic Graphs. Meeting of IFIP Working Group 2.1 on Algorithmic Languages and Calculi, October 16-20 2017, Lesbos. Abstract
- DSLM - Presenting Mathematical Analysis Using Functional Programming. Meeting of IFIP Working Group 2.1 on Algorithmic Languages and Calculi, August 24-28 2015, Göteborg. With Patrik Jansson.
- Domain-Specific Languages of Mathematics. DTP 2014, Vienna, 2014-07-13.
- Dependently-typed Programming in Scientific Computing. IFL 2012, August 30 - September 1 2012, Oxford. Video.
- Increasingly Correct Scientific Programming. Invited talk, CICM 2012, July 8-13 2012, Bremen.
- Calculation and Communication. Meeting of IFIP Working Group 2.1 on Algorithmic Languages and Calculi, February 6-10 2012, Rome.
- Economic Equilibria in Type Theory. Types 11, September 12-15 2011, Bergen.
- Dependently-typed Programming in Economic Modelling. DTP11, August 27 2011, Nijmegen. Code.
- Proving versus testing in climate impact research. Types 10, October 13-16 2010, Warsaw.
- Using dependent types in models of climate change impacts. DTP10, July 9-10 2010, Edinburgh.
- Vulnerability modeling and monadic dynamical systems. Freie Universität Berlin, 2009.
- Machine Learning, University of Oxford, Trinity 2019.
- Introduction to Artificial Intelligence. Kellogg College and the United Arab Emirates. Professional developement course, aimed at UAE government officials, with the goal of ensuring that "government officials across the government are not just 'A.I. informed', but 'A.I. literate and capable'"(the UAE Minister of State for Artificial Intelligence)
- Domain-Specific Languages of Mathematics, Chalmers, Spring 2017.
- Domain-Specific Languages of Mathematics, Chalmers, Spring 2016.
- Category Theory for Functional Programming, Chalmers, Fall 2014.
- Introduction to Programming, University of Gothenburg, Fall 2014.
- Introduction to Programming, University of Gothenburg, Fall 2013.
- Advanced Functional Programming, FU Berlin, Summer Semester 2013.