Optimisation

More comming soon!

The NordConsNet Workshop 2017
The NordConsNet Workshop 2017 of The Nordic Network for researchers and practitioners of Constraint programming held Monday 22 May 2017 in Uppsala, Sweden. Click here for more information.

Mathematical optimisation consists of theories, models, and methods for formulating and solving optimisation problems that arise in a wide spectrum of applications.

The optimisation arena at the Department of Information Technology serves as a platform to enable researchers to collaborate and network. The purpose of the arena is:

• to strengthen the research by bringing together knowledge in optimisation and various application domains.
• to identify optimisation problems of relevance and develop problem-solving techniques.
• to pursue synergy effects and added value in research where optimisation is a significant component.

Division of Computing Science

• Models and methods for fundamental capacity characterisation and optimisation for information and communication technology and networks.
• Large-scale optimisation for transportation systems and logistics, and applications in biology, medicine, and healthcare.
• SAT/SMT techniques for analysis, synthesis, and repair of programs or models. This includes the development of new solvers in this area, in particular for data-types like floats, bit-vectors, and strings, and considering extensions like interpolation and fixed-point solving.

The ASTRA Group on Combinatorial Optimisation addresses practical applications and the following research topics in combinatorial (or: discrete) optimisation:

• Improved inference for predicates for constraints on integer timeseries, and inference for predicates for constraints with decision variables of type String.
• High-level language for specifying local-search heuristics as annotations to declarative constraint-based models, and extension of our back-box local-search backend to the MiniZinc language to support search annotations, string variables, and string constraints.

Division of Computer Systems

• Optimisation problems in sensing and communication in Internet of Things (IoT), including incentive allocation in mobile crowdsourcing, coordination of stationary and mobile sensors in sensing and communication, etc.
• Optimisation techniques for smart-city applications and city planning. 

Division of Scientific Computing

• Parameter estimation and likelihood maximisation in Bayesian inference with (ordinary or partial) differential equation modelling. The models call for simplification to be included in an optimisation loop of solving repeating equations.
• Form and topology optimisation with partial differential equations (PDE) as constraints, and PDE-constrained optimisation problem in general with many control variables (with applications within geophysics).

Division of Systems and Control

• Estimation of parameters in static as well as dynamic models, ranging from linear models resulting in convex problems, to nonlinear ones for which the corresponding non-convex problems require more intelligent algorithms.
• Formulating real-world problems as tractable optimisation problems that can be solve within a reasonable time frame, and developing fast application-specific minimisation methods.
• The target applications are machine learning, system identification, automatic control, Markov chain and sequential Monte Carlo, network inference and control, target tracing, filter design, beam forming and array processing, spectral analysis, etc.

Visual Information and Interaction

• TBA

Algorithms and Data Structures III (1DL481, 5 credits) is taught every spring term. The course includes introductory material on combinatorial optimisation: mixed integer linear programming (MIP), stochastic local search (SLS), Boolean satisfaction (SAT), and SAT modulo theories (SMT).

Combinatorial Optimisation using Constraint Programming (1DL441, 10 credits) is taught every autumn term. The course explains in detail the algorithms behind solvers of constraint programming (CP) technology.

Modelling for Combinatorial Optimisation (1DL449, 5 credits) is taught every spring term. It addresses declarative problem modelling, with experiments on solvers from a wide range of combinatorial optimisation technologies: CP, MIP, SAT, SMT, SLS, and hybrids.

Optimisation (1TD184, 5 credits) is taught every autumn term. The course covers mathematical modelling and formulation, and basic concepts and methods in optimisation.