Advisor
Research Topic
Global optimization, Numerical analysis, Space-filling curves.
Research Abstract
Multi-extremal problems (global optimization problems) are very important for many practical applications. The aim of global optimization is to find algorithms to solve problems where the object function has many local extrema and its analytic expression is not known (black-box function). In the literature there are several optimization techniques but they are not always efficient because in order to have more accuracy, many function evaluations are necessary and this produces a huge computational cost.
For these reasons, innovative approaches and new computational paradigms are of particular interest.
Methods that will be taken into consideration in my PhD research are those based on a fractal approach to reduce the dimension of the problem (using adaptive diagonal curves and space filling curves) and methods which tune the behaviour of the algorithm according to the shape of the objective function and the constraints of the problem.
An emerging computational paradigm called Infinity Computing will be applied for this purpose.