Stripinis Linas - Institute of Data Science and Digital Technologies

Doctoral studies

Theme Improvement, development and implementation of derivative-free global optimization algorithms

Supervisor dr. Remigijus Paulavičius

Study time 2016 10 01–2020 09 30

Aims:

To increase efficiency of state-of-the-art DIRECT-type global optimization algorithms on optimization problems with many local minima, and where the solution with high accuracy is needed;
To extend DIRECT-type algorithms on global optimization problems with general and hidden constraints;
To develop efficient open-source derivative-free algorithms by taking into account algorithmic improvements, efficient data structures, and parallelization techniques.

Main problems:

To evaluate the performance of the existing state-of-the-art DIRECT-type global optimization algorithms and determine their weaknesses;
To improve existing and develop new algorithms considering identified drawbacks;
To develop a general constraint-handling strategy in the DIRECT algorithmic framework;
To develop an auxiliary functions-based DIRECT-type algorithm for optimization problems with hidden constraints;
To implement efficient sequential and parallel versions of proposed algorithms, and compare their performance to other related approaches;
To efficiently solve challenging practical (potentially black-box) optimization problems using implemented and openly accessible tools.

Scientific and pedagogical background

Research Interests

Presentations at Scientific Conferences

2019: The International EURO mini Conference Modelling and Simulation of Social-Behavioural Phenomena in Creative Societies, September 18–20, Vilnius, Lithuania. Importance of optimization techniques for the social sciences.
2018: 10th International workshop on Data Analysis Methods for Software Systems (DAMSS), November 29 – December 1. Druskininkai, Lithuania. Improved DIRECT-type algorithms for generally constrained global optimization problems
2018: 16th EUROPT Workshop on Advances in Continuous Optimization, 2018 July 12-13, Almeria, Spain. Improved DIRECT-type algorithm for constrained globalo ptimization problems.
2017: 9th International workshop on Data Analysis Methods for Software Systems (DAMSS), November 30 - December 2, 2017. Druskininkai, Lithuania. Improved DIRECT-type algorithms for generally constrained global optimization problems.

Doctoral conferences:

Annual Reporting Doctoral Conference of Informatics, October 22, 2020 Vilnius „Improvement, development and implementation of derivative-free global optimization algorithms“ (Slides)
Annual Reporting Doctoral Conference of Informatics, October 30, 2019, „Vilnius „Improvement, development and implementation of derivative-free global optimization algorithms“ (Technical Report, Slides)
Annual Reporting Doctoral Conference of Informatics, October 24, 2018, „Vilnius „„Improvement, development and implementation of derivative-free global optimization algorithms“ (Technical Report, Slides)
Annual Reporting Doctoral Conference of Informatics, October 17, 2017, „Improvement, development and implementation of derivative-free global optimization algorithms“ (Technical Report, Slides)

Scientific publications

Scientific publications in peer-reviewed periodicals journals with citation index (Impact Factor) indexed by Clarivate Analytics Web of Science database: