Ніколаєвська Олена Анатоліївна
Leading Researcher

Nikolaievska Olena Anatoliivna

Candidate (PhD) of Physical and Mathematical Sciences, Senior Researcher

Email: elena_nea@ukr.net, NikolaevskaOA@nas.gov.ua

Biography

Nikolaevska Olena Anatoliivna, born in 1983, has been working at the VM Glushkov Institute of Cybernetics of the National Academy of Sciences of Ukraine since 2008. As a software engineer since May 2008, since November 2008 she has been transferred to the position of junior researcher, as a researcher since 2011, from 2020 to the present she has been transferred to the position of senior researcher

Specialty by diploma - "Mathematics and Fundamentals of Economics".

In 2010 she successfully defended her dissertation on specialty 01.05.02 - "Mathematical modeling and computational methods".

Scientific bases

  

Research interests

  • Least squares problem, weighted least squares problem, weighted pseudo-inverse matrix of Moore-Penrose;
  • methods and algorithms for solution SLAE with approximate initial data, singular value decomposition of the matrix, weighted singular value decomposition of the matrix;
  • mathematical models with approximate initial data;
  • algorithms for solution SLAE for computers with parallel architecture;
  • programming with multiple precision, GMP, MPFR
  • machine learning, AI, GPT models

Main results

  •  perturbation theory and error theory for weighted least squares problems are developed;
  • the errors of the weighted normal pseudoresolution in the case of mismatch of the rank of the original and perturbed matrix are obtained;
  • an original approach to proving the existence and uniqueness of a normal pseudo-solution in the case of a problem with positively determined weights is proposed;
  • developed algorithms for solving problems of weighted least squares with positively determined weights with matrices of arbitrary structure and arbitrary rank on computers of MIMD-architecture using arbitrary bit size;
  • a study of the dependence of the efficiency of algorithms on the decomposition of the original data in solving these problems, as well as on the structure and architecture of MIMD-computers.