29.5.2020 Development of nuclear energy density functionals from optimization to uncertainty analysis (Haverinen)

The nuclear energy density functionals are still the only microscopic models that can be applied throughout the nuclear chart, including even the heaviest nuclei. These models can be used to give predictions for nuclear bulk properties such as binding energies and radii. The long-term goal is to further enhance the prediction power of these models, and thus provide even more accurate and precise information for the needs of nuclear astrophysics e.g. for the studies of the r-process, which is responsible for the production of many heavy elements in stars. The doctoral thesis of Tiia Haverinen at the Ä¢¹½Ö±²¥ deals with the development of nuclear energy density functionals from multiple points of view.
Published
29.5.2020

The development of a model can be divided roughly into four phases: theoretical formulation and derivation of physical quantities, optimization of the model parameters, testing and application of the model, and uncertainty analysis. All these phases are covered in the thesis.

During the doctoral studies, certain physical quantities were derived from a theoretical model. These quantities are used in the optimization of the model parameters. The nuclear energy density functionals include a set of parameters whose values cannot be derived from theory at present. In addition, a computer program for fitting the model parameters to experimental data, was written and implemented. The calculations are computationally demanding, and the optimization work was therefore performed on the super computers of IT Center for Science CSC.

The uncertainties and error composition of the previously developed UNEDF models were studied as well. It was discovered that the theoretical uncertainties of binding energies increase rapidly when the neutron number of the nucleus of interest increases. The error composition has been changed between the older UNEDF0 and the newer UNEDF2 model: the errors of binding energies have decreased and the error is split more equally among the model parameters.

The research as a whole has been versatile.

"I could combine theoretical physics, high-performance computing, statistics and optimization methods in the project. These kinds of projects could be carried out together with other departments and faculties", Haverinen remarks.

The dissertation is published in JYU Dissertations series, number 222, 2020, Jyväskylä. ISBN 978-951-39-8170-9 (PDF), URN:ISBN:978-951-39-8170-9, ISSN 2489-9003
Link to publication:

The doctoral thesis was supported by Finnish Cultural Foundation (North Karelia Regional Fund) and the Ä¢¹½Ö±²¥. The CSC-IT Center for Science Ltd., Finland, is acknowledged for the allocation of computational resources.

Tiia Haverinen graduated from the high school of Valtimo in 2011. She completed her M.Sc. in theoretical physics in the Ä¢¹½Ö±²¥ in August 2015 and worked there as a doctoral student at the Department of Physics until December 2019. Starting from January 2020 Haverinen has been employed as a data scientist at Gofore.

M.Sc. Tiia Haverinen defends her doctoral dissertation in theoretical physics "Development of nuclear energy density functionals from optimization to uncertainty analysis" on Friday 29 of May starting at 12 o'clock at the Department of Physics at the Ä¢¹½Ö±²¥. The Opponent is Professor Gianluca Colo from the University of Milan (Italy) and the Custos is Senior Researcher Markus Kortelainen from the Ä¢¹½Ö±²¥. The doctoral dissertation will be held in English.

The dissertation is held online. Link to the Zoom Webinar (Zoom application or Google Chrome web browser recommended):
Webinar ID: 625 0790 1866

For further information
Tiia Haverinen, tiia.k.haverinen@student.jyu.fi

Communications officer Tanja Heikkinen, tanja.s.heikkinen@jyu.fi, tel. 358 50 581 8351
The Faculty of Mathematics and Science:
/science/en
FB: Twitter: