Dimensional Problems in Stellar Modelling: The Computational Challenge

TY - BOOK

T1 - Dimensional Problems in Stellar Modelling

T2 - The Computational Challenge

AU - Winther, Mark Lykke

PY - 2023/12

Y1 - 2023/12

N2 - Understanding stars is key to unravelling the mysteries of the Universe. From the discovery of exoplanets, to understanding large scale structure formation, and predicting the evolution of our own Sun, it all hinges on how well we can describe the structure and evolution of stars through numerical models. With the leaps forward in quantity and quality of observations of stars being made in the preceding and coming decade, it continuously challenges the precision of our models. We therefore seek to improve the models, which in this work is done in two different ways, with the keyword for both being dimensions.The first theme is related to “traditional” stellar models, in which stars are described from centre to surface as one-dimensional stellar models. The main advantage is that these are computationally inexpensive to calculate, even when spanning the entire evolution history of the stars. There are however a plethora of parameters to adjust when producing the models, such as mass and chemical composition. All of these constitute the parameter space of the models, which has a dimension for each parameter. Producing appropriate models to match the large quantities of observed stars therefore becomes a question of efficiently filling up and utilizing this parameter space.For this purpose, I have developed two new methods, which are presented in this work. Firstly, is the interpolation of stellar models, which seeks to efficiently fill up the parameter space of the models by interpolating between them, when the density of models is deemed insufficient. Secondly, I have developed a framework for matching frequency phase shift differences, a more accurate version of frequency ratios, to better utilize the present models with asteroseismology. Both of these methods contribute to increasing the usability of one-dimensional stellar modelling.The second topic of this thesis revolves around the Sun and how it can be more accurately determined in simulations of higher dimensions, specifically three-dimensional simulations. Because the Sun is our closest and best observed star, it is therefore also notoriously difficult to model correctly. Here, it is presented how I have expanded on the magneto-hydrodynamical simulation code Arepo, to allow it to accurately model the Sun. The focus is on the convective envelope, to address the discrepancies between observations and models of its convective flow. As this is an extensive project, it is currently still ongoing, and therefore this work details the current implementations, mainly being radiative transport, and the expected development of the project.

AB - Understanding stars is key to unravelling the mysteries of the Universe. From the discovery of exoplanets, to understanding large scale structure formation, and predicting the evolution of our own Sun, it all hinges on how well we can describe the structure and evolution of stars through numerical models. With the leaps forward in quantity and quality of observations of stars being made in the preceding and coming decade, it continuously challenges the precision of our models. We therefore seek to improve the models, which in this work is done in two different ways, with the keyword for both being dimensions.The first theme is related to “traditional” stellar models, in which stars are described from centre to surface as one-dimensional stellar models. The main advantage is that these are computationally inexpensive to calculate, even when spanning the entire evolution history of the stars. There are however a plethora of parameters to adjust when producing the models, such as mass and chemical composition. All of these constitute the parameter space of the models, which has a dimension for each parameter. Producing appropriate models to match the large quantities of observed stars therefore becomes a question of efficiently filling up and utilizing this parameter space.For this purpose, I have developed two new methods, which are presented in this work. Firstly, is the interpolation of stellar models, which seeks to efficiently fill up the parameter space of the models by interpolating between them, when the density of models is deemed insufficient. Secondly, I have developed a framework for matching frequency phase shift differences, a more accurate version of frequency ratios, to better utilize the present models with asteroseismology. Both of these methods contribute to increasing the usability of one-dimensional stellar modelling.The second topic of this thesis revolves around the Sun and how it can be more accurately determined in simulations of higher dimensions, specifically three-dimensional simulations. Because the Sun is our closest and best observed star, it is therefore also notoriously difficult to model correctly. Here, it is presented how I have expanded on the magneto-hydrodynamical simulation code Arepo, to allow it to accurately model the Sun. The focus is on the convective envelope, to address the discrepancies between observations and models of its convective flow. As this is an extensive project, it is currently still ongoing, and therefore this work details the current implementations, mainly being radiative transport, and the expected development of the project.

M3 - Ph.D. thesis

BT - Dimensional Problems in Stellar Modelling

PB - Århus Universitet

ER -