京都大学 京都大学

Simulation and Data Science

Numbering Code Year/Term 2022 ・ Second semester
Number of Credits 2 Course Type Lecture
Target Year Master's students Target Student
Language English Day/Period Thu.3
Instructor name IMADERA KENJI (Graduate School of Energy Science Associate Professor)
Outline and Purpose of the Course Simulation and data science are main research approaches based on computers in modern science. Simulation science can be defined as a deductive methodology to find an approximate solution of a given governing equation by using some numerical methods. On the other hand, data science is an inductive methodology that extracts and estimates the rules behind obtained data by using some mathematical statistics.
This course will help students to understand the basic theory of such simulation and data science and write the program by themselves, aiming to acquire the abilities at practical level.
Course Goals By the end of this course, students should be able to:
(1) acquire the abilities to write the program by themselves with the programing language Python;
(2) understand some methods for solving the ordinary differential equations numerically as the basics of simulation science;
(3) understand regression analysis and neural network model as the basics of data science;
(4) apply the learned skills to their own research field by studying how the methodologies in (2) and (3) are utilized at the forefront of research.
Schedule and Contents Week-1：Guidance
After the overall guidance of this course, Jupyter Notebook, a web-based interactive computing platform for the Python programming will be installed to student's laptop.

Week-2：Basics of hardware and numerical calculation
In addition to CPU/GPU hardware system and memory architecture, the floating-point representation of numbers will be learned.

Week-3, 4, 5：Basics of Python
Students will learn variable setting, iterations and branching in the Python programming and then write the related program by themselves.

Week-6, 7：Numerical methods for ordinary differential equations
The Euler and Crank-Nicolson schemes will be learned as numerical methods for solving ordinary differential equations. Students will also write the related program by themselves.

Week-8, 9：Regression analysis
In addition to the simple regression model based on the least-square method and the gradient-descent method, the statistical regression model based on the logistic regression will be learned. Students will also write the related program by themselves.

Week-10, 11, 12：Neural network model
As one of main methodologies in data science, the Neural Network (NN) model will be learned. Construction of the NN model for recognizing hand-written data and that for inferring the ordinary differential equation from obtained data will be also made.

Week-13, 14：Simulation and data science in plasma physics
In plasma physics, simulation and data science have become an important methodology. Students will study how both are utilized at the forefront of plasma research and acquire the abilities for applying these approaches to their own research field.

Week-15：Feedback
Evaluation Methods and Policy The grading policy will be as follows:
(A) Mini test (every lecture) 30%
(B) Homework of programming (roughly once per two lectures) 30%
(C) Final report 40%
Course Requirements Basics of calculus are required.
Experiences for programming are expected but not required.
Study outside of Class (preparation and review) Students are strongly expected to pre-study the lecture slides on the PandA and solve the quiz in advance.
Textbooks Textbooks/References Lecture slides will be uploaded on the PandA before the class.
PAGE TOP