Numbering Code	G-INF00 58020 LJ11 G-INF00 58020 LJ10	Year/Term	2022 ・ First semester
Number of Credits	2	Course Type	Lecture
Target Year		Target Student
Language	Japanese	Day/Period	Thu.5
Instructor name	YAMASHITA NOBUO (Graduate School of Informatics Professor) SATO HIROYUKI (Graduate School of Informatics Program-Specific Associate Professor) USHIJIMA SATORU (Academic Center for Computing and Media Studies Professor) FUJIWARA HIROSHI (Graduate School of Informatics Associate Professor) ENOMOTO TAKESHI (Disaster Prevention Research Institute Professor)
Outline and Purpose of the Course	The aim of this course is to obtain the basic knowledge and skills for computational science. For this purpose, we give lectures and exercises as follows. In lectures, we give the following topics: numerical algorithms, computer architecture, basics of high precision computation, high performance computing, parallel computing techniques, and applications of computational science in various fields. In exercises, using the C language, students acquire optimization techniques for serial computing, which is the most important technique to utilize computers. Students also acquire parallel computing techniques for computers with multicore CPUs and computers with distributed memory.
Course Goals	Students will acquire the basic knowledge and skills for computational science. The goal is to obtain optimization techniques for serial computing, and to be able to use two parallel computing techniques, OpenMP and MPI.
Schedule and Contents	All schedules are as follows. ・Lecture of numerical computation (6 times; Lecture) (1) Accuracy and stability of numerical computation (2) Speed-up of numerical computation, BLAS&LAPACK (3) Optimization algorithms ・Lecture of computational science (3 times; Lecture) (1) Speed-up of serial computation and parallel computation (2) Multi-thread parallel programming (OpenMP) (3) Multi-process parallel programming (MPI) ・Supercomputer exercise (3 times; Exercise) (1) Speed-up of serial computation (2) OpenMP (3) MPI ・Case study (3 times; Lecture)
Evaluation Methods and Policy	The understanding level of "Lecture of numerical computation" is evaluated by reports. Students also submit three reports corresponding to "Supercomputer exercise" (Speed-up of serial computation, OpenMP, MPI). Reports about "Lecture of numerical computation": it is worth 30 points. Reports about "Supercomputer exercise": it is worth 45 points. Attendance and active participation through questions is worth 25 points.
Course Requirements	ECS-ID is required for all attendees prior to the exercise classes. A laptop computer is used in this course. It is better to bring a laptop computer.
Study outside of Class (preparation and review)	For exarcise, students should write C language or Fortran programs. Students are required to prepare one of these programming language.
Textbooks	Textbooks/References	Handouts to be distributed. Not specified.
References, etc.	Not specified.