Parallel Iterative Computation (CS623)

NOTE TO STUDENTS: This course is is about "Parallel Iterative Computing". It is a scientific computing oriented courses and teaches numerical technical techniques to solve large sparse linear systems on parallel computers. Please read the following materials to see if you are REALLY interested in taking this course.

Semester: Spring, 2008.
Class: 2:00pm - 2:50pm, MWF.
Classroom: C226 OHR.

Instructor: Jun Zhang, E-mail:jzhang@cs.uky.edu, Tel:257-3892.
Office: 763F Anderson Hall.
Office Hours: WF: 1:00pm - 2:00pm, and by appointment.

IMPORTANT NOTES:

Please note that, in addition to the two 500 level CS courses as prerequiste, you need to have some background in parallel programming using MPI or to spend some time learning MPI programming. A very simple tutorial book is the MPI Tutorial (pdf file).

Text Book: Briggs and Saad's books will be used as text book. Class notes will be available online. Nevertheless, here is a list of reference books:

A Multigrid Tutorial, by William L. Briggs; This is the slide in a pdf file.

Iterative Methods for Sparse Linear Systems, part1.pdf, part2.pdf, part3.pdf, part4.pdf, by Yousef Saad.

Template: Iterative Solution of Linear Systems, by R. Barrett, et al.

Parallel Computer: We plan to use the university's NCX parallel computer. MPI programming will be required.

Here is the syllabus.ps (syllabus.pdf). You should also read the rule.ps (rule.pdf) on homework and grading.

Class Notes

Lecture 1, and the pdf file
Lecture 2, and the pdf file
Lecture 3 Part 1, a multigrid code for 2D Poisson equation
Lecture 3 Part 2
Lecture 3 Supplement
Lecture 4
ps file , Lecture 5
ps file, Lecture 6
ps file , Lecture 7
Lecture 8 Part 1
Lecture 8 Part 2
Lecture 9 Part 1
Lecture 9 Part 2
Lecture 9 Suplement 1
Reading Materials Developments and Trends in the Parallel Solution of Linear Systems, by I. S. Duff and H. van der Vorst.
MPI Supplemental Reading: part 1, part 2, part 3, part 4.

Note: The key components of this course is to study two most popular iterative methods: Krylov subspace methods and multigrid method. The following are two text books:

Iterative Methods for Sparse Linear Systems, by Yousef Saad. This book can be purchased from many online book stores at about $72 a piece.
A Multigrid Tutorial, By William Briggs. This book can be purchased from SIAM online at about $5 a piece.

None of these books is absolutely necessary, as we have the online copy of their essential parts.

Warm-Ups

Set up your NCX parallel computer Account.
At this time, all students should have access to the university's NCX Supercomputer.

Homework and Projects

Final in 2001 sample test, pdf file.

Homework 1, due on February 11, 2001.
Project 1, and code, due on February 22, 2001.
Project 2, pdf file, , due on March 1, 2001.
2001 Midterm.
Homework 2, pdf file, due on March 10, 2002.
Project 3, due on March 24, 2002.
Homework 3, due on April 25, 2001.

PLEASE NOTE

Here are some exciting research projects in scientific and parallel computing and my ambitious research team.

Supercomputing Technology News

On February, 1998, IBM announced the world's first 1000 MHZ chip, three times faster than Intel's fastest chip. Don't worry about that, your parallel computers could be even faster.
In May, 1998, Los Alamos built a mail-order supercomputer and it is among the word's fastest. Take a look at the new supercomputer, named Avalon.
The TOP500 List of Supercomputer Sites in June 1998.
Newsgroup comp.parallel archieve.

The World's Fastest Supercomputers

Here is a list of parallel scientific computing courses offered at several universities:

On-line study materials on parallel computing:

Designing and Building Parallel Programs by Ian Foster.

Last modified on Monday, January 7th, 2008.