Illinois Institute of Technology: CS 595

CS 595 - Advanced Scientific Computing

Fall 2010

The Boss is: OUT

Site Map

Contact Information

Hong Zhang : hzhang@mcs.anl.gov

Office : SB 235C
Office Hours : R 3:00-5:30

Michael McCourt : mccomic@mcs.anl.gov (or mccomic@iit.edu)

Office : E1 105d
Office Hours : MW 10:00-1:00, TR 2:00-5:00

Lectures

Like most graduate classes, this course will have the lectures used posted here so you have access to them while you are working on your own. New lectures will be posted after the class they were presented in, and Dr. Zhang will in general bring copies of her lectures to class so that you don't have to print them out yourself.

Lecture 1 - Aug. 26

Overview of parallel computing
Classifications of parallel computers
Common terminology

Lecture 2 - Sep. 2

Introduction to MPI
Programming with MPI
Important MPI functions
Understanding data movement/communcation

Lecture 3 - Sep. 9

PETSc Introduction
Examples of PETSc applications
SNES & KSP objects
Sparse numerical linear algebra

Lecture 4 - Sep. 16

PETSc Follow-Up
Vector and Matrix Objects
Various Underlying Implementations
Object Assembly and Parallel Communication

Lecture 6 - Sep. 30

KSP Basics
Linear Solvers in PETSc
Runtime Options
Block Solvers
Intro to Preconditioning

Lecture 6 - Sep. 30

PETSc Programming Aids
Debugging
Profiling, -log_summary
Parallel Data Layout
Distributed Arrays

Lecture 7 - Oct. 7

PETSc Nonlinear Solvers
Newton-based methods
User-defined context
Jacobian evaluation via finite differences

Lecture 8 - Oct. 14

Software Engineering in C++
Data types and pointers
Memory allocation and leaks
Arrays and Pass-By-Reference

Lecture 9 - Oct. 21

Distributed Version Control
Intro to Mercurial
Revisions, branching and merging
Creating and maintaining a repository
Managing users, workflow, patches

Lecture 10 - Oct. 28

Parallel Algorithm Design
Programming Models
Data vs. Functional Decomposition
Mapping processes to processors
Application structure/Methods of parallelism

Lecture 10 - Oct. 28

Time stepping solver in PETSc
Example 7 from PETSc source

Lecture 11 - Nov. 4

SLEPc: An eigensolver using PETSc objects
Intro to eigenvalue/singular value problems
Basic usage of SLEPc: EPS, ST, SVD objects
Subspace extraction/generation

Lecture 12 - Nov. 11

Intro to Design Patterns
Facade Pattern, Bridge Pattern

Lecture 12 - Nov. 11

Complete paper describing Jacobian coloring graph theoretically

Lecture 12 - Nov. 11

Software engineering as problem solving
Dealing with complexity of software systems
Software design goals