CSE Undergrad News

From: Randy LeVeque <rjl@uw.edu>
Date: Fri, Dec 11, 2009 at 12:40 PM
Subject: New course on High-Performance Scientific Computing coming
Spring Quarter
To: amath-current@amath.washington.edu

I will be teaching a new course *Spring Quarter* 2010 on High-Performance
Scientific Computing, appropriate for advanced undergraduates and graduate
students.  It is intended to be a broad-brush survey course as described
further below, for students who have had some programming experience.

Please help spread the word about this course.  Feel free to contact me
with suggestions for the course as well, since it is still in the planning
stage.

A pdf version of this announcement suitable for posting can be found
on the webpage
  http://www.amath.washington.edu/~rjl/hpsc10.html
where there is also a pointer to the seminar on this topic we ran last spring
as a warmup to this class, which may give more idea of the intended level.

Thanks,
 Randy LeVeque

--------------------------------------------------------------------------

NEW COURSE --- SPRING QUARTER 2010

Applied Mathematics 483/583
High-Performance Scientific Computing

Instructor: Prof. Randy LeVeque
Time: Spring Quarter 2010, MWF 8:30am (tentative), EDGE Classroom TBA
Webpage: http://www.amath.washington.edu/~rjl/hpsc10.html

This class will cover a selection of topics in high-performance computing
(HPC), briefly introducing many of the issues that arise when solving
large scale computational problems in science and engineering. In
particular, the following topics will be touched on:

 - Computer languages and issues affecting the choice of language, e.g.
  compiled vs. interpreted, procedural vs. object oriented.

 - Programming in Fortran 95 and Python/Sage as representative languages
  (prior programming experience in some language is a prerequisite!)

 - Computer architecture issues relevant to HPC, e.g., cache and memory
  hierarchies, shared vs. distributed memory, vector pipelines, GPUs, parallel
  computers from multicore laptops to supercomputers with 100,000+ cores.

 - Languages for parallel computing, in particular MPI and OpenMP.

 - Tools for managing large computer programs, e.g., makefiles, debuggers,
  version control systems such as Mercurial or Subversion.  Best practices for
  reproducible research.

 - Dealing with large datasets arising from computation or scientific
  observations.

 - Graphics and visualization of scientific data.

This is a lot of material to cover in one quarter.  The emphasis will
be on seeing key concepts, getting started using a variety of tools, and
becoming familiar with the documentation and online resources available
for further learning.  Homework assignments will involve using many of
these tools.  Other courses, such as CSE 524 (Parallel Algorithms), go
into more details of some aspects of this class and would be a natural
next step.

Prior programming experience is required, at the level of CSE 142,
AMath 301, or AMath 481/581.  Students should be comfortable installing
software on their own computers and/or using ssh for remote access
to linux machines.  Assistance and documentation will be available
(including an introduction to linux/unix), but students averse to
exploring new software and overcoming the frustrations that typically
accompany this will probably not enjoy the class.

Some background in linear algebra at the level of Math 308 or AMath
352 is recommended.  Linear algebra is the basis for much of scientific
computing and we will study examples related to matrix multiplication
and solving linear systems in particular.

CSE Undergrad News

Fwd: [Escience_bbl] New course on High-Performance Scientific Computing coming, Spring Quarter