Announcements

Schedule

• Saturday, August 17       :   9:00 am Summer School begins at Brown. Watson CIT Building, Room 265
• Saturday, August 17       :   6:00 pm Welcome Barbeque behind the Applied Math Building
• Monday, August 19        :   6:00 pm Karoke at Hot Club ( WALKING DIRECTIONS )
• Tuesday, August 20       :   7:15 pm Workshop Dinner at Al Forno restaurant ( WALKING DIRECTIONS )
• Wednesday, August 20  :  Presentations by students on their project so far
• Monday, August 26        :   9:00 am Summer School begins at Kobe. Location TBD
• Monday, August 26        :   Time TBD Party. Location TBD
• Wednesday, August 28  :   Time TBD Dinner at a typical Japanese restaurant. Location TBD
• Friday, August 30           :   Time TBD Final Presentation on Group Projects
• Friday, August 30           :   Time TBD Farewell Barbeque at Kobe. Location TBD

Class Files

Presentations

• Monte Carlo Calculation of PI Using MPI
( typo on Monte Carlo slide: inequality should be less than 0.5, not 1! )
• Linear Algebra and MPI
• Introduction to Differential Equations
• Numerical Solution of PDE: Spectral Element Method
• Introduction to ScaLAPACK

Resources for Presentations

• Solution to Finding PI using MPI
• Example Matlab implemenation of Jacobi and Gauss Elimination
• Air Resistance Video
• Heat Equation Video
• Wave Equation Video
• Spectral Element Video

ScaLAPACK Tutorial Files

• Instructions for Using OSCAR
• ScaLAPACK Header Files
• Instructions for Running ScaLAPACK example programs
• ScaLAPACK Example Programs: Personal Computer Version
Note: these example programs were converted into C from the NERSC Fortran ScaLAPACK examples
• ScaLAPACK Example Programs: OSCAR Version
Note: the difference between the "personal computer" and "Oscar" versions of these programs is the target to the ScaLAPACK libraries in the Makefile. Everything else is identical.
• Laplace's Equation Solver Template
• Laplace's Equation Solver Solution: Personal Computer Version
• Laplace's Equation Solver Solution: OSCAR Version

Resources

High Performance Computing

• ETH HPC Course Notes

MPI

• MPI Quick Refrence Guide
• MPI beginner tutorial

ScaLAPACK

• ScaLAPACK Users' Guide
• Index of ScaLAPACK Double Precision Routines
• Netlib C Scalapack Documentation
• IBM ScaLAPACK Doumentation
• Example Fortran ScaLAPACK programs
• Example C ScaLAPACK program
• BLACS Quick Reference Guide
• PBLAS Quick Reference Guide
• ScaLAPACK Quick Reference Guide
• Netlib Scalapack installation instructions

Project Resources

Project: Oscillons

• Tomography of Reaction-Diffusion Microemulsions Reals Three-Dimensional Turing Patterns. Physical Review E, 80, 066204 (2009).
• Instabilities of a three-dimensional localized spot. Science 331, 1309-312 (2011).

Before You Arrive

1. Has access to a machine with a gcc and MPI compiler and the ability to link to a ScaLAPACK library. You have the option of either using a supercomputer or getting the libraries working on your computer. Kelly and Nat can help you with either option, but we suggest using a supercomputer because everything is already set up on the machine and the process to get an account is very simple.
   
   
   1. Using Oscar (supercomputer option for Brown students):
      If you do not already have an an Oscar account, please fill out the CCV form Use Bjorn Sandstede as the supervisor and say that it is for the summer school in scientific computing at Brown in August 2013. Once you have an account, please refer to the CCV user manual to get started compiling and running programs on Oscar.
   
   
   2. Installation on your personal computer:
      If you decide to install ScaLAPACK on your own machine, ScaLAPACK can be installed by following the Netlib Scalapack installation instructions. We recommend that you use the ScaLAPACK installer if you have a linux machine and that you download the blas and lapack directories using the installer. It can otherwise be really tedious to try to link the libraries properly by hand.
   
   
2. Refamiliarize yourself with the basics of:
   1. Navigating a Unix shell
   2. Programming in C/C++
   3. Basic MPI programming
   
   
   To do this we recommend the MPI beginner tutorial. If you understand the code in that tutorial and can successfully compile and run them in the command line, then you will be fine for the course. If any of the Brown students who are less familiar with scientific computing are interested, we will organize a one afternoon remedial session to help step through the tutorial. Please let us know if you are interested in this session. When the workshop begins, we will spend a morning doing some exercises using MPI to make sure that everyone is comfortable before we begin working on projects. We will also spend a day or so teaching some foundations in numerical linear algebra and numerical solution of partial differential equations so that we have a good theoretical foundation before we start working on our projects.