Matlab Distributed Computation Tutorial

Description

This tutorial is aimed at researchers and engineers who want to use Matlab for distributed applications, such as the generation and processing of data on multiple machines. This kind of setup is especially useful when calculations can be executed in parallel, speeding up the process.

The solutions developed in the examples are written in Java RMI and controlled by a Matlab script. This is a good fit, as the Matlab interface has been based on Java for quite a while now. In addition, Java RMI rescues the programmer from having to write low-level communications protocols on his or her own. Since toolboxes for parallel computation have been available for some time, this tutorial does not lead up to a program for out-of-the-box usage. Instead, it explores different possibilities for writing various kinds of distributed Matlab applications using different concepts.

The use of Java will also allow our Matlab scripts to communicate with elements of other Java programs, be it business objects or GUIs on remote machines.

For simply using the examples in the text, Matlab and Java knowledge is required. A deeper understanding of the Java code requires a basic knowledge of Java RMI.

Contacting the Author

If you have any questions, comments or suggestions (for changes, enhancements, examples, etc.) concerning this tutorial, found an error or simply would like to tell how you used it, you are welcome to contact the author, Gabor Cselle. Send rants and flames to gabor (at) student [dot] ethz [dot] ch.

Acknowledgements

Thanks go out to the following people: Florian Herzog, Gabriel Dondi, and Simon Keel of the Institute for Measurement and Control at ETH Zurich, who gave tasks massive enough to justify writing distributed Matlab applications; plus Fabian Siegel, Markus Egli, and Daniel Dalquen, who saw through the drafts and gave encouraging feedback.

Copyright

The tutorial is copyright © 2004 Gabor Cselle, but may be copied and distributed in any form, as a whole or in part, as long as this document is cited as the source. Matlab is a registered trademark of MathWorks, Inc., Natick, MA, USA. The author gives no guarantee as to the functionality of the contents of this tutorial and disclaims all liability.

Version Information

• Version 1, February 3, 2004: First version
• Version 2, March 18, 2004: Several small corrections, renamed "calcstations" to "workers" in scenario 3.