Solving PDE Problems on Networks by Mixed Integer Programming Techniques<BR>

<P>
Aexander Martin, TU Darmstadt<BR>

<p>
Abstract: <BR>

<p>

Partial differential equations are used to mathematically describe
physical phenomena such as the flow of water or gas in pipelines.
When combinatorial decisions in networks such as where to open or
close valves or pipes must be modeled mixed integer programming
techniques are short-listed. Models and methods that incorporate both
structures are hardly available. In this talk we want to exploit mixed
integer programming techniques and investigate whether it is possible
to approximate partial differential equations in an appropriate way in
order to come up with the right combinatorial decisions. We will study
some real-world examples of this kind and discuss pros and cons of
this approach.