Re: Special matching problem

Zig wrote:

> I do not quite understand the question.
> According to (2), what if we choose m(t) = max_p { a(t, p) } ? Isn't
> that the answer?

Hi Zig,
thank you for your fast answer!

Well, that's not really the solution, because this could lead to an
assignment where multiple tasks are mapped to one processor. Surely, I
could think of a heuristic which handles this case, but this could lead to
other conflicts => bipartite matching should be the right modelling from my
point of view.

Maybe my question was not well stated. I want to assign the tasks to the
processors such that:
- a processor is only used by one task, i.e. there is a 1:1 relation between
tasks and processors
- each task t is satisfactory with the assignment, i.e. there is no other
task t' which can use a processor p while the affinity of t to p is higher
than the affinity of t' to p.

Michael
.