[Haskell-cafe] Re: finding the right mathematical model
apfelmus at quantentunnel.de
Wed Jul 7 05:42:01 EDT 2010
Günther Schmidt wrote:
> Hi list,
> the problem I have stems from the app I had developed. What my app does
> is to split the money a hospital receives for a case to the departments
> involved in a fair way.
> An additional requirement however was to allow the users of the app to
> re-map any revenue shares credited to certain departments to other
> departments. Such cases are sometimes due to politics within the
> hospital and also have more legitimate reasons, like saying the
> radiology should not receive shares for surgical procedures but those
> shares should be redirected to the "General surgery" department.
> The feature is already implemented, but I'm not pleased with it,
> especially since I did not develop a mathematical model for it.
> It boils down to model mappings, or rather what sort of data structure
> would be suited for this kind of thing.
> Dept A is "mapped" to itself
> A -> A
> Dept B is mapped to Dept C
> B -> C
> Dept C is mapped to Dept C
> C -> C
> Dept D is mapped to Dept A
> D -> A
> It should not be possible to construct looping mappings, ie.
> 1. A -> B
> 2. B -> C
> 3. C -> A
> What sort of model would be suitable to describe this, some sort of matrix?
You probably want a graph where the nodes represent departments and
edges represent the mappings. To implement graphs in Haskell, have a
look at the functional graph library
If that's too complicated for you and your graphs are really small, you
can also use a toy implementation like
type Graph = [(Node, -- Department
[Node]) -- List of Departments it shares revenue to
To test whether a graph has cycles ("looping mapping"), you can use a
More information about the Haskell-Cafe