[CM] Arno in Grace?

Torsten Anders torsten.anders at beds.ac.uk
Wed Dec 28 15:14:50 PST 2011


Dear Bill,

Thanks for your feedback. I would certainly appreciate seeing any further work of you in this area! 

> One thing I know I'd change -- any search would use the genetic algorithm, not brute force!  

You are of course right that brute force can result in unreasonably slow performance. However, there are many smarter approaches out there, not only genetic algorithms. For example, constraint propagation can reduce the size of the search space by orders of magnitudes. Also, the order in which variables are visited can have a similar impact. Here is a link to a book chapter I wrote for musical problems on the latter factor (draft version of [1]).

 http://cmr.soc.plymouth.ac.uk/tanders/publications/Anders-VariableOrderings-2011-draft.pdf 

If you are looking for highly efficient and flexible constraint solver you may want to check out http://www.gecode.org/.

Best wishes,
Torsten

[1] Anders, T. (2011). Variable Orderings for Solving Musical Constraint Satisfaction Problems. In G. Assayag, C. Truchet (Eds.), Constraint Programming in Music. Wiley.

On 28 Dec 2011, at 19:30, Bill Schottstaedt wrote:
> I notice you mention my old counterpoint solver in your essay.
> Several times since then I've started on a new version, aimed
> at Bach, not Fux, but I get side-tracked.  One thing I know I'd
> change -- any search would use the genetic algorithm, not
> brute force!  




More information about the Cmdist mailing list