MAGIC Magic Mailing List
 
 
From: R. Timothy Edwards (tim AT stravinsky DOT jhuapl.edu)
Date: Wed Feb 27 2002 - 12:38:54 EST

  • Next message: Robert Penny: "Tile splitting question" 

    Just a thought that occurred to me:

   Often what drives magic into a "worst-case scenario" of an overabundance
of tiles is not really the fault of the layout design but is the fault of
magic's tile splitting/merging policy "maximal horizontal strips".  The
instantiation of this policy is restricted to just a few routines in a few
files in the database/ directory.  So an interesting question is whether
there exists a policy that, on average, produces a cleaner tiling of the
planes and tends not to fall into these "worst-case scenario" traps.  A
fractal-like distribution of tiles should be optimal over the largest
ensemble of cases, if there is a consistent split/merge policy that
produces this kind of distribution.

						---Tim

    •  
     
    Questions? Contact Rajit Manohar     		cornell logo