úÎ ¨ ¶ Needs a rectangular array of  nonnegative weights, which D encode the weights on the edges of a (complete) bipartitate graph.  The indexing should start from (1,1). 1 Returns a minimal matching, and the cost of it. QUnfortunately, GHC is opposing hard the polymorphicity of this function. I think 3 the main reasons for that is that the there is no Unboxed type class, and  thus the contexts IArray UArray e and MArray (STUArray s) e (ST s) do not 6 know about each other. (And I have problems with the forall s part, too).  The same as 'hungarianMethod<Type>'), but uses boxed values (thus works with $ any data type which an instance of ). G The usage of one the unboxed versions is recommended where possible,  for performance reasons.       Munkres-0.1Data.Algorithm.MunkreshungarianMethodInthungarianMethodFloathungarianMethodDoublehungarianMethodBoxedswap complement mergeUnioninsertremove firstJust alternatethawSTthawSTU newSTArray_ newSTUArray_baseGHC.RealRealhungarianMethodSharedmunkers