SQL Help!!!

[tektipuser2008]

Using Teradata ordered analytical functions , is it possible to get the result set below. If yes, can someone please provide me with some hints or the SQL itself.
Any help with this is much appreciated.

Sample Data :

Col1 Col2

D1 K1
D2 K1
D3 K2
D3 K4
D4 K1
D5 K2

Result set :

Col1 Col2

D1 K1
D3 K2

Explanation : Since K1 is associated with D1, all other associations of K1 should be dropped, similarly since D3 is associated with K2 all other associations of D3 should be dropped.

 

[j355ga]

select col1, col2 from foo qualify rank() over (partition by col2 order by col1, col2) =1
order by 1