I agree with Cem Kaner.
Domain Partitioned means you are dividing the input values to test the target function.
- We will take one valid and two invalid values for testing the target function.
e.g. I am having the input value say "a" for testing a given function.
So I divide the input value into three values for testing this function and these would be - ("a", "-a", "+a") or ("a", "Value less than a", "Value greter than a"). this is also termed as boundary value analysis.
Domain partitioning is the most widely practiced software test design method.
We partition (subdivide) a domain (the set of all possible inputs to a function) into sub-domains (subsets, aka equivalence classes) and then choose members of each set to test, as representatives of the set. Typically in a linearizable domain (one whose inputs can be mapped to a number line), we choose boundary values as the representatives of the set.
The best single article on the topic is probably
J. Ostrand and M. J. Balcer. The category-partition method for specifying and generating functional tests. Communications of the ACM,. Communications of the ACM, 31(6):676 -- 686, June 1988
There are extended discussions of the approach in Glen Myers, Art of Software Testing; Kaner et al's Testing Computer Software; Binder's Object-Oriented Testing; Jorgensen's Software Testing: A Craftsman's Approach, and Beizer's Black Box Software Testing.
For the mathematically unchallenged, I think Jorgensen provides the best presentation in print. For gentler introductions, I think Myers and my books are both suitable.