Local exhaustive testing
Thomas Wood, Keith Miller, Robert E. Noonan · 1992
We introduce local exhaustive testing as a simple strategy for creating test cases that uncover faults (a deficiency in the code that is responsible for incorrect behavior) with a higher probability than tests chosen randomly. To use local exhaustive testing, we identify certain inputs points as "critical," and then test all inputs close to that point. We expect that this strategy will be particularly effective in applications that include an emphasis on geometric or other regular organization. We demonstrate the effectiveness of local exhaustive testing on a collection of programs that are all implementations of a single specification, the proportional navigation problem.