A problem whose scoring criterion involves running the output of a program through a large number of test cases.

Test battery challenges involve running a program through a large number of test cases (usually at least 5 or 6 digits) and measuring how it performs in some respect of its output.

For example, a problem might require you to output the smallest operation tree that results in a number, and the solutions compete on the total size of trees for 100,000 of these given numbers.