Automatic structural testing with abstraction refinement and coarsening

White box testing, also referred to as structural testing, can be used to assess the validity of test suites with respect to the implementation. The applicability of white box testing and structural coverage is limited by the difficulty and the cost of inspecting the uncovered code elements to either generate test cases that cover elements not yet executed or to prove the infeasibility of the elements not yet covered. My research targets the problem of increasing code coverage by automatically generating test cases that augment the coverage of the code or proving the infeasibility of uncovered elements, and thus eliminating them from the coverage measure to obtain more realistic values. Although the problem is undecidable in general, the results achieved so far during my PhD indicate that it is possible to extend the test suites and identify many infeasible elements by suitably combining static and dynamic analysis techniques, and that it is possible to manage the combinatorial explosion of execution models by identifying and remove elements of the execution models when not needed anymore.