With the increasing trend of parallel programming, techniques to analyze and optimize parallel programs are gaining vast importance. Compilers perform a variety of optimizations and analyses in each phase to produce a semantically equivalent representation of the source code. These are often interdependent, and the order of performing them varies the final result. Finding the order among the analyses which gives the most precise result is challenging. Moreover, the optimal order is often application dependent. In some cases where the analyses are mutually dependent, no single order or repetitive application can result in optimal results. Such a dependency is called phase ordering.

May-Happen-in-Parallel(MHP) analysis and points-to analysis are two highly popular analyses with broad applications. MHP analysis is a static analysis that helps identify the statements in a parallel program that may run in parallel. Points-to analysis is a static analysis that identifies which variables may point to which memory locationsThey are interdependent. We show that they exhibit phase ordering and propose a combined MHP and points-to analysis for Java programs.