Data structure synthesis

Abstract

All mainstream languages ship with libraries implementing lists, maps, sets, trees, and other common data structures. These libraries are sufficient for some use cases, but other applications need specialized data structures with different operations. For such applications, the standard libraries are not enough. I propose to develop techniques to automatically synthesize data structure implementations from high-level specifications. My initial results on a large class of collection data structures demonstrate that this is possible and lend hope to the prospect of general data structure synthesis. Synthesized implementations can save programmer time and improve correctness while matching the performance of handwritten code.