Analyzing the effect and computational cost of famous heuristic graphsearch algorithms

Defining the concept of problem, program, solution. Partial and complete correctness of sequential , alternative, loop, parallel structure.

Our goal is to teach machines to understand, write, and fix code. Some currently running projects: - synthesizing code based on input-output examples (given the inputs and outputs, the system writes code to solve the problem) - synthesizing code based on natural language (generating code to solve programming assignments automatically) - detecting semantic bugs which cannot be detected by a conventional static analyzer - idiomatizing code: detecting and fixing non-idiomatic code fragments - detecting code duplication (code clones), even when the code clones are syntactically very different