This is the implementation that aims to support my 3rd year dissertation, Machine learning functional prgrams. The implementation is still a prototype, so code is a bit messy and comments are lacking.
Rigth now, the system is capable of finding programs that are part of a restricted program space. There are 3 hard coded metarules (a future version will allow the user to supply them) that describe how the functions should look like: function composition, map and filter (more will be addded). The input for those functions are either user defined background functions or functions that have been created through the synthesis process (functions whose name starts with gen). The language that we induce programs in is a pure language that maps to a subset of Haskell (a description could be found in the file src/Language/Syntax.hs and an usage example can be found in the file test/droplasts.test).
Once a program has been successfully found, it will be pretty-printed in the console.
The only build dependency is The Haskell Platform, which will also install the stack build system, used to run the program.
When in the project directory, running the command stack run input will synthesize a program based on the input file (relative path). For an example of how the input file should look like, look at the file test/droplasts.test.