This repository is made to continue research project of Q4 2025.

The goal of this research is to conduct a comparison of compresson/refactoring method versus aulile/library learning method

There are three kinds of experiments conducted:

single problem (aulile with compression)

• modify aulile to take N best programs vs
• use the N best programs for compression and add that to the grammar
  • with different timing for compression

dream coder

work with N synth problems for a specific grammar. Pause on each on them to find the best program (according to Aulile)

• add those programs completely (aulile like) vs
• run compression on them

compare aulile aux. functions

• how close it is to solving all examples vs
• the number of examples solved completely vs
• combined with priority to the number of solved examples