This python program uses genetic algorithm to evolve an instruction set for a virtual machine that executes them and creates a set of movement instructions for a guy that collects treasures on a map. Get it? Okay, again.

We use genetic algorithm to evolve a set of instructions – 64 bytes. 1 instruction is one byte. The first 2 bits determine the type of instruction:

• 0 = increment
• 1 = decrement
• 2 = jump
• 3 = print

The remaining 6 bits point to an instruction in the set that is to be affected. This means that the program is self-modifying.

A virtual machine takes these instructions, evaluates them, and generates a set of movements – up, right, down, left. This movements are generated using the print instruction.

Then there's a map with treasures and a greedy guy that wants to collect them. He takes the generated set of movements and moves around the level. He collects a treasure when they are on the same tile. The number of treasures collected is the fitness function for the evolution algorithm.