gastd / ga-hls Goto Github PK

Grammar-based fuzzing is about using a grammar (set of rules) to guide the mutation of a formula.

What if we try to use something similar when mutating the individuals?
In this way we may guarantee that the mutations will not result in syntactically wrong formulations.
After all, we already have the HLS grammar.
What would be the effort to implement the "fuzzer"? Are there any off-the-shelf artifact tha twe can use to implemetn the fuzzer?

Resources that may help with the investigation (Andreas Zeller):

• https://www.fuzzingbook.org/
• https://mutation-workshop.github.io/2019/Grammars_for_Mutation_Testing.pdf

The crossover should split the chromosome into two parts. The parts are merged with other parent parts to create two new offspring. Our individual is a tree, so we randomly choose a node and split the tree into this node. The same is done for the second parent. We create the offspring by merging the tree with the subtree, creating a new individual.

To mutate an individual, we randomly choose a node to mutate and randomly choose the new operator using a list of operators such as '[ForAll, And, Or, Implies, Exists]'.

Come up with a helpful fitness function to guide the GA's evolution:
Our goals:

• A runnable formula