Refactoring class GeneticAlgorithm

[alguiguilo098]

• create interface Genétic Algorthim

• create testcase method class Genetic Algorthim

• split Genetic Algorthim in more sub-class

[alguiguilo098]

@gnypit Feel free to provide feedback on this issue.

[gnypit]

@alguiguilo098 I'm a mathematician and (partly) a physicist. I'd love to make this repo into a package/library/program available for other researchers to use. However, I'm not at all familiar with how to structure code in terms of interfaces, testcases, etc. I know problems which are well-solved with genetic algorithms and yesterday finally the endurance of the material worked.

In the past I had this code working for the labyrinth as well. Howewver, it was not scalable, and so I started learning OOP. At this point I can see how to generalise and develop:

• genome generators,
• more functions with selection, crossover and even mutation operators,
• various metrics for genetic diversification and variation between rival generations.
  Other evolutionary programming techniques could be added in the future, because they are useful for optimising parameters of simulations of physical systems on classical computers, like Quantum Key Distribution protocols and some experiments in spectroscopy, in combination with evolutionary reinforced learning.

Once labyrinth works well again, I'll want to work again on BB84 simulators, but it would be great to keep developing the genetic algorithm part independently, for general use too. This is where your help could be critical, as in past weeks.

One thing I was considering is providing users with 2 possibilities: to either import pyqkd.genal and working with it in their own code or just providing code for their simulation/problem, e.g. in terms of genome generator and fitness function, and run them with the genetic algorithm in a GUI (high-level approach, I suppose).

Could you elaborate on splitting the Genetic Algorithm into more sub-classes? How could the interface and testcase look like?

[alguiguilo098]

@gnypit I am a computer science student, and there is a set of software metrics called CK metrics, which indicate the quality of object-oriented software. By using an interface, we enforce a set of classes that inherit this interface to maintain the same signature. This is a way to standardize the classes Genetic Algorithm and Parallel Genetic Algorithms, as both exhibit the same behavior. I have a course in my undergraduate program where I am required to contribute to an open-source software project.

[alguiguilo098]

@gnypit A set of test cases allows us to verify if the outputs of a method in a class are correct. I believe that the Genetic Algorithm class has many methods, which, according to the WMC metric, is an indication of a problem. Furthermore, we can divide the Genetic Algorithm class into smaller classes that access this method.

[alguiguilo098]

you like send link ?

[gnypit]

Okay, go ahead! How about we create a branch for this purpose (linked to this issue)? I think it's a great idea to use a metric from the outside 😉

[gnypit]

you like send link ?

Which link? For the joblib?

[alguiguilo098]

about interfaces