PyScan: Search-based Tools (in Python)
Author: "There are some who call me... Tim?".
Under Construction
This site will be a "howto" guide on combine SBSE and data mining tools to explore models. Right now, its much less than that. But it will be the toolkit used for the Fall 2015 graduate subject Automated (Model-Based) Software Engineering. Share and enjoy... but not quite yet.
Contents
This site contains numerous links to example models.
As to the rest, it comprimises:
- 10% intro Python examples;
- 20% theory of model-based and search-based SE;
- 20% how to build models (using Python DSLs, using raw programming, uses interfaces to other systems);
- 30% how to search models (multi-objective optimization, plus some data mining);
- 20% how to watch model output (experimental methods to understanding models).
Anything Here that is Special and Different?
One big emphasis in my work is the comparative analysis of different approaches:
- It seems to me that you can't just describe (say) two optimization algorithms-- you also have to provide the machinery that can assess which of these algorithms works best on different problems.
- Hence, much of this code concerns the rig around the algorithms that watches and assesses their performance.
Also, there is stong connection between data mining can optimization:
- While this theoetical connection is widely acknowledged, I am unaware of work takes the next step to to combine/ clarify/ simplify these two approaches.
- My suspicions are that both approaches are really combinations of a lower-level set of primitive operators.
- The goal of this work is to indentify those operators, and offer sample implementations.
- Watch this space!
Before you start
Time to freshen your Python skills:
- Do you understand my intro Python examples?
- Note that some of the techniques in the intro are reused (extensively) in the rest of this code base.