Iteration 3 Tests - mazmaz2k/Minimal-balanced-node-separator GitHub Wiki

Iteration 3

In this iteration, we will have a functional algorithm at our position.We will test our Algorithm against existing Algorithms that find vertex separator in graph like metis, KaHIP.
In addition, we will be working with system that computes all minimal models of a propositional theory -WASP and test our own theory against it, we will check on same examples and see if there is improvement in run time.

Total Estimation

We estimate the work length to be at 4 weeks The software would:

At the end of this iteration, we would have also defined software requirements and also our software design, according Mentors requests.

Retrospective - goals and results:

Process, Methods & Tools

  • After testing with Grid graph issue, Omri had to change algorithm to find node separator issue.Now tests will postpone after construction the new algorithm, and after they will be made on the new algorithm.
  • We used Eclipse as our main developing environment. Each of us written his code on his own local machine while we performed the actual merging, building and compiling on a single machine (Eclipse is a resource hog program).

Rules:

  • Adi will work on Rules data structure creation.
  • Omri will work on graph data structure creation.

Missions finished until now:

  • Graph return set of sources to rules data structure and it find minimal model for set of rules by using davis putnam
  • we are able to find minimal module to small set of clauses .
  • create method that dismantle Strongest Connected component
  • Receive a set of vertexes from graph DS and put values on each vertex- also checks if the values we put in the variables return SAT if so we change rules DS by the values we found and if not we try different values for the variables

Issue Management:

link to issues

Documentation

Manuals:

Iteration Conclusion:

Omri have to change Theory to the "Theory that use K-edge connected component" to "Finding small balanced separators" link. After test with grid like graph failed to find node separator but return a result with that separate one node each time. Now I implement the new theory and preform test on it.
Our team works very well together, we have via meetings on skype an also have a Whatsapp group and Gitter group , where all the members always stay connected and updated. We also work closely with our mentors and as a result we found a bug after test were done on "K-edge connected component",we Created new method to find node separator.

Next iteration Iteration 4

⚠️ **GitHub.com Fallback** ⚠️