This week we focused on the implementation details of the Vogel’s algorithm. I will get back to this algorithm at a later stage. To start off, last week we had a problem in generating the planar diagram code. But that was resolved using the braid as input. It seems braid word has all the information to generate the planar diagram (unique). The problem with other inputs like the gauss code or the dt code is we get a planar diagram but which component of the over-crossing comes first when we move in the clockwise direction is what is the problem. But in braidword we have this information in the form of -1 and +1, which happens to provide this information. This was achieved at the very beginning of the last week. Ever since it has been the implementation of Vogel’s algorithm, which takes in the oriented gauss code and converts it to the braid word. We have had to work on the oriented gauss code as the parameter to the method, it is not in a great shape but still the main idea has been achieved. (The minor problem is we have to pass the oriented gauss code as a parameter for the method). The difference here is, for every crossing we take how the crossing is oriented, so the name oriented gauss code. So this allows us to convert from oriented gauss code to planar diagram. The next thing was to determine the regions and then the Seifert circles. There are two parts to the algorithm of which one is to identify the unoriented Seifert circles and then perform a Reidmester move. Once this move is done then look for other unoriented Seifert circles. Once we end up with no unoriented circles we get the braidword which forms the second part. So in the first part we are able to convert the oriented gauss code to Planar Diagram, we are working on looking into how we can get regions which are bounded by the components. Next step is to identify the Seifert circles and then detect the unoriented pair. We have used different approaches for detecting the regions, like the Directed graphs where the crossings form the vertices and they have an edge if they have a component or two in common. But this made things a bit complicated as directed graphs cycles (g._show_all_cycles() ) returned more data and it was a bit difficult to get the exact regions. We are working on this as of now. Miguel has sent in his ideas on this matter and I am looking into it. The idea now is to move around the crossings in a given fashion and then work out the regions and then move on. I had another approach which I took time to read, this was the approach of Andrew who has implemented a version of Vogel’s algorithm in the Braid programme. It is somewhat on the similar lines but we choose to re think as it had some huge information to calculate after each step. I have sent in a pull request with the work
Documentation work still remains of the newly implemented methods and we just got these methods working we still need to re think on the design, two to three things that we are lacking as of now:
2. Redesigning the currently implemented methods.
3. Setting the minor issues which have been already identified from the previous pull requests. (in-line comments on the previous pull requests have to be worked upon).