Google Summer of Code Blog Posts

GSoC Week 14
19 August 2016

This is the final week of GSoC! I have the write up of the work done on a different page.

GSoC Week 13
12 August 2016

This week I spent a lot of time working on FeatherstonesMethod and its component parts. I started off by moving a bunch of spatial vector functions from another PR I have to the featherstone PR and used some of those functions to calculate the spatial inertia of Body objects. The next thing I worked on was completely rewriting the internals of the joint code. The joints now consist of 4 reference frames and points (one set at each of the bodys mass centers and one set per body at the joint location).

GSoC Week 12
05 August 2016

This week my main work was on Featherstone’s articulated body algorithm. I started by prototyping what I thought his algorith might look like in python code (the algorithm was pulled from chapter 7 of his book). With the passes prototyped it was apparent that I would need a full description of the kinematic tree and so I prototyped the building of the kinematic tree from a single “base” body. I then went on to see what it would look like if the kinematic tree was built during the first pass of his articulated body algorithm and decided that keeping the two separate would result in cleaner code.

GSoC Week 11
29 July 2016

Somehow I think I was off by a week. I think last week’s blog post covers week 9 and 10 and this week’s covers week 11. This week I created a full draft for all components of the SymbolicSystem class that will take the place of a equations of motion generator “base class” that was discussed in my project proposal. I began by creating all of the docstrings for the class followed by the test code. With the documentation and test code written it was a simple matter to finish off the code for the class itself. Lastly I added documentation to two places in sympy, one place contains the autogenerated documentation from the docstrings and the other place I adapted an example from pydy to show how to use the new class.

GSoC Week 8 & 9
22 July 2016

Last week I did not end up writing a blog post and so I am combining that week’s post with this week. Last week I attended the SciPy 2016 conference and was able to meet my mentor, and many other contributers to SymPy, in person. I was also able to help out with the Pydy tutorial. During this time at the conference (and this current week) I was able to flesh out the remaining details on the different portions of the project. I have updated PR #353 to reflect the api decisions for SymbolicSystem (previously eombase.EOM).

GSoC Week 7
08 July 2016

This week the focus was on the support code for the featherstone method. I finished adding examples to the docstrings of every function I made. I then wrote up test code for all of the new functions primarily focusing on expected outputs but included some expected error messages for one of the functions. Lastly I have coded up the functions themselves. This work can be seen in PR #11331.

GSoC Week 6
01 July 2016

The main theme of this week is Featherstone’s method. I have finished reading all of the text book that I feel I need to in order to finish my project. After reading I realize that I have been improper about addressing my project. Instead of saying I am introducing Featherstone’s method to SymPy, it would be more accurate to say that I am introducing one of Featherstone’s methods. The book introduced two equations of motion generation methods for open loop kinematic trees and one method for a closed loop kinematic tree (I stopped reading after chapter 8 and so there may have been even more methods). For my project I have decided to focus on the articulated body method of equation of motion generation for kinematic trees. This method is presented as being more efficient than the composite body method and the closed loop method seems rather complicated.

GSoC Week 5
24 June 2016

Well I started this week off by getting sick and as such productivity took a little bit of a hit. The majority of this week was spent reading Featherstone’s text book. The example documentation showcasing the base class API still hasn’t been reviewed and so that part of the project will just have to be set aside until later. Overall the project will not suffer, however, because of my progress in learning Featherstone’s method.

GSoC Week 4
17 June 2016

I started off this week writing the example code for a pendulum defined by x and y coordinates instead of an angle, theta. This was to show how the eombase.EOM class would handle a differential algebraic system. I also altered the simple pendulum example I made early on in the project to show how it would look as an eombase.EOM example. Of the examples I have made for the base class this one stands out as currently being the only one making use of the equations of motion generators (the other two have the equations of motion entered by hand). While addressing comments on the PR, it was mentioned that a more traditional documentation approach would allow greater visibility of the desired results of the code as the output could be explicitly shown. I agreed and moved all three examples to a single .rst document in PyDy and changed the code to represent the documentation format over example code format. At this point I made a list of all of the attributes and methods I though the base class should represent and made sure they were represented in the example documentation. In addition I included multiple ways I thought error messages should be brought up for incorrect uses of the base class. This information is currently awaiting review.

GSoC Week 3
10 June 2016

Today is Friday the 10th and thus marks the end of the third week of Google Summer of Code. This week started off with continuing work on making test code and a base class for KanesMethod and LagrangesMethod. The work took a turn early in the week when I started working on an example problem that would use the base class instead of working on the base class and test code itself. This resulted in more reading and studying of code examples. This week I also had the opportunity to review multiple shorter PR’s in addition to a longer one that dealt directly with code in KanesMethod.

GSoC Week 2
03 June 2016

This week involved more research and learning than coding and as such the results are less visible. I made some changes to the test code for under mechanics, I spend time looking through the structure and workings of KanesMethod and LagrangesMethod, and created a PR to hold the EOM class development discussion.

GSoC Week 1
27 May 2016

The first week of the Google Summer of Code is now coming to an end and I feel like I’ve hit the ground running and made a great head start. Most of the week revolved around work with creating a way to benchmark KanesMethod and LagrangesMethod classes so that activities aimed at enchancing the speed performance of these classes can be tracked. I also worked on moving some code from the pydy repository to the sympy repository and made my first attempt at reviewing a pull request. Lastly I continued researching Featherstones Method of equation of motion generation and started digging into the structure of KanesMethod and LagrangesMethod as I work towards making a base equations of motion class.

Community Bonding Period
20 May 2016

The community bonding period is coming to a close and so I’d like to write about what I’ve done/learned during this time. I’ve had the opportunity to create my first blog, have my first meeting with my mentors, submit a couple of minor pull requests to pydy and sympy, add an example script to the pydy repository, begin learning about spatial vectors and begin work on some benchmarking code.

GSoC Acceptance
26 April 2016

I am excited to announce that I have been accepted for the the Google Summer of Code program for the summer of 2016. I will be working with the Sympy open source project’s equation of motion generators. For the project I will mainly be focusing on creating a shared base class for the current equation of motion generators and adding an additional generator.