I am Sarrah Bastawala, a second year undergraduate engineering student pursuing Bachelors in Information Technology in Veermata Jijabai Technological Institute.
This is a companion discussion topic for the original entry at https://sarrahbasta.home.blog/2022/06/10/introduction-of-the-project/
Thanks for the post. The GSoC coding period is starting in a couple days, so I suggest that the next steps is to made a full work plan including goals for the midterm and final
evaluations and have your mentors approve it. That would help you start on the right track.
Thank you! I do have the original work plan and timeline I outlined in my proposal, would a discussion over that by my mentors @cdf be possible or should I revise it on my own?
looking at that timeline, there were quite a few things you had listed in the Community Bonding period leading up to Monday’s start of the Coding period. I guess it might make sense to look at that, update based on where things actually stand, and review with @cdf, et al.
Yes, while I had listed quite a few things to be done in the Community bonding period, unfortunately I wasn’t able to fulfil them. On account of my exams being extended and me spending some more time understanding Mercurial as I had originally made some mistakes, I did want to enquire if for the rest of the project I would be working on the git mirror or sending changes via Mercurial ? Also, although I had written about some test cases, I would require some guidance in which direction to research as I don’t have much experience with such mathematical problems. Does the original timeline and scope of project seem okay to you, or would some changes be required before further work ? @nrjank @cdf I am currently re-reading the research paper provided to me to improve the decic function.
Hello Mentors @cdf @nrjank @siko1056 ,
As the coding period started the previous week, I have tried to adhere to the original timeline in my proposal and have been working on understanding the research paper for the first phase of my project (i.e improving the decic function)
I have written a small pseudocode outlining my understanding of the cic function currently used in Matlab with the trust region method. I think the next step in it would be to understand which predefined functions from Octave I can use to implement the algorithm (such as for QR decomposition, etc) and to get the pseudocode I have come up with checked by a mentor with the mathematical knowledge for it (which I unfortunately do admit I lack )
Also, before that, while researching on this in the last week, I came across this mail thread from 2019 with @cdf that outlines a plan to use the IDACalcIC function from Sundials to get the consistent initial conditions required. I also found that Sundials now uses IDACalcIC to correct values and has a function IDAGetConsistentIC to calculate initial conditions. While as outlined in my proposal, I was going to use this function as a benchmark to test decic.m, I would like to ask as to what the conclusion of this mail thread was and if implementing the function directly would in any way be better or should be considered first?
I also had a small doubt regarding whether the code contributions I would make would be via Mercurial in the form of patches or on the git mirror as I am a bit more comfortable with Github workflow…
@sarrah-basta did you have any conversation with @cdf since the GSoC start yet?
No, unfortunately not. The last communication I received from @cdf was when I reached out on the octave-maintainers mailing list before the GSoC Proposal submission deadline and @cdf left a comment after having a look at my this bug here. As was expected of me, I have left messages on the main discourse topic created for the project and also written the blog as advised by you. I have tried starting the project myself as advised by @nrjank to follow the set timeline, following the update above, till we have a discussion with @cdf for approval, should I start with the second phase of my project, namely trying to use Octave’s own matrix/vector classes for the solvers? Would some guidance from you and @nrjank be possible with respect to Octave’s pre-existing classes, coding environment expectations and gnu autotools in that regard? Also, should I try mailing the maintainer’s thread once again if that may be likely to catch @cdf 's attention?
Hi, Sorry for being absent lately, real life issues have been preventing me from getting online often for the last month.
It seems we had cross-posted, as I had sent you two messages via the bug tracker and on discourse, few minutes before seeing this post.
can you reply there so we may get going with the work?