Half the “summer” gone by

Hello, before I blog about my SymPy work, its going to be an exciting new semester in college (I hope) , and a set of new interesting courses (or courses with interesting names atleast) to be bunked. Last semester, just flew away,  I have the faintest idea of what happened , and I’m just hoping this time, it wouldn’t be the same. The summer wasn’t any different, and I have no idea how it went by. I had thought maybe I could learn something other than my SymPy project, like revising my C concepts, or maybe basics of ML, but if there is anything that I suck at, it is definitely time management, for some strange reason I prefer to while away the entire mornings and afternoons, listening to music, on Quora, Facebook, and in the evenings when the guilty conscience of having done nothing pricks me, I start to write some SymPy code.  There are of course others things that I suck at, however this post is going to about my SymPy project and not about things that I suck at and don’t.

Well poor jokes apart,  I guess this was the most productive week of my GSoC project (Was it something to do with the mid-term deadline? ). After numerous changes, I finally got my sphinx PR https://github.com/sympy/sympy/pull/2282 and refactor PR https://github.com/sympy/sympy/pull/2286 merged. Other than that I read five new heuristics that Raoul gave me, and pushed it to a single PR https://github.com/sympy/sympy/pull/2308. Since almost of all the algorithms are pretty straightforward, I’d just focus on the one that I found most interesting, and the one that I found well, not so interesting.

1. Linear:

This one assumes $\xi$ to be $ax + bf(x) + c$ and $\eta to be fx + gf(x) + h$, This is similar to the bivariate heuristic, except for the fact that, that $h$ need not be a rational function.  This would mean cases even in which the exponents are symbolic constants would satisfy this heuristic. Substituting, $\xi$ and $\eta$ , the PDE is simplified to, $f + (g - a)h - bh^{2} - (ax + by + c)\frac{\partial h}{\partial x} - (fx + gy + h)\frac{\partial h}{\partial y}$  And as usual. grouping the coefficients and by using solve, one could get the value of the constants, $a, b, c, f, g, h$

2. Abaco2_unique_general:

This algorithm seems like these huge formulae in my mechanical engineering exams (that nobody has any idea how it came into place, in which you just substitute things and get the answer, (I get most of them wrong anyway). This gist would explain the algorithm better than me, https://gist.github.com/Manoj-Kumar-S/6095045

I guess thats it for now. I am waiting for Sean to give his comments, before I can go any further in my project. Cheers.