Sumith : Optimized Awesomeness

GSoC Progress - Week 10 and 11


Hello all. Here are the most recent developments in the Polynomial wrappers.

Report

  • The Polynomial wrappers was using piranha::hash_set as the Polynomial wrappers, hence when there was no Piranha as a dependence, the Polynomial wouldn't compile. The fix to this was to use std::unordered_set with -DWITH_PIRANHA=no so that there would be atleast a slow version available.

  • Another issue was Travis testing of Polynomial. Since we depend on Piranha, we had to setup Travis testing with Piranha included and Polynomial tests run. This was done in the merged PR 585.

  • Before we get the Polynomial merged we have to add mul_poly, improve printing, and test exhaustively. The mul_poly is ready here, will be merged once more tests are prepared.

For mul_poly, previously we never checked the variables corresponding to the hash_sets, which implies you could only multiply a n variable polynomial with another n variable polynomial with the variable symbols same in both. When the variables of two hash_sets are different, a work around would be needed. This would result in slow down if done directly.

As suggested by Ondřej, mul_poly now calls two functions _normalize_mul and _mul_hashest. Here _noramlize_mul sees to it that the hash_sets satisfy the afore mentioned criteria and then _mul_hashset operates
For example, say mul_poly is called, then _normalize_mul converts {1, 2, 3} of x, y, z and {4, 5, 6} of p, q, r to {1, 2, 3, 0, 0, 0} and {0, 0, 0, 4, 5, 6} and _mul_hashset multiplies the two hash_set. The speed of benchmarks determined by _mul_hashset.

  • The printing needs improvement. As of now the polynomial 2*x + 2*y gets printed as 2*y**1*x**0 + 2*y**0*x**1.

  • Not all that was planned could be completed this summers, mostly because of my hectic schedule after the vacations ended and institure began. I am planning to work after the program ends too, when the workload eases. As the final deadline week of GSoC is coming up, I need to ensure at least the PRs on hold gets merged.I am planning to continue after the period ends so as

That's all I have
See ya