Rcpp version of mpoly and mpoly arithmetic

[Blaza]

No description provided.

[dkahle]

Any idea why ^.mpoly() is so much slower than previously?

[Blaza]

Haven't really done the benchmark or profiling, but when I look at it, it seems that the problem may be that the original R version would just do *.mpoly in a loop, which implies doing mpoly for every multiplication, thus simplifying the polynomial for the next loop iteration, whereas the Cpp version just does the Cpp multiplication in a loop without simplifying the polynomial and the complexity of the mpoly list object increases substantially at each iteration. And it may be the case that the overhead of this complexity beats the overhead of doing the R loop and mpoly each time, as each operation in the loop is pretty fast.

Those are just my thoughts, as the only difference I can see between the implementations is the lack of mpoly after each multiplication in the Cpp version.

Do you have any concrete benchmarks or profiling results for this?

Perhaps sticking with the R version of ^.mpoly is better if the looping doesn't add too much overhead. I don't have the time right now to do a good benchmark of that.

The multiplication *.mpoly works faster than the original version though, right?

[Blaza]

That really turned out to be the issue, I added poly reducing now in every loop iteration and it works fast now.