[WIP] Combined addition and multiplication for fmpz_mpoly #994
Conversation
multiplication; no optimization for small coefficients or small exponents; no documentation
1. maxfields wasn't computed correctly
2. exponent offsets are multiplied by "N"
3. testing a candidate multiindex to see it it's on an edge
is more complicated than just checking if it's less than 0
… exponents, that were commented out anyway because they haven't been ported
Works on existing single-term test cases, but hasn't been tested multi-term yet.
…ixes on output_function
and instead pass 'output_function' as a function argument
…aces at the ends of lines
time with an index of -1 to indicate end-of-polynomial
…tput_function is set
|
@wbhart could you approve this for CI? |
|
The code looks ok so far, however it fails on Windows (take a look at the appveyor CI). I think this is likely due to the function pointers. I'm not sure how to do those on Windows but I assume there's tutorials out there somewhere. Of course I didn't check this, so it could be something else of course. |
|
How do you envision documentation working? It's a very specific application. To make it a permanent part of Flint, we probably need to document it clearly in its own section so that people know what it is for and how to use it (with examples). |
That sounds good. I've started work on expanding the documentation in MPolyAlgorithms, but a new section in the main documentation sounds like a good idea. |
I just noticed that we've got appveyor running on this PR. I haven't tried to build on Windows at all; I'll take a look at it. |
…d of slong, which makes more sense and helps silence compiler warnings in SAGE
It looks like it just doesn't include the new source files in the library, and doesn't find the functions when linking the test programs. I just added the new source files to the Is there anything else that needs to be done for Windows? |
|
Possibly the CMakeLists file needs updating. Not sure. |
|
Are there any profiling programs with this against |
fmpz_mpoly_addmul_multi and its variants constructs the sum of a list of products of fmpz_mpoly's without storing intermediate results