tscrim / coxeter Goto Github PK

View Code? Open in Web Editor NEW

11.0 11.0 5.0 424 KB

Coxeter is a library for the study of combinatorial aspects of Coxeter group theory

TeX 4.20% C++ 94.89% Makefile 0.91% Shell 0.01%

coxeter's People

Contributors

Stargazers

Watchers

Forkers

jdemeyer sashakolpakov bollu davidb2 adamnemecek

coxeter's Issues

Separate library and top-level interface into builds

Fix deprecations during the build process

As noted on http://trac.sagemath.org/ticket/19178.

Use std::copy over memcopy

I've seen some posts saying that the former is faster than the latter. Moreover, there are some places where handling the copying directly could be faster (e.g., List::erase).

Crash on specific matrix input

Hi!

Entering the following matrix

yields an Address Boundary Error.
Running ddd on that seems to indicate that the error comes from

(anonymous namespace)::InitMinTable::fillReflectionRow (this=0x7ffff7f46650, G=..., r=5, s=2 '\002') at minroots.cpp:837

I've looked at the code at this place but couldn't easily find the source of the bug.
I'll try to look into it more carefully.

EDIT: (the matrix was input interactively)

EDIT2: After further investigation, it seems that at this line

coxeter/minroots.cpp

Line 835 in 7b5a1f0

y = min(y,s);

, the value for y is undef_minnbr, which I think shouldn't happen since we're then using it as an index.

Implement parabolic KL polynomials

Convert to autotools/configure for creating makefiles

Implement timing test suite

Because speed is the main priority, we should add a test suite to generate timings and tests for speed regressions.

Cleanup namespaces

There are far too many of them and all of the code should be under one unified namespace.

Implement push and pop for List

These will be very fast list operations that we currently do not have. In particular, Bruhat ordering in #10 pops elements off the back and comments have said this is a bottleneck.

Cleanup documentation

We should use a more standard C++ documentation formatting, such as that compatible with Doyxgen.

Add unit tests

Because a good testing framework is good to have. We should consider using a standard C++ unit test framework to do so.

The basic implementation of MinTable::inOrder() uses recursion and copying of the Coxeter element at every step. We can avoid the recursion and the extra copying by making an entry point which makes one copy and then uses an iterative version of the check. We can also do a similar improvement to the one which records the positions.

Consider moving sagemath gitlab org?

Hi @tscrim ,

We've been putting some math packages for which maintenance has been largely taken over by the Sage community (including, similarly to this library, symmetrica: https://gitlab.com/sagemath/symmetrica)

Would you be okay if we set up a repo for this on gitlab and moved all the issues over there? That will make it easier also to have a new release upstream for use in Sage as well as downstream packagers. I could also take over reworking the build system if you want. This is motivated in part by https://trac.sagemath.org/ticket/29150