Custom Query (545 matches)

Doxygen has bibtex support since release 1.7.5.

It breaks my heart, but we might want to consider replacing our custom tool to Doxygen's built in one.

The min cost flow algorithms implemented in CycleCanceling are rather slow in practice, their only advantage is that two of them is proved to be strongly polynomial if the applied min mean cycle algorithm is also strongly polynomial. Although HowardMmc is clearly faster than HartmannOrlinMmc in practice, its theoretical running time is not polynomial. I think it is more important to ensure that these implementations are indeed strongly polynomial than using Howard's min mean cycle algorithm for efficiency reasons.

Actually, the current state of the library is not consistent. The doc says that two cycle canceling implementations are strongly polynomial and expresses their theoretical running time, while the implementation uses Howard's algorithm, so it is _not_ polynomial. I'm sorry for introducing such an inconsistency.

Therefore, I suggest to use HartmannOrlinMmc in the cycle canceling implementations.

The current installation guide should be updated with respect to the change in LEMON 1.3, i.e. the autotool-based build system becomes deprecated. The current CMake-based part of the guide should also be updated according to the INSTALL file of the distribution (linux/unix systems should also be considered, not only Windows with Visual Studio).