Stop using INT_MIN/INT_MAX for floats . INT_MIN/INT_MAX are meant for the int type, and thus casting them to a different type (float) means loss of precision which may lead to failed comparisons later on. . Instead use the proper limits using std::numeric_limits.
Use a locally installed mathjax, instead of a remote version diff --git a/docs/doxyfile.in.cmake b/docs/doxyfile.in.cmake index 235484b..9dd8550 100644
Fix the installation of the Python module - move the installation root ($DESTDIR) as --root parameter - specify --single-version-externally-managed, so the module is not installed as single egg file
