Closed jeffhammond closed 7 years ago
In Molecule::initSymmetry
, group
never gets initialized. If I add group = &PointGroup::C1();
around https://github.com/devinamatthews/aquarius/blob/master/src/input/molecule.cxx#L334`, then the code runs and shows the molecule is C1.
I will keep debugging, but it seems that the symmetry detection logic is incomplete. At the very least, there should be an error check around https://github.com/devinamatthews/aquarius/blob/master/src/input/molecule.cxx#L1174 that prevents a segfault if group
is uninitialized.
The culprit is this little gem:
399 /*
400 * Spherical rotors: Td, Oh, Ih
401 */
-> 402 assert(0);
403 }
I'll work on actually implementing that.
Are expecting to run in full Ih symmetry or is D2h enough?
It will make it a lot easier to beat NWChem if I can run with Ih but I'll certainly settle for D2h, since that is all any other CC code supports.
Side note: I've never really understood why non-Abelian groups were hard in CC. I get that one needs to generalize a bit, but it's not that complicate. Am I missing something?
It's not too hard in C++ when you make the irrep an object, but I imagine it is a PITA in Fortran. In CFOUR we can't use non-Abelian groups because we use DPD.
On 12/14/16 2:28 PM, Jeff Hammond wrote:
It will make it a lot easier to beat NWChem if I can run with Ih but I'll certainly settle for D2h, since that is all any other CC code supports.
Side note: I've never really understood why non-Abelian groups were hard in CC. I get that one needs to generalize a bit, but it's not that complicate. Am I missing something?
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/devinamatthews/aquarius/issues/19#issuecomment-267147171, or mute the thread https://github.com/notifications/unsubscribe-auth/AFAMocuM9sAY5DHgm8KmWrXK4wH1-Bhfks5rIFF1gaJpZM4LL8aF.
Fixed by 7bbf94f in master.
GDB Backtrace
Input File