This is the permanent web page for BAGEL (Brilliantly Advanced General Electronic-structure Library) and related programs. The repositories are currently hosted at github. See also related publications and developers.
BAGEL, Brilliantly Advanced General Electronic-structure Library.
SMITH3, an automated code generator for spin-free multireference theories.
ZQUATEV, code to diagonalize large quaternionic matrices to yield symmetry-adapted eigenvectors.
Jae Woo has recently implemented the code for computing derivative couplings using XMS-CASPT2. He has also interfaced the program to surface-hopping dynamics packages (e.g., Newton-X) and to molecular mechanics packages (e.g., Gromacs). We will shortly report these features!
We have generalized the second CASSCF implementation to relativistic CASSCF! So far it is only with the Dirac-Coulomb Hamiltonian, but it will be extended to the Gaunt and Breit interaction, too. I expect that the code is still somewhat buggy, so let us know if you hit some strange behavior.
We have just implemented the second-order algorithm for CASSCF orbital optimization! Special thanks to Takeshi Yanai.
We have recently implemented nuclear gradients for MS- and XMS-CASPT2 with full internal contraction (i.e., Molcas-type CASPT2). Vertical shifts can be used. We will shortly report the details!
We have recently implemented the extended multistate CASPT2 (XMS-CASPT2) with full internal contraction in the BAGEL package. In addition, we have parallelized the SMITH-based code (intra- and inter-node hybrid parallelization). The code for the nuclear energy gradients is currently under development.
Today BAGEL reached 6000 commits!
We have finished implementing relativistic CASPT2, NEVPT2, and ic-MRCI. NEVPT2 is already parallelized and ready for practical use. CASPT2 and ic-MRCI needs a bit more work. See details in on arXiv.
We have just made an extension for internally contracted MRCI so that it could be used with multiple electronic states!
We have implemented the multi-reference configuration interaction method using SMITH3. Using a commutator trick, our implementation requires only up to 4RDM. Multi-state and relativistic extensions are on the way.
We have recently updated our DF-MP2 gradient code, which is now very efficient in parallel environments. In addition, we have created an option so that the user can compute nuclear forces alone.
1-10 of 24