Known Issues With AIMAll

Known Issues With AIMAll (Version 10.07.25)

Some Limitations

AIMAll is currently only available for Windows 2000/XP/Vista/Windows7 (32-bit or 64-bit). Ports to Linux and MacOSX are being worked on.
Wavefunctions are limited to S, P, D, F, G and H Gaussian basis function types. Support for higher angular momentum Gaussian basis functions and Slater basis functions is planned.
Non-Nuclear Attractors are not handled in a fully automatic manner. They need to be manually added to the wfnfile or wfxfile as atoms with nuclear charge = 0.0 and names beginning with NNA, e.g. NNA13. Fully automated support for non-nuclear attractors is planned.
Wavefunctions derived from effective core potential (ECP) calculations or semi-empirical calculations are currently not supported. Support for wavefunctions derived from ECP calculations is planned.
Conflict catastrophes and other catastrophes in the electron density topology are not fully supported yet. Conflict catastrophes are the most important since they can be non-isolated. Conflict catastrophes can occur for certain symmetric structures, usually involving a weak BCP between an atom and another BCP. The default "Proaim" atomic integration algorithm will not work for atoms connected by a BCP whose interatomic surface contains another BCP, as occurs in a typical conflict catastrophe. The "Promega" atomic integration algorithm should be used for such atoms. Support for conflict catastrophes using the "Proaim" atomic integration algorithm is being worked on.
The calculation of some IQA (Interacting Quantum Atoms, A. Martin Pendas et al) energy contributions are not yet implemented. In particular, diatomic two-electron interaction energy contributions Vee(A,B) are currently not calculable by AIMAll. Intratomic two-electron interaction energy contributions Vee(A,A) are calculable as are atom | whole-molecule two-electron interaction energy contributions Vee(A,Mol) and therefore the atom | rest-of-molecule two-electron interaction energy contributions Vee(A,A') = Vee(A,Mol) - Vee(A,A). The current implementation of Vee(A,A') calculations is, however, extremely expensive and it is therefore not currently advisable to do Vee(A,A) and Vee(A,A') calculations except for small molecules with a small number of MOs. A more efficient implementation for Vee(A,A) calculations is being worked on and the efficient calculation of Vee(A,B) is being implemented.
For post-Hartree-Fock (e.g., MP2, CCSD, etc.), natural orbital AIM "wavefunction" files the Muller approximation of the two-electron density matrix (2EDM) in terms of natural orbitals of the one-electron density matrix (1EDM) is used to calculate 2EDM-dependent atomic properties, i.e., the Vee(A,Mol), Vee(A,A) and Vee(A,A') contributions and the electronic localization and delocalization properties such as LI(A), DI(A,B), D2(A,B), etc. This is necessary because the 2EDM is not available in AIM wavefunction files. Better approximations of the 2EDM in terms of the 1EDM are being considered, but ultimately the availability of the actual post-Hartreee-Fock 2EDM in AIM wavefunction files is most desirable.

Some Other Issues and Some Tips

Some versions of g03 may produce traditional AIM .wfn files with an inconsistency between the nuclear coordinate orientation and the orientation that the MO coefficients correspond to, i.e., the nuclear coordinates are written in the input orientation while the written MO coefficients correspond to the Gaussian standard orientation (unless the Gaussian keyword NOSYMM is specified). This issue is usually caught by AIMQB, which performs a normalization check on traditional AIM .wfn files. To avoid this and other issues with traditional AIM .wfn files, it is strongly recommended to open Gaussian formatted checkpoint files (.fch or .fchk files) with AIMQB instead and then use the corresponding .wfx files for all AIMAll calculations. Opening formatted checkpoint files with AIMQB and then using the .wfx files with the AIMAll programs is beneficial in several other ways as well.
For g03 post-SCF, correlated calculations (e.g., MP2, CCSD, etc.) one must use the Gaussian keywords DENSITY=CURRENT (in addition to OUTPUT=WFN) in order for the necessary "natural orbitals" of the correlated first-order total density matrix to be written to the traditional AIM .wfn file. Use of DENSITY=CURRENT is also necessary for single-point, post-SCF correlated calculations in order for the correlated first-order density matrix to be written to the checkpoint file and thus the corresponding formatted checkpoint file potentially used by AIMQB.
For g03 unrestricted, post-SCF, correlated calculations (e.g., UMP2, UCCSD, etc.) one must use the Gaussian keywords POP=NOAB (in addition to DENSITY=CURRENT and OUTPUT=WFN) in order for the necessary Alpha and Beta "natural spin orbitals" of the correlated Alpha and Beta first-order density matrices to be written separately to the traditional AIM .wfn file. POP=NOAB is not necessary if one intends to open the formatted checkpoint file with AIMQB and then use the .wfx file for all AIMAll calculations.
For g03 post-SCF, correlated calculations the SCF energy and SCF virial ratio (-V/T) are always written to the traditional AIM .wfn file instead of the necessary total correlated energy and correlated virial ratio. To correct the traditional AIM .wfn file written by g03, one must manually change the energy and virial ratio at the bottom of the file to the correlated values. The correlated virial ratio can be calculated as -(E/T)+1, where E is the total correlated energy and T is the correlated electronic kinetic energy (search for KE= in the correlated population analysis section of the Gaussian log file). This issue can be avoided by opening the formatted checkpoint file with AIMQB and then using the .wfx file for all AIMAll calculations.
When doing single point calculations with g03 for which a subsequent AIMAll analysis of the corresponding fchk file is planned, it is recommended to use the FORCE keyword so that the energy gradients with respect to nuclear coordinates appear in the fchk file and hence the wfx file.