Closed thomas-dresselhaus closed 3 years ago
Please can you give enough detail (bond distance, basis sets, code version) so that I can reproduce the error. The quadrature is adaptive so should only fail to converge if something is going seriously wrong.
Of course, sorry for not providing this information straight away.
The calculation is a standard Hartree--Fock calculation on AuI in the Def2-SVP basis set (as obtained from the widely used basis set exchange). The calculation shows the above error for various bond lengths (2, 3, 4 Angstrom, and also the optimal bond length of 2.405 Angstrom).
The libecpint version used is the most recent master version, i.e. commit fa8df43c0adc7587a4a44b87fc9567a33dcc003d.
This issue is now resolved in the new patch (pr #23). I get no more convergence issues, and an RHF energy that agrees with molpro to the microhartree.
In a calculation on AuI, multiple outputs of an error message "Failed to converge:" appeared, which seem to originate from a grid integration. The program continues to run, but the resulting energy seems to be off by about 1 mHartree.