A black-box, general purpose quadratic self-consistent field code with and without Cholesky Decomposition of the two-electron integrals
Tommaso Nottoli (1), J\"urgen Gauss (2), Filippo Lipparini (1) ((1), Universit\`a di Pisa, Pisa, Italy, (2) Johannes Gutenberg-Universit\"at, Mainz, Mainz, Germany)

TL;DR
This paper introduces a robust, black-box quadratic SCF algorithm applicable to various Hartree-Fock methods, capable of exploiting Cholesky decomposition for larger systems and ensuring high convergence without parameter tuning.
Contribution
The implementation of a general-purpose, quadratically convergent SCF algorithm that is black-box, adaptable to different Hartree-Fock references, and utilizes Cholesky decomposition for efficiency.
Findings
The QCSCF algorithm achieves high convergence without parameter tuning.
Cholesky decomposition enables calculations on larger molecular systems.
The method effectively handles difficult convergence cases in high-accuracy computations.
Abstract
We present the implementation of a quadratically convergent Self-consistent field (QCSCF) algorithm based on an adaptive trust-radius optimization scheme for restricted open-shell Hartree-Fock (ROHF), restricted Hartree-Fock (RHF), and unrestricted Hartree-Fock (UHF) references. The algorithm can exploit Cholesky decomposition (CD) of the two-electron integrals to allow calculations on larger systems. The most important feature of the QCSCF code lies in its black-box nature -- probably the most important quality desired by a generic user. As shown for pilot applications, it does not require one to tune the self-consistent field (SCF) parameters (damping, Pulay's DIIS, and other similar techniques) in difficult-to-converge molecules. Also, it can be used to obtain a very thigh convergence with extended basis set - a situation often needed when computing high-order molecular properties -…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
