Economical quasi-Newton unitary optimization of electronic orbitals
Samuel A. Slattery,
Kshitijkumar A. Surjuse,
Charles C. Peterson
et al.
Abstract:We present an efficient quasi-Newton orbital solver optimized to reduce the number of gradient evaluations and other computational steps of comparable cost. The solver optimizes orthogonal orbitals by sequences of...
“…Although the near-zero gap displayed for 1FUL in section is typical for zwitterions in vacuum, we have found that for a different zwitterion our local SCF solver is able to converge to the “physically correct” charge-separated solution. In Figure we display the HOMO and LUMO for 1RVS when the QUOTR solver is given the usual extended-Hückel-like initial guess for the orbitals, but without perturbation . We see that there is no unphysical mixing of the occupied orbital on the CO 2 – with the unoccupied orbital on the NH 3 + .…”
Section: Resultsmentioning
confidence: 97%
“… a The standard (RH) SCF solver values are from ref , whereas the quasi-Newton (QUOTR) SCF solver values are from this work. For all cases where the standard solver failed to converge to a solution, the quasi-Newton solver located a solution with a vanishing HOMO–LUMO gap. …”
Section: Resultsmentioning
confidence: 99%
“…This work reexamined the SCF convergence problems for polypeptides in the gas phase in conjunction with modern nonhybrid and hybrid DFAs. While standard SCF solvers typically fail catastrophically, , using a robust quasi-Newton SCF solver we were able to obtain SCF solutions when the conventional solvers fail; in such cases, the KS solutions always had a vanishing HOMO–LUMO gap. Deeper analysis of these solutions using novel natural deformation orbitals obtained from the HF–KS density matrix difference reveals which regions of the system donate and accept electron density in the unphysical KS DFT solution and are thus the culprits in the delocalization error.…”
Section: Discussionmentioning
confidence: 99%
“…HF and KS DFT computations on polypeptides in section 3.1 were performed with a developmental version of the Massively Parallel Quantum Chemistry (MPQC) version 4 program package 48 using the recently developed QUOTR SCF solver. 46 The maximum L-BFGS history size was set to 15 (parameter m), and the initial guess was the unperturbed version of the extended-Huckel-like guess used previously, except for 1RVS, which used the perturbed guess. The regularizer threshold (t r ) was lowered to 0.15, and the history was also reset whenever the RMS gradient crossed 1 × 10 −6 (either crossing below or coming back up again).…”
Section: Technical Detailsmentioning
confidence: 99%
“…In this work, we revisit the convergence problems of SCF for local and semilocal DFAs applied to medium-sized biomolecules in a vacuum. In section we examine solutions for a set of 17 polypeptides considered in ref using a robust local-convergence SCF solver (QUOTR), thus allowing us to characterize the “true” (energy-minimized) solutions even for cases where no solutions could be found in ref ; the SCF convergence issues are correlated to the unphysically small (or even vanishing) HOMO–LUMO gap. Although the incorrect density predicted by approximate functionals has been investigated for zwitterions, , the orbital structure (at least for the semilocal functionals) was not examined in detail.…”
We revisit the problem of unphysical charge density delocalization/fractionalization induced by the self-interaction error of common approximate Kohn−Sham (KS) density functional theory functionals on simulation of small to medium-sized proteins in a vacuum. Aside from producing unphysical electron densities and total energies, the vanishing of the HOMO−LUMO gap associated with the unphysical charge delocalization leads to an unphysical low-energy spectrum and catastrophic failure of most popular solvers for the KS self-consistent field (SCF) problem. We apply a robust quasi-Newton SCF solver [Phys. Chem. Chem. Phys. 2024, 26, 6557] to obtain solutions for some of these difficult cases. The anatomy of the charge delocalization is revealed by the natural deformation orbitals obtained from the density matrix difference between the Hartree−Fock and KS solutions; the charge delocalization not only can occur between charged fragments (such as in zwitterionic polypeptides) but also involves neutral fragments. The vanishing-gap phenomenon and troublesome SCF convergence are both attributed to the unphysical KS Fock operator eigenspectra of molecular fragments (e.g., amino acids or their side chains). Analysis of amino acid pairs suggests that the unphysical charge delocalization can be partially ameliorated by the use of some range-separated hybrid functionals but not by semilocal or standard hybrid functionals. Last, we demonstrate that solutions without the unphysical charge delocalization can be located even for semilocal KS functionals highly prone to such defects, but such solutions have non-Aufbau character and are unstable with respect to mixing of the non-overlapping "frontier" orbitals. Caution should be exercised when unexpectedly small (or vanishing) HOMO−LUMO gaps and atypical SCF convergence patterns (e.g., oscillatory) are observed in KS DFT simulations in any context (bio or otherwise).
“…Although the near-zero gap displayed for 1FUL in section is typical for zwitterions in vacuum, we have found that for a different zwitterion our local SCF solver is able to converge to the “physically correct” charge-separated solution. In Figure we display the HOMO and LUMO for 1RVS when the QUOTR solver is given the usual extended-Hückel-like initial guess for the orbitals, but without perturbation . We see that there is no unphysical mixing of the occupied orbital on the CO 2 – with the unoccupied orbital on the NH 3 + .…”
Section: Resultsmentioning
confidence: 97%
“… a The standard (RH) SCF solver values are from ref , whereas the quasi-Newton (QUOTR) SCF solver values are from this work. For all cases where the standard solver failed to converge to a solution, the quasi-Newton solver located a solution with a vanishing HOMO–LUMO gap. …”
Section: Resultsmentioning
confidence: 99%
“…This work reexamined the SCF convergence problems for polypeptides in the gas phase in conjunction with modern nonhybrid and hybrid DFAs. While standard SCF solvers typically fail catastrophically, , using a robust quasi-Newton SCF solver we were able to obtain SCF solutions when the conventional solvers fail; in such cases, the KS solutions always had a vanishing HOMO–LUMO gap. Deeper analysis of these solutions using novel natural deformation orbitals obtained from the HF–KS density matrix difference reveals which regions of the system donate and accept electron density in the unphysical KS DFT solution and are thus the culprits in the delocalization error.…”
Section: Discussionmentioning
confidence: 99%
“…HF and KS DFT computations on polypeptides in section 3.1 were performed with a developmental version of the Massively Parallel Quantum Chemistry (MPQC) version 4 program package 48 using the recently developed QUOTR SCF solver. 46 The maximum L-BFGS history size was set to 15 (parameter m), and the initial guess was the unperturbed version of the extended-Huckel-like guess used previously, except for 1RVS, which used the perturbed guess. The regularizer threshold (t r ) was lowered to 0.15, and the history was also reset whenever the RMS gradient crossed 1 × 10 −6 (either crossing below or coming back up again).…”
Section: Technical Detailsmentioning
confidence: 99%
“…In this work, we revisit the convergence problems of SCF for local and semilocal DFAs applied to medium-sized biomolecules in a vacuum. In section we examine solutions for a set of 17 polypeptides considered in ref using a robust local-convergence SCF solver (QUOTR), thus allowing us to characterize the “true” (energy-minimized) solutions even for cases where no solutions could be found in ref ; the SCF convergence issues are correlated to the unphysically small (or even vanishing) HOMO–LUMO gap. Although the incorrect density predicted by approximate functionals has been investigated for zwitterions, , the orbital structure (at least for the semilocal functionals) was not examined in detail.…”
We revisit the problem of unphysical charge density delocalization/fractionalization induced by the self-interaction error of common approximate Kohn−Sham (KS) density functional theory functionals on simulation of small to medium-sized proteins in a vacuum. Aside from producing unphysical electron densities and total energies, the vanishing of the HOMO−LUMO gap associated with the unphysical charge delocalization leads to an unphysical low-energy spectrum and catastrophic failure of most popular solvers for the KS self-consistent field (SCF) problem. We apply a robust quasi-Newton SCF solver [Phys. Chem. Chem. Phys. 2024, 26, 6557] to obtain solutions for some of these difficult cases. The anatomy of the charge delocalization is revealed by the natural deformation orbitals obtained from the density matrix difference between the Hartree−Fock and KS solutions; the charge delocalization not only can occur between charged fragments (such as in zwitterionic polypeptides) but also involves neutral fragments. The vanishing-gap phenomenon and troublesome SCF convergence are both attributed to the unphysical KS Fock operator eigenspectra of molecular fragments (e.g., amino acids or their side chains). Analysis of amino acid pairs suggests that the unphysical charge delocalization can be partially ameliorated by the use of some range-separated hybrid functionals but not by semilocal or standard hybrid functionals. Last, we demonstrate that solutions without the unphysical charge delocalization can be located even for semilocal KS functionals highly prone to such defects, but such solutions have non-Aufbau character and are unstable with respect to mixing of the non-overlapping "frontier" orbitals. Caution should be exercised when unexpectedly small (or vanishing) HOMO−LUMO gaps and atypical SCF convergence patterns (e.g., oscillatory) are observed in KS DFT simulations in any context (bio or otherwise).
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.