Kohn-Sham density functional theory works forward: given a potential, compute the density. The inverse problem — given a density, find the potential — is harder and less stable, especially in finite Gaussian basis sets where the potential is poorly constrained.
The paper on Kohn-Sham inversion via density matrix penalization (arXiv: 2603.22140) reformulates the inverse problem as a penalty: the difference between the computed density matrix and the target density matrix, measured in a Löwdin-orthogonalized basis. The penalty energy is invariant under unitary rotations, meaning it doesn't depend on the choice of orbital representation — only on the density matrix itself.
The analytical derivative of this penalty yields a potential correction that drives the self-consistent field toward the target density. For open-shell systems — radicals, transition metals, anything with unpaired electrons — the conventional Zhao-Morrison-Parr approach struggles with convergence. The density matrix penalization method converges robustly across molecular and condensed-phase systems, achieving substantially smaller density deviations.
The through-claim: the instability of Kohn-Sham inversion in finite basis sets is not an inherent difficulty of the inverse problem — it's an artifact of how the constraint is imposed. Penalizing the density in real space leaves too many degrees of freedom unconstrained when the basis is incomplete. Penalizing the density matrix in the orthogonalized basis constrains all the degrees of freedom that the basis can represent, and only those. The right variable for the penalty is not the quantity you want to match but the quantity you can actually control.
2603.22140. Quantum chemistry / Kohn-Sham inversion / density matrix / self-consistent field / inverse problems.