One Basis Set to Rule Them All: Toward an Affordable Platinum Standard via Natural Orbital Thresholds

We propose a strategy to mitigate the prohibitive cost of coupled cluster methods, including up to quadruple or pentuple excitations, often referred to as the platinum standard, by combining a single sufficiently large basis set with frozen natural orbitals. In our framework, natural orbitals are generated from low-order perturbative or coupled cluster density matrices, truncated according to an occupation threshold, and recanonicalized to yield correlated orbitals fully compatible with standard coupled cluster algorithms. Using occupation thresholds rather than fixed-percentage truncations preserves the size consistency and enables systematic accuracy control. Benchmarks on standard data sets and challenging molecular systems show that large basis sets with aggressively reduced numbers of correlated orbitals consistently outperform smaller basis sets with the same number of correlated orbitals. These results support a unified workflow in which a single basis set is retained for all excitation levels with progressively larger thresholds to reduce the virtual space. Although optimal thresholds are not yet predefined, this framework paves the way toward unsupervised protocols aimed at approaching platinum standard accuracy for unprecedented system sizes at nonprohibitive cost.