The Paradox of\nGlobal Multireference Diagnostics

Modern computational chemistry methods are a useful tool\nfor modeling\nmany chemical systems, but they are challenged by multireference species\n(e.g., transition metals). A variety of diagnostics have been formulated\nto identify such cases. They are typically developed by analyzing\nmultireference characters of small molecules, and many provide an\naverage picture of the entire system. We caution the use of such diagnostics\nfor large systems because large systems may include parts with varying\ndegrees of multireference characters. Specifically, a small but highly\nmultireference component may yield a large error in absolute terms,\nwhich may be masked in an average value over the entire molecule.\nAs the calculation of molecular relative energies often concerns errors\nin absolute terms, such a false sense of safety may be detrimental.\nA prospective means to tackle this challenge is to use fractional\noccupation density to identify potentially problematic components\nin a system and then examine this moiety with higher-level computations\non appropriately constructed smaller models.