The theoretical description of strongly correlated materials relies on the ability to simultaneously capture, on equal footing, the different competing energy scales. Unfortunately, existing approaches are either typically extremely computationally demanding, making systematic screenings of correlated materials challenging or are limited to a subset of observables of interest. The recently developed ghost Gutzwiller ansatz (gGut) has shown great promise to remedy this dichotomy. It is based on a self-consistency condition around the comparatively simple static one-particle reduced density matrix, yet has been shown to provide accurate static and dynamical observables in one-band systems. In this work, we investigate its potential role in the modeling of correlated materials, by applying it to several multiorbital lattice models. Our results confirm the accuracy at lower computational cost of the gGut, and show promise for its application to materials research.
