Seminarium Fizyki Materii Skondensowanej

Strong electronic correlation underlies the behaviour of solids and molecules of technological interest, such as multiferroics, magnetic materials or enzymatic centres. These are characterized by a huge tuneability: they can change between markedly different electronic states through slight variations in external parameters like temperature and pressure, enhancing their potential for the design of more efficient and sustainable devices. Computational models should play a central role in this design, alleviating the challenge of searching for and trying out different compounds in the lab. However, theoretically describing strongly correlated electrons is an extremely complex endeavour, requiringspecialized methods. While various techniques for accurately capturing strong correlation exist, they are typically associated with a steep computational cost, limiting the possible exploration of material families. For this reason, the development of more approximate formulations, reducing the computational overhead while remaining qualitatively accurate, is an important complementary strategy towards realizing the technological promise of correlated materials. In this talk,I will discuss the basis and application of one such model: the ghost Gutzwiller (gGut) framework. This can be seen as an embedding approach deriving from the traditional Gutzwiller variational Ansatz. gGut captures strong electronic correlation in terms of an effective, non-interacting quasi-particle Hamiltonian. The key here is the introduction of auxiliary states, the eponymous ghosts, which describe correlation in terms of one-body fluctuations, and which are ultimately projected-out when computing observables. Crucially, despite this comparatively simple structure, gGut can still provide qualitatively accurate spectra of correlated models in both low and high energy regimes. I will show how it reproduces some of the key phenomenological ingredients in multi-orbital models relevant for describing iron pnictides or perovskites. Furthermore, I will introduce a new approximation to recover non-local correlation effects, which are crucial to understanding the properties of enzymatic centres and multi-layered materials. I will discuss the scope and limitations of this approximation, and exemplify its quality on examples of bond dissociation in small molecules. The qualitative reliability of the gGut framework and its comparatively modest computational cost make it a promising addition to the theoretical tool-set for material exploration.