DESigning Correlated And Reduced Theories for Electronic Structures – DESCARTES

For n-interacting electron systems, numerical diagonalization is limited to Hamiltonians of few tens of electrons on classical computers. Thus a physically motivated and transferable approach to divide this intractable N-interacting particle problem into multiple sub interacting embedded systems that can be efficiently diagonalized is crucially needed as an alternative to the current state-of-the-art methods. In that context, the DESCARTES project aims at developing new embedding methods solving the drawbacks of the recently proposed Density Matrix Embedding Theory (DMET). We propose a variational strategy based on the properties of the Householder unitary transformation to design uniquely-defined embedded cluster. Test case as well as cutting edge problems will be addressed in the context of model Hamiltonians (e.g. Hubbard model). Once the optimal strategy is set, extension to real Quantum Chemistry problem will be performed and compared with results obtained from standard DMET.