Hubbard model dynamical correlation function via Lanczos Techniques

given by Mario Collura

Term: 0

Start: 11-01-2021 End: 26-02-2021 Room: 131

Credits: 3

Schedule: TBA


This hands-on course has the aim of guiding the students to apply numerical iterative techniques
to face with a standard problem which can be encounter in typical condensed matter setup: Compute dynamical correlation functions.

The approach will be the following:


  1. Consider the Hubbard Hamiltonian with PBC and implement symmetries in
    constructing the Hilbert space of the system.
  2. Implement an efficient procedure to compute the action of the Hamiltonian to a generic state in the
    previously constructed symmetric Hilbert space.
  3. Write an iterative algorithm (e.g. Davidson) to find out the ground state |GS> of the model.
    Working at HALF-FILLING, find out the |GS>
  4. Consider the operator

    c(q)_s = Sum_{j = 1}^{N} \exp(-i q j) c(j)_s

    where “s” is the spin index; and compute the dynamical correlation function (or its Fourier transform)

    < GS | c(0,t)_s c^{\dag}(0,t’)_s | GS >

    with t > t’, where c(0,t)_s = exp(i H t) c(0)_s exp(-i H t)

    via the Green function

    < GS | c(0)_s (z - H)^{-1} c^{\dag}(0)_s | GS >

    using the continued fraction methods. Notice that the number of particle is left unchanged;so the dynamical procedure works in the half-filling sector.

  5. Study the previous points as function of the Hubbard interaction U.
  6. [Optional] Study the previous points as function of the momenta “q”.


Reference for deep details: Rev.Mod.Phys.66.763; Rev.Mod.Phys.68.13;