1: \begin{abstract}

2: The development of polynomial cost solvers for correlated quantum impurity models, with controllable errors, 

3: is a central challenge in quantum many-body physics, where these models find applications ranging from nano-science to the dynamical mean-field theory (DMFT). Here we describe how 

4:  configuration interaction (CI) approximations to exact diagonalization (ED)

5: may be used as solvers in DMFT. CI approximations retain the main advantages of ED, such as the ability to treat

6: general interactions and off-diagonal hybridizations and to obtain real spectral information, but are of polynomial cost. Furthermore,

7: their errors can be controlled by monitoring the convergence of physical quantities as a function of the CI hierarchy.

8: Using benchmark DMFT applications, such as single-site DMFT of the 1D Hubbard model and  $2\times 2$ cluster DMFT of the 2D Hubbard model,

9: we show that CI approximations allow us to obtain near-exact ED results for a tiny fraction of the cost.

10: This is true over the entire range of interaction strengths including ``difficult'' regimes, such as in the pseudogap phase of the 2D Hubbard model. 

11: We use the ability of CI approximations to treat large numbers of orbitals to 

12:  demonstrate convergence of the bath representation in the $2\times 2$ cluster DMFT using a 24 bath orbital representation.

13: CI approximations thus form a promising route to extend ED to problems

14: that are currently difficult to study using other solvers  such 

15: as continuous-time quantum Monte Carlo, including  impurity models with large numbers of orbitals and general interactions.

16: \end{abstract}

17: