1: \begin{abstract}

2: A simple impurity solver is shown to capture impurity-induced superconducting subgap states in quantitative agreement with the numerical renormalization group and quantum Monte-Carlo simulations. The solver is based on the exact diagonalization of a single-impurity Anderson model with discretized superconducting reservoirs including only a small number of effective levels. Their energies and couplings to the impurity $d$-level are chosen so as to best reproduce the Matsubara frequency dependence of the hybridization function.

3: %local BCS Green function. 

4: We provide a number of critical benchmarks and demonstrate the solvers efficiency in combination with the reduced basis method [Phys. Rev. B 107, 144503

5: (2023)] by calculating the phase diagram for an interacting three-terminal junction. 

6: 

7: %A projection onto a ``reduced basis" of select sub-gap state vectors at different parameter points further lowers the effective model's dimensionality. ...

8: 

9: %The solver may be used as-is to produce fast and accurate fits of experimental data for systems of quantum dots coupled to superconducting leads. 

10: 

11: 

12: 

13: %The resulting effective model is solved efficiently using matrix-product-state techniques at both zero and finite temperatures and in all analyzed situations is observed to rapidly converge with increasing number of effective levels.

14: 

15: %It may also be incorporated, as an accelerator, into existing numerically expensive approaches.

16: 

17: \end{abstract}

18: