1: \begin{abstract}

2:   We present efficient methods for Brillouin zone integration with a non-zero

3:   but possibly very small broadening factor $\eta$, 

4:   focusing on cases in which downfolded Hamiltonians can be evaluated

5:   efficiently using Wannier interpolation.

6:   We describe robust, high-order accurate algorithms automating convergence to a

7:   user-specified error tolerance $\varepsilon$, emphasizing an efficient

8:   computational scaling

9:   with respect to $\eta$. After analyzing

10:   the standard equispaced integration method, applicable in the case

11:   of large broadening, we

12:   describe a simple iterated adaptive integration algorithm effective in the

13:   small $\eta$ regime. Its computational cost

14:   scales as $\OO{\log^3(\eta^{-1})}$ as $\eta \to 0^+$ in three

15:   dimensions, as opposed to $\OO{\eta^{-3}}$ for equispaced

16:   integration. We argue that, by contrast, tree-based

17:   adaptive integration methods scale only as $\OO{\log(\eta^{-1})/\eta^{2}}$ for

18:   typical Brillouin zone integrals. In

19:   addition to its favorable scaling, the iterated adaptive algorithm is straightforward to

20:   implement, particularly for integration on the irreducible Brillouin

21:   zone, for which it avoids the tetrahedral meshes required for tree-based

22:   schemes.

23:   We illustrate the algorithms by calculating the spectral function of

24:   SrVO$_3$ with broadening on the meV scale.

25: \end{abstract}

26: