1: \begin{abstract}

2: We present a real-space computational method

3: called treecode-accelerated Green Iteration (TAGI)

4: for all-electron Kohn-Sham Density Functional Theory.  

5: TAGI is based on a reformulation of the Kohn-Sham equations

6: in which the eigenvalue problem in differential form

7: is converted into a fixed-point problem in integral form

8: by convolution with the modified Helmholtz Green's function.

9: In each self-consistent field (SCF) iteration,

10: the fixed-points are computed by Green Iteration,

11: where the discrete convolution sums are efficiently evaluated 

12: by a GPU-accelerated barycentric Lagrange treecode.

13: Other techniques used in TAGI include

14: adaptive mesh refinement, Fej\'er quadrature,

15: singularity subtraction,

16: gradient-free eigenvalue update,

17: and

18: Anderson mixing to accelerate convergence of 

19: the SCF and Green Iterations.

20: Ground state energy computations of several atoms (Li, Be, O) 

21: and 

22: small molecules (H$_2$, CO, C$_6$H$_6$) 

23: demonstrate TAGI's ability to efficiently achieve chemical accuracy.

24: \end{abstract}

25: