MECCA can handle large-atoms per cells with inhomogeneous, multisublattice disorder. MECCA is O(N) even for metals using screened k-space KKR. Self-Consistent, total energy or finite-T Grand Potential is handled by Green's Function Contour Integration. Use of Screened-CPA permitted more correctly accounts for charge-correlation effects in metals. (See Publications for original KKR-CPA and screened-KKR-CPA.)

Versions of Code are Downloadable in Summer of 2008: (Software Archive)
Regarding Use: User License
(Help files are contained in the MECCA source code)
A user help page will be established for VERSION 2.

A KKR-CPA code (Version 1)-- from the original self-consistent, green's function KKR with Coherent Potential Approximation with Atomic-Sphere and Muffin-Tin Approximations for addressing chemical disorder. (See Publications for original fast KKR-CPA by contour integration with total energy, Second Broyden's Methods, etc.)

MECCA (Version 2) -- includes Dynamical Cluster Approximation (KKR-DCA/Non-local CPA) for correlated atoms (atomic short-range order).

• O(N) screened k-space KKR-DCA/NLCPA. (2005-present)
  Developers: Dominic Biava, S. Ghosh and D.D. Johnson
  • KKR-DCA-ASA: Uses Dynamical Cluster Approximation (DCA) to handle Atomic Short-Range Order (SRO), i.e. correlated atoms instead of correlated electrons. Also called the Non-Local CPA. (2005)
  • Developed Generalized Lloyd's Formula and Grand Potential/Total Energy for SRO Alloys. (2006)
  • Finite-T Grand Potential Prediction of SRO parameters. (2008)
    Extensions: Aftab Alam and D.D. Johnson
    • Reduced Storage for Potentials. (2007)
    • Starting potentials by overlapping atomic densities. (2008)
    • Large-scale parallelization. (2008)
    • Pseudo-Full-Potential KKR-CPA/DCA. (current)
    • Augmented KKR for reduced and more accurate basis. (current)

• O(N) Multisublattice, Screened k-space KKR-CPA --- for metals, too! (1999)
  Developers: A.V. Smirnov and D.D. Johnson
  • Scalar version for Workstations. (1998)
    in the Atomic Sphere Approximation (ASA) and Muffin-in (MT) Corrected.
    For 10 Energies on the contour integral (fine for Cu) (values are obtained on DEC 500 MHz AU):
    • 1-atom/cell FCC   Cu about     7.5 secs/scf-iteration*
    • 8-atom/cell DO22 Cu about 140.0 secs/scf-iteration*
      (*DO22: 800 secs/scf for more accurate DOS near Real E axis, only needed at convergence.)
  Scalable-parallel version for workstation clusters
  Developers: A.V. Smirnov and D.D. Johnson
  • KKR requires Energy contour integration which trivially parallelizes over workstation cluster per E.
  • Self-consistent, finite-T Grand Potential for multisublattices. (1998)
  • Sparse Matrix Techniques with Iterative Methods for Parallel Matrix Inversion. (1999)
  • 2000 all-electron Cu atoms on DEC 500 Workstation. (1999)

• One-atom/cell KKR-CPA-ASA/MT for Mainframes (1983)
  Developer: D.D. Johnson
  • Utilized contour integration for "Fast KKR method". (1984)
  • First variational and self-consistent KKR-CPA calculations. (1985)
    Old Benchmarks
  • Self-consistent KKR-CPA Finite-T Grand Potential. (1986)