opendf

Official Resources

• Homepage: https://github.com/CQMP/opendf
• Documentation: https://github.com/CQMP/opendf/blob/master/README.md
• Source Repository: https://github.com/CQMP/opendf
• License: GNU General Public License v2.0

Overview

opendf is a condensed matter physics code that solves strongly correlated lattice problems (such as the Hubbard model) in finite dimensions using the dual fermion method. It extends DMFT by including non-local correlations through diagrammatic extensions, providing a systematic way to treat spatial correlations beyond local DMFT approximations.

Scientific domain: Strongly correlated systems, dual fermion method, diagrammatic extensions of DMFT Target user community: Researchers studying non-local correlations in strongly correlated materials

Theoretical Methods

• Dual fermion (DF) method
• Diagrammatic extensions of DMFT
• Non-local correlation effects
• Ladder approximation
• Hubbard model in finite dimensions
• Spatial correlations beyond DMFT
• Vertex function calculations

Capabilities (CRITICAL)

• Dual fermion calculations: Systematic inclusion of non-local correlations.
• Hubbard model solutions: Supports 2D and 3D lattice models.
• Diagrammatic extensions: Includes ladder diagrams and higher-order correlations.
• Data Standard: Utilizes the Open Data Format (ODF) for data exchange.
• Parallelization: MPI parallelization for intensive vertex calculations.

Inputs & Outputs

• Input formats:
  • ODF Packages (.odf.zip): A zipped archive containing:
    • Data (.csv): Primary numerical data.
    • Metadata (.xml): Descriptive metadata using DDI-Codebook schema.
    • Version (.json): Versioning information.
  • Configuration files for solver parameters.
• Output data types:
  • Self-energies (momentum-dependent)
  • Green's functions
  • Vertex functions
  • Observables
  • ODF-compliant output archives.

Interfaces & Ecosystem

• ALPSCore: Built on ALPSCore libraries for Monte Carlo and grid utilities.
• C++: Modern C++ implementation.
• HDF5: Standard data format for internal storage.
• MPI: Parallel execution.

Limitations & Known Constraints

• Computational Cost: Higher than standard DMFT due to vertex function calculations.
• Dependency: Requires ALPSCore libraries, which can be complex to install.
• Scope: Primarily focused on model Hamiltonians (Hubbard) rather than full ab-initio materials (unless interfaced).

Performance Characteristics

• Cost: Significantly higher than standard DMFT due to vertex function calculation ($O(N^4)$ complexity).
• Parallelization: MPI parallelization crucial for diagrammatic summations.
• Scaling: Scales well on large clusters.

Comparison with Other Methods

• vs Standard DMFT: opendf includes non-local spatial correlations via Dual Fermions; standard DMFT is purely local.
• vs GW+DMFT: Dual Fermion is a diagrammatic extension of DMFT; GW+DMFT combines GW results with DMFT.
• Unique strength: Systematic inclusion of non-local correlations using the Dual Fermion method.

Verification & Sources

Primary sources:

1. GitHub repository: https://github.com/CQMP/opendf
2. README and code documentation.
3. Open Data Format specifications.

Verification status: ✅ VERIFIED

• Source code: OPEN (GitHub)
• Method: Dual Fermion implementation
• Data Standard: Adopts ODF