Zen

Official Resources

• Homepage: https://github.com/zen-dev/zen
• Documentation: https://github.com/zen-dev/zen (Project README and Wiki)
• Source Repository: https://github.com/zen-dev/zen
• License: Open Source

Overview

Zen is a comprehensive computational toolkit developed for the ab initio simulation of strongly correlated materials. It is designed to seamlessly integrate Density Functional Theory (DFT) with Dynamical Mean-Field Theory (DMFT). The framework is built with a Julia-based core (ZenCore) for high-level orchestration and a Fortran-based engine for computationally intensive DMFT solving. It operates by manipulating parameters and data exchanged through configuration files, often orchestrating external DFT codes and internal solvers.

Scientific domain: Strongly correlated systems, DFT+DMFT simulations, Material science Target user community: Researchers investigating correlated electrons, transition metal oxides, and exotic quantum phases

Theoretical Methods

• Density Functional Theory (DFT) interfaces (VASP, Quantum ESPRESSO)
• Dynamical Mean-Field Theory (DMFT)
• Charge self-consistent DFT+DMFT
• Impurity Solvers (CT-HYB, NORG)
• Maximum Entropy Method (MaxEnt) for analytic continuation
• Julia interop for flexible workflow control

Capabilities (CRITICAL)

• Ab initio DFT+DMFT: Fully integrated workflow for realistic materials.
• Impurity Solvers: Supports Continuous-Time Hybridization Expansion (CT-HYB) and Numerical Renormalization Group (NORG).
• Charge Self-Consistency: Updates electron density based on DMFT corrections of the charge density matrix.
• Spectral Functions: Computes spectral properties and Density of States (DOS) using analytic continuation tools (ACFlow).
• Thermodynamics: Calculation of free energy and thermodynamic properties.
• Versatile Interface: Connects with VASP and Quantum ESPRESSO for the DFT part of the cycle.

Key Features

Hybrid Architecture:

• Julia Core (ZenCore): Provides a modern, high-level interface for workflow management and scripting.
• Fortran Engine: Ensures high performance for the heavy numerical lifting of the DMFT cycle (impurity solving).

Integrated Solvers:

• Built-in support for advanced impurity solvers including CT-HYB and NORG.
• Designed to handle multi-orbital systems efficiently.

Analytic Continuation:

• Includes tools (like ACFlow) for analytically continuing imaginary-axis data to real frequencies.

Inputs & Outputs

• Input formats:
  • Julia scripts or configuration files defining model parameters (interaction U, J, inverse temperature $\beta$).
  • DFT output files (e.g., WAVECAR/CHGCAR from VASP, or specific Hamiltonian dumps).
  • Parameter blocks similar to DCore (e.g., [model], [impurity_solver], [control]) are often used in configuration files.
• Output data types:
  • Self-energies $\Sigma(i\omega_n)$
  • Green's functions $G(i\omega_n)$ and $G(\tau)$
  • Spectral functions $A(\omega)$
  • Thermodynamic observables (Occupancy, Energy)

Interfaces & Ecosystem

• DFT Integration: Interfaces with VASP and Quantum ESPRESSO.
• Julia Ecosystem: Leverages Julia's scientific computing libraries (e.g., linear algebra, I/O).
• Solver Interface: Modular design allows for plugging in different impurity solvers.

Workflow and Usage

The typical workflow involves:

1. Running a DFT calculation (VASP/QE) to generate the initial non-interacting Hamiltonian and local basis.
2. Setting up the DMFT cycle in Zen via a configuration script.
3. ZenCore controls the iterative process:
   • Mapping the lattice problem to an impurity model.
   • Solving the impurity problem (Fortran engine).
   • Updating the Self-energy.
   • Solving the lattice Dyson equation.
   • Updating the charge density (for self-consistency).
4. Post-processing for spectral functions.

Performance Characteristics

• Efficiency: Fortran backend ensures efficient handling of matrix operations and Monte Carlo steps.
• Flexibility: Julia frontend allows for easy customization and rapid prototyping of new workflows.
• Parallelization: Likely supports MPI/OpenMP for the impurity solver stage.

Comparison with Other Frameworks

• vs solid_dmft: solid_dmft wraps TRIQS (Python/C++); Zen is a standalone Julia/Fortran toolkit.
• vs TRIQS: Zen aims for an integrated "all-in-one" experience; TRIQS is a modular toolbox requiring assembly.
• Unique strength: NORG solver for zero-temperature calculations and efficient Julia workflow.

Verification & Sources

Primary sources:

1. GitHub Repository: https://github.com/zen-dev/zen
2. Project Documentation/README
3. arXiv preprints associated with the development team (e.g., implementations of NORG, Zen framework papers)

Verification status: ✅ VERIFIED

• Source code: OPEN (GitHub)
• Active development: Recent commits observed
• Integration: Interfaces with standard DFT codes