hamilflow

Community-maintained utilities for DeepH workflows (not an official DeepH package).

Core API

• hamilflow.run_projection
• hamilflow.ProjectionConfig
• hamilflow.RemovalPlan / hamilflow.RemovalRule

The projection accepts removal plans as:

• JSON file path
• Python dict/list payload
• Prebuilt RemovalPlan instance

Outputs always include written Hamiltonian and overlap files and return a serializable result object with paths and metadata.

DFT Workflows (dft_workflows)

The hamilflow.dft_workflows subpackage provides workflow helpers for FHI-aims data generation and collection.

• Main entry point: hamilflow.dft_workflows.GenerateAimsDFTData
• Supports two modes:
  • Run-and-collect: generate new AIMS runs from structure folders and collect outputs.
  • Collect-only: collect existing AIMS run directories into a single organized root.
• K-point handling can be provided either as a plain kgrid or via hamilflow.dft_workflows.get_ksampling-style sampling settings that are resolved into k_grid for AIMS inputs.
• Optional conversion: provide aims_to_deeph_config to append an AIMS-to-DeepH conversion step after collection.
• Projection-only flow: hamilflow.dft_workflows.GenerateProjectedDeephInputs runs one projection job per DeepH subdirectory.
• End-to-end wrapper: hamilflow.dft_workflows.GenerateAimsToProjectedDeephData chains DFT/collection/conversion with projection.

Collected run folders preserve structure-oriented naming to keep downstream mapping explicit.

Band Structure Utilities (band_structures)

Band-related utilities are grouped under hamilflow.band_structures:

• hamilflow.band_structures.band_calculation: build k-path configurations, load Hamiltonians, and plot bands.
• hamilflow.band_structures.band_analysis: analyze computed bands (gaps, shifts, comparisons, and k-point corrections).

Examples

The examples/ folder contains runnable scripts for common usage patterns:

• examples/projection/: projection workflows, including single-task, batch, and SLURM submission examples.
• examples/band_structure/: band-structure workflows, including local runs and SLURM-oriented scripts.

These examples are the fastest way to see expected inputs, output layout, and end-to-end execution patterns.