HONPAS

Official Resources

• Homepage: https://honpas.ustc.edu.cn/
• Documentation: https://github.com/honpas/honpas/wiki
• Source Repository: https://github.com/honpas/honpas
• License: Academic/Open Source

Overview

HONPAS is an ab initio electronic structure calculation software package designed for linear-scaling first-principles DFT calculations on large-scale systems. Developed at the University of Science and Technology of China (USTC), it uses numerical atomic orbitals (NAOs) and is built upon the SIESTA methodology. HONPAS enables calculations on systems with tens of thousands of atoms using hybrid functionals.

Scientific domain: Large-scale materials, nanostructures, extended defects, hybrid functional calculations
Target user community: Researchers requiring linear-scaling DFT with hybrid functionals for very large periodic systems

Theoretical Methods

• Density Functional Theory (DFT)
• Numerical Atomic Orbitals (NAOs)
• Linear-scaling O(N) algorithms
• Density matrix purification methods
• LDA and GGA exchange-correlation functionals
• Hybrid functionals (PBE0, B3LYP, HSE06)
• NAO2GTO scheme for exact exchange
• Periodic boundary conditions

Capabilities (CRITICAL)

• Ground-state electronic structure
• Linear-scaling DFT calculations
• Tens of thousands of atoms
• Hybrid functional calculations (PBE0, B3LYP, HSE06)
• Density matrix purification algorithms
• Band structure and DOS
• Geometry optimization
• Periodic systems (1D, 2D, 3D)
• Electric field calculations in solids
• Post-SCF linear-scaling methods

Sources: USTC group, Frontiers in Chemistry publications, ResearchGate

Key Strengths

Linear-Scaling Algorithms:

• Trace-preserving canonical purification
• Trace-correcting purification (TC2)
• Trace resetting density matrix purification
• O(N) computational scaling
• Parallel implementation

Hybrid Functionals at Scale:

• PBE0 hybrid functional
• B3LYP hybrid functional
• HSE06 screened hybrid
• NAO2GTO scheme for ERI
• Low-rank HFX approximation

NAO Basis Sets:

• SIESTA-compatible basis
• Compact localized orbitals
• Systematic quality control
• Efficient matrix sparsity

High-Performance Computing:

• Massively parallel execution
• MPI parallelization
• Supercomputer-ready
• Excellent parallel efficiency

Inputs & Outputs

• Input formats:

  • SIESTA-compatible input files
  • Structure files
  • Pseudopotential files (Troullier-Martins)
  • Basis set definitions

• Output data types:

  • Total energies
  • Density matrices
  • DOS and PDOS
  • Band structure
  • Forces and stresses

Interfaces & Ecosystem

• SIESTA compatibility:

  • Based on SIESTA methodology
  • Similar input format
  • NAO basis technology

• Integration:

  • DeepH machine learning Hamiltonians
  • Post-processing tools
  • Visualization compatibility

Advanced Features

Density Matrix Purification:

• Multiple purification algorithms
• Trace-preserving methods
• Trace-correcting methods
• Trace resetting methods
• Optimal algorithm selection

NAO2GTO Scheme:

• Gaussian fitting of NAOs
• Efficient ERI evaluation
• Exact exchange calculations
• Hybrid functional enabling

Electric Field Treatment:

• Berry phase approach
• Modern theory of polarization
• Electric field in periodic systems
• Ferroelectric studies

Linear-Scaling Post-SCF:

• O(N) property calculations
• Efficient post-processing
• Large-scale analysis

Performance Characteristics

• Speed: O(N) linear scaling achieved
• Accuracy: Comparable to standard DFT
• System size: Tens of thousands of atoms
• Memory: Scales linearly with size
• Parallelization: Excellent MPI scaling

Computational Cost

• Linear scaling: True O(N) for large systems
• Hybrid DFT: Enabled via NAO2GTO
• Break-even: ~1000-2000 atoms vs cubic codes
• Target systems: 10,000+ atoms feasible

Limitations & Known Constraints

• Small systems: Overhead for systems < 1000 atoms
• Metallic systems: Linear scaling less efficient
• Documentation: Primarily academic
• User base: Specialized community
• Learning curve: Familiarity with SIESTA helpful
• Availability: Academic distribution

Comparison with Other Codes

• vs SIESTA: HONPAS adds linear-scaling hybrid functionals
• vs CONQUEST: Both O(N), different hybrid implementations
• vs ONETEP: Both O(N), HONPAS NAO vs ONETEP NGWF
• vs OpenMX: Both NAO-based, HONPAS focuses on hybrids
• Unique strength: Large-scale hybrid functional DFT

Application Areas

Extended Defects:

• Grain boundaries
• Dislocations
• Stacking faults
• Interface reconstructions

Nanostructures:

• Large nanoparticles
• Nanowires
• 2D materials with defects
• Heterostructures

Biomolecular Systems:

• Large proteins
• DNA segments
• Drug-receptor complexes
• Enzyme active sites

Disordered Materials:

• Amorphous semiconductors
• Random alloys
• Disordered interfaces
• Glass structures

Best Practices

Linear-Scaling Setup:

• Ensure localized density matrix
• Use appropriate purification method
• Set localization radius correctly
• Monitor sparsity

Hybrid Functionals:

• Use NAO2GTO with care
• Balance accuracy and cost
• Test on smaller systems first
• Compare with standard GGA

Large-Scale Calculations:

• Start with smaller test systems
• Optimize parallel settings
• Use restart capabilities
• Monitor convergence

Community and Support

• USTC Electronic Structure Group
• GitHub repository
• Published methodology papers
• DOE and NSFC support
• Chinese computational chemistry community

Verification & Sources

Primary sources:

1. GitHub: https://github.com/honpas/honpas
2. USTC group page: https://honpas.ustc.edu.cn/
3. Frontiers in Chemistry: Linear-scaling hybrid DFT publications

Secondary sources:

1. ResearchGate publications
2. Chinese Academy of Sciences collaborations
3. Computational materials science journals

Confidence: VERIFIED - Active GitHub, recent publications

Verification status: ✅ VERIFIED

• Source code: OPEN (GitHub)
• Academic use: USTC and collaborators
• Documentation: Wiki and papers
• Active development: Recent commits
• Specialty: Linear-scaling hybrid DFT with NAO basis