Fireball

Official Resources

• Homepage: https://github.com/FIREBALL2020
• Documentation: https://thunder-dft.github.io/
• Source Repository: https://github.com/FIREBALL2020/thunder-master
• License: GPLv3

Overview

Fireball is an efficient ab initio tight-binding Density Functional Theory (DFT) code designed for molecular dynamics simulations of large systems. It utilizes a local-orbital formulation of DFT, enabling the simulation of supercells containing thousands of atoms with linear-scaling O(N) computational cost.

Scientific domain: Materials science, surface science, biomolecules, nanostructures Target user community: Researchers performing large-scale MD simulations and studying extended systems

Theoretical Methods

• Density Functional Theory (DFT)
• Tight-Binding (TB) formalism
• Local-orbital basis sets (Numerical Atomic Orbitals)
• Sankey-Niklewski (SN) approach
• Pseudopotentials (Norm-conserving)
• Molecular Dynamics (MD)
• Lewis-structure initialization

Capabilities (CRITICAL)

• Large-scale simulations: Efficiently handles thousands of atoms (up to 10,000+).
• Linear scaling: O(N) cost for total energy and forces, avoiding cubic scaling of standard DFT.
• Basis sets: Optimized numerical atomic orbitals (NAOs), "Fireballs", which are strictly localized.
• Dynamical reconstruction: Suitable for surface catalytic processes and reconstruction.
• Charge transport: Investigations in amorphous systems (e.g., DNA) and molecular junctions.
• QM/MM: Interface with AMBER for hybrid quantum/classical simulations of biomolecules.
• Van der Waals corrections: Implementation available in newer versions (Fireball2020).
• Electronic structure: Band structures, Density of States (DOS).
• Transport: Conductance and STM image simulations.

Sources: Official GitHub (https://github.com/FIREBALL2020), cited in 4 sources.

Key Strengths

Efficiency via Local Orbitals:

• Fireball Orbitals: Numerical atomic orbitals with strict cutoff radii.
• Pre-computed Integrals: Three-center integrals are pre-calculated and stored, speeding up runtime.
• Sparse Hamiltonian: Local nature leads to sparse matrices, enabling O(N) scaling.

Large-Scale MD:

• Mesoscale Systems: Designed for systems too large for plane-wave DFT but requiring quantum accuracy.
• Long Timescales: Efficiency allows for longer molecular dynamics trajectories.
• Dynamical Evolution: Ideal for studying kinetic processes and surface restructuring.

Hybrid Applications (QM/MM):

• Biomolecular Focus: Strong integration with AMBER for enzymatic reactions and DNA studies.
• Active Site Treatment: Quantum treatment of active sites with classical environment.

Inputs & Outputs

• Input formats:

  • fireball.in: Main control file (SCF settings, time steps, temperature).
  • structure.inp: Coordinate file.
  • Basis set files (pre-generated).
  • Pseudopotential files (Fdata).

• Output data types:

  • param.dat: Output parameters.
  • dynamics.dat: MD trajectory.
  • Energies and forces logs.
  • Electronic structure data (eigenvalues, DOS).
  • STM images (if requested).

Interfaces & Ecosystem

• QM/MM Integration:

  • AMBER: Direct interface for hybrid calculations.

• Tools:

  • Lightning: Fast visualization tool/pre-processor.
  • ASE (Atomic Simulation Environment): Scripting and workflow control.
  • FireballTG: Fork with enhanced transport capabilities (Transiesta-like).

Workflow and Usage

Typical MD Workflow:

1. Preparation: Generate initial structure, select basis set and pseudopotentials.
2. Setup: Configure fireball.in for iensemble (NVE/NVT) and dt (timestep).
3. Execution: Run Fireball (parallel execution supported).
4. Analysis: Extract dynamics.dat for visualization and structural analysis.

Transport Workflow:

1. Lead Definition: Define semi-infinite leads and scattering region.
2. Calculation: Compute Green's functions.
3. Output: Transmission spectra and I-V curves.

Advanced Features

Surface Science:

• Dynamical Reconstruction: Observing surface atom rearrangement in real-time.
• STM Simulation: Generating theoretical Scanning Tunneling Microscopy images.
• NEB: Nudged Elastic Band for reaction barriers (in some versions).

Transport (FireballTG):

• NEGF Formalism: Non-Equilibrium Green's Function for molecular electronics.
• Conductance: Landauer-Buttiker formalism.

Performance Characteristics

• Speed: Significantly faster than plane-wave DFT (e.g., VASP, QE) for large systems.
• Scaling: Strictly linear O(N) for matrix construction; diagonalization can be O(N^3) or O(N) depending on solver.
• System Size: Routine handling of 1,000-5,000 atoms.

Computational Cost

• Single-Point: Seconds to minutes for hundreds of atoms.
• MD Step: Rapid enough for ps-scale dynamics in reasonable wall time.
• Comparison: Slower than DFTB+ (which uses parameter tables), but more rigorous (explicit integrals).

Limitations & Known Constraints

• Accuracy vs. Efficiency: "Fireball" approximations (basis cutoff) can lead to reduced accuracy compared to converged plane-wave results.
• Self-Consistency: Some older versions or modes (non-SCC) lack full charge self-consistency, though modern versions address this.
• Basis Set Optimization: Requires careful selection of orbital radii; poor choices lead to errors.
• TDDFT: Limited excited state capabilities compared to specialized codes.
• Documentation: Can be fragmented between different versions (Fireball vs Fireball2020 vs FireballTG).

Comparison with Other Codes

• vs DFTB+: Fireball calculates integrals ab initio (using efficient numerics) rather than using fitted tables. It is generally more transferable but computationally heavier than parameterized DFTB.
• vs SIESTA: Both use numerical atomic orbitals. SIESTA is more general-purpose; Fireball is highly specialized for MD speed in large systems.
• vs VASP/QE: Fireball is much faster for >500 atoms, but less accurate for high-precision electronic structure (e.g., band gaps).

Application Areas

Biomolecular Simulations:

• Enzymatic Catalysis: QM/MM study of reaction mechanisms.
• DNA Damage: Photolesions and charge transfer in DNA.
• Protein Solvation: Water-protein semi-quantum interactions.

Nanomaterials:

• Surface Reconstruction: Semiconductor surface annealing.
• Nanowires: Conductance and structural stability.
• Defects: Diffusion of defects in bulk materials.

Molecular Electronics:

• Single-Molecule Junctions: I-V characteristics of organic molecules between leads.

Best Practices

Basis Set Selection:

• Validation: Test numerical basis sets against plane-wave results for small systems first.
• Cutoffs: Ensure orbital cutoff radii are sufficient to capture bonding but short enough for efficiency.

Convergence:

• SCF: Use mixing (Pulay/Broyden) cautiously; difficult systems may require increased electronic temperature (smearing).
• MD Stability: Use smaller timesteps (0.5-1.0 fs) for variable-charge/flexible-basis MD to conserve energy.

QM/MM Setup:

• Partitioning: carefully define the QM region to minimize boundary errors.
• Link Atoms: Use standard link atom schemes (hydrogen caps) at the QM/MM boundary.

Community and Support

• Project Structure: Open-source, hosted on GitHub.
• Primary Group: Fireball2020 / Thunder-DFT team.
• Support: GitHub issues, academic collaboration networks.

Verification & Sources

Primary sources:

1. Official GitHub: https://github.com/FIREBALL2020
2. Documentation: https://thunder-dft.github.io/
3. Comparative Reviews: Included in lists of O(N) DFT codes.

Confidence: VERIFIED