SPHINX

Official Resources

• Homepage: https://sxrepo.mpie.de/projects/SPHInX/wiki
• Source Repository: https://sxrepo.mpie.de/projects/SPHInX
• Documentation: https://sxrepo.mpie.de/projects/SPHInX/wiki
• License: Apache License 2.0

Overview

S/PHI/nX is a C++ library for materials simulation, with its dominant feature being a plane-wave density functional theory (DFT) code. Developed at the Max Planck Institute for Sustainable Materials (formerly MPIE), it provides efficient electronic-structure calculations using both norm-conserving pseudopotentials and the projector augmented-wave (PAW) formalism. The code is designed as an object-oriented C++ library that can be used both as a standalone DFT program and as a foundation for developing new physics codes.

S/PHI/nX supports a range of exchange-correlation functionals including LDA, GGA-PBE, and hybrid functionals (PBE0, HSE) as experimental features. It also includes k·p multiband modeling capabilities for semiconductor heterostructures, making it versatile for both bulk and nanostructured materials. The package includes dozens of specialized add-on programs for setup, analysis, post-processing, and other simulation types.

Scientific domain: Electronic structure theory, materials science, semiconductor physics
Target user community: Computational materials scientists, semiconductor physicists

Theoretical Methods

• Kohn-Sham density functional theory with plane-wave basis
• Norm-conserving pseudopotentials and PAW formalism
• k·p multiband model Hamiltonians
• LDA, GGA-PBE, PBE0 and HSE hybrid functionals
• DFT+U for molecular orbitals
• Strain calculations
• Generalized dipole correction for charged slabs

Capabilities (CRITICAL)

• Plane-wave DFT calculations with PAW and norm-conserving pseudopotentials
• Reads PAW setups from VASP, ABINIT, AtomPAW, and CPPAW formats
• k·p multiband modeling with flexible N-band Hamiltonians
• Very robust and fast minimizers with efficient preconditioning
• Atomic spin constraints
• Geometry optimization and band structure calculations
• Defect supercell corrections (sxdefectalign, sxdefectalign2d)
• Charged slab and 2D material corrections
• Object-oriented C++ library for custom code development

Inputs & Outputs

Input formats:

• S/PHI/nX native input files (.sx format)
• Structure files with flexible geometry via material maps
• PAW potential files (VASP, ABINIT, AtomPAW formats)

Output data types:

• Total energies and forces
• Band structures
• Charge densities
• Wavefunctions
• Strain fields
• Defect correction energies

Interfaces & Ecosystem

• Programming language: C++
• Parallel computing: MPI parallelization, GPU acceleration support
• PAW format compatibility: VASP, ABINIT, AtomPAW, CPPAW
• Add-on tools: sxdefectalign, sxdefectalign2d, structure setup tools
• Build system: CMake with Open Build Service binary packages

Limitations & Known Constraints

• Hybrid functionals (PBE0, HSE) still experimental
• Smaller user community compared to VASP or Quantum ESPRESSO
• Documentation primarily in reference manual format
• Some features require building from source for specific architectures

Performance Characteristics

• Optimized numerical algorithms exploiting CPU vectorization
• Efficient preconditioners for wavefunction minimization
• GPU acceleration for key routines
• FFT operations at 25-50% of theoretical peak
• BLAS operations at 80-90% of theoretical peak

Comparison with Other Codes

• vs VASP: S/PHI/nX is open-source (Apache 2.0); VASP is commercial. Both use PAW; S/PHI/nX offers k·p modeling unique to it
• vs Quantum ESPRESSO: Both open-source plane-wave codes; S/PHI/nX has object-oriented C++ architecture vs QE's Fortran
• vs ABINIT: S/PHI/nX offers unique k·p and defect correction add-ons; ABINIT has broader GW/BSE support

Best Practices

• Use PAW formalism for accurate total energies and forces
• Leverage sxdefectalign for charged defect calculations
• Use k·p module for semiconductor heterostructure modeling
• Build from source for MPI-capable and GPU-enabled versions
• Use the efficient preconditioned minimizers for faster convergence

Verification & Sources

Primary sources:

1. Official website: https://sxrepo.mpie.de/projects/SPHInX/wiki
2. S. Boeck et al., Comput. Phys. Commun. 182, 543-554 (2011)
3. O. Marquardt et al., Comp. Mat. Sci. 95, 280 (2014)

Confidence: VERIFIED - Official website and documentation confirmed accessible