PEtot

Official Resources

• Homepage: https://psi-k.net/codes/petot (Archive/Reference)
• Documentation: Available within distribution packages (often manual.pdf)
• Source Repository: No central modern repo; historically distributed by L.-W. Wang's group (LBNL)
• License: Open Source (specifics vary by distribution, often academic/research use)

Overview

PEtot is a plane-wave pseudopotential Density Functional Theory (DFT) code specifically designed for large-scale materials simulations. It employs norm-conserving and ultrasoft pseudopotentials and is renowned for its efficient parallelization capabilities, allowing it to scale to thousands of processors. It serves as a foundational engine for other codes, such as PWtransport for quantum transport calculations.

Scientific domain: Materials science, condensed matter physics, quantum transport (as backend) Target user community: Researchers studying large systems, nanostructures, and needing a highly parallelizable plane-wave code

Theoretical Methods

• Density Functional Theory (DFT)
• Plane-wave basis sets
• Norm-conserving pseudopotentials
• Ultrasoft pseudopotentials
• LDA, GGA functionals
• Iterative diagonalization (Conjugate Gradient, DIIS)
• Empirical pseudopotential method (EPM) capabilities (in some variants)
• Real-space multigrid options (in related specialized versions)

Capabilities

• Ground-state electronic structure
• Large-scale system simulation (thousands of atoms)
• Geometry optimization (atomic relaxation)
• Ab initio Molecular Dynamics (AIMD)
• Band structure calculation
• Total energy and force calculations
• Massive parallelization (MPI)
• Support for isolated and periodic systems
• Electronic density generation

Key Strengths

Scalability and Parallelism:

• Designed for High-Performance Computing (HPC)
• Efficient MPI parallelization over G-vectors, bands, and k-points
• Scales well to thousands of cores
• Suitable for large supercomputers

Application Areas

• Semiconductor Physics: Large supercell calculations for defects and dopants.
• Nanostructures: Quantum dots, nanowires, and isolated clusters.
• Quantum Transport: Electronic structure generation for NEGF codes like PWtransport.
• Alloys: Disorder modeling using large supercells.

Large System Handling:

• Optimized for systems with high atom counts
• Efficient memory management
• Fast iterative solvers for large Hamiltonians

Algorithm Versatility:

• Multiple wave function solution methods (Band-by-band, All-band CG, All-band DIIS)
• Flexibility in handling different system types (insulators, metals)

Inputs & Outputs

• Input formats:

  • Main input file (control parameters)
  • Atom configuration files (coordinates)
  • Pseudopotential files (UPF or native formats)

• Output data types:

  • Standard output (energies, forces, convergence info)
  • Charge density files
  • Wavefunction files
  • Band structure data
  • Relaxed coordinates

Interfaces & Ecosystem

• Integrations:
  • Foundational engine for PWtransport
  • Used in conjunction with various post-processing tools in the L.-W. Wang group ecosystem
  • Interfaces with visualization tools capable of reading standard charge/density formats

Performance Characteristics

• Speed: High performance on parallel architectures due to optimized MPI communication.
• System size: Capable of handling thousands of atoms, making it competitive with other flagship codes for large-scale problems.
• Parallelization: Multi-level parallelization ensures efficient resource utilization.

Computational Cost

• Scaling: Generally $O(N^3)$ with system size, but efficient prefactors due to optimized FFTS and linear algebra.
• Memory: Distributed memory model allows handling systems that would exceed single-node RAM.
• Efficiency: Plane-wave basis requires large grids for "hard" pseudopotentials, but ultrasoft/norm-conserving potentials mitigate this.

Best Practices

Parallelization Strategy:

• Hybrid MPI: Use MPI for inter-node communication; pure MPI is often standard for PEtot.
• K-point Distribution: For smaller systems, distribute k-points first for near-linear scaling.
• G-vector/Band: For large setups (single Gamma point), rely on G-vector and Band parallelization.

Optimization:

• Process Binding: Disable hyperthreading for improved floating-point performance.
• Memory: Ensure sufficient memory per core; excessive swapping kills performance.
• Start-up: Use a converged wavefunction from a smaller cutoff or similar system to speed up SCF.

Community and Support

• Primary Hub: Psi-k Portal.
• Development: Historically centered at LBNL (Lin-Wang Wang group).
• Support Channels: Direct contact with developers or academic collaboration; less community forum traffic than VASP.

Limitations & Known Constraints

• Availability: Standard distribution is less centralized than codes like VASP or Quantum ESPRESSO; often obtained via direct academic contact or specific archives.
• Documentation: Less comprehensive or modern online documentation compared to major community codes.
• User Interface: Typically relies on traditional text-based input files without a modern GUI.

Comparison with Other Codes

• vs VASP/QE: PEtot focuses heavily on large-scale parallel performance specifically for plane-wave calculations, though it lacks the vast feature set (e.g., advanced functionals, extensive post-processing) of VASP or QE.
• vs BigDFT: PEtot uses standard plane waves, whereas BigDFT uses wavelets. PEtot is a traditional plane-wave code optimized for size.

Verification & Sources

Primary sources:

1. Psi-k Code Database: https://psi-k.net/codes/petot
2. L.-W. Wang Group Publications (LBNL)
3. "PEtot: A plane-wave pseudopotential density functional theory program for large systems" (Description in literature)

Confidence: VERIFIED - Code existence and primary features are well-documented in scientific literature and community databases.

Verification status: ✅ VERIFIED

• Existence: CONFIRMED
• Domain: DFT/Plane-Wave
• Key Feature: Massive Parallelism